User not logged in - login - register
Home Calendar Books School Tool Photo Gallery Message Boards Users Statistics Advertise Site Info
go to bottom | |
 Message Boards » » arghx's technical ramblings thread Page 1 2 3 4 5 6 [7] 8 9, Prev Next  
arghx
Deucefest '04
7584 Posts
user info
edit post

The last generation Honda Prelude Type SH uses what Honda calls the Advanced Torque Transfer System, or ATTS. This is a front wheel drive vehicle, and front wheel drive has a tendency to understeer. To address this, ATTS uses clutches to engage a planetary gearset in order to transfer torque between the two axles and increase the speed of the outside wheel in a turn. The system uses a lot of elements you find in an automatic transmission.



You can see its location in the front of the vehicle and the hydraulic lines to cool the fluid for the wet multiplate clutches (similar to what's in an automatic trans).



Here you see the two half shafts, a set of planetary gears, the pump used to generate hydraulic pressure, solenoid valves and fluid pressure sensors like you would see in an automatic transmissions' valve body.



Here's a basic diagram of a planetary gearset. In the middle is the sun gear, then the planetary gears connected to a common gear carrier, then a ring gear on the outside. For purposes of this discussion, the ring gear is irrelevant for understanding how this system transfers torque between axles.

A simplified way to explain it is that, if you think 3 dimensionally, there are 3 sun gears next to each other: left, right, and central. There are two sets of pinion gears, left and right. The clutches will engage the left side with the central, or the right side with the central. If I engage the left clutch, I transfer torque/driving force from the left wheel to the right wheel, and vice versa. Due to carefully-chosen differences in the number of gear teeth, it works out that the outside wheel in a turn will be able to spin at ~1.15 times the speed of the inside wheel when the clutch is fully engaged. This transfer of torque and rotation to the outside wheel reduces the understeer tendency.



The above diagram shows the basic structure of the control system. This generation Prelude came to market as a 1997 model, and these are pretty advanced controls for back then. A basic engine torque model and wheel driving force/torque model are used to determine how to engage the clutches. The system also had a yaw sensor and a G sensor to better understand the current driving maneuver. These sensors are more likely to be found in the more advanced AWD or other vehicle dynamics systems of today.

3/1/2015 12:56:22 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

Compare the timing and accessory drive of the Audi 4.0 liter twin turbo V8 engine (420hp), which is a dual overhead cam engine, and the GM Gen V 6.2 LT1 engine (450hp) with pushrod valve actuation. Both engines have cylinder deactivation and variable valve timing.

Here's the Audi chain drive:



So you've got 4 chains, labeled A-D

Chain A - connects from the crank to intermediate gears on the two banks
Chain B - driven by Chain A, drives the cams on Bank 2
Chain C - driven by Chain A, drives the cams on Bank 1
Chain D - drives all accessories except for the alternator

You can see the system of accessory gears connected to Chain D:



The alternator is the only accessory with conventional serpentine belt and tensioner:



Why the complexity? I would think that the use of intermediate gears and separate chains for the cams reduces noise and doesn't have the stress of one big chain driving all the gears, especially with the variable valve timing cam phasers and high pressure fuel system.

For the chain driven accessories I can only speculate that it may have been a packaging issue of some sort--reduces the number of pulleys. Or maybe the chain drive as implemented here was quieter.

Now, look at this cutaway of a GM Gen V LT1 engine:



You can see on the left a small chain driving a single camshaft in the valley, with a simple dual-equal phaser (retards both intake and exhaust valve timing at the same time). The high pressure fuel pump is cam driven but not shown here. Then you have a simple serpentine belt system to drive the accessories. This is a less expensive and less complex design than what Audi uses, especially since it is a large displacement naturally aspirated pushrod V8.

5/10/2015 11:00:03 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

Porsche was the first manufacturer to have a turbocharged engine for racing. The 1973 Porsche 917 was a 5.4 liter flat 12 cylinder rated at 1100 horsepower. It used a setup that may be familiar to many of you with flat engines: an external wastegate controlled by a manual boost controller.



An external wastegate bypasses exhaust gases from the turbocharger wheel to control boost. A spring-loaded poppet valve controls flow. The stiffness of the spring indirectly controls the amount of boost pressure to the engine. The diaphragm chamber is divided into an upper portion and a lower portion. The lower portion can receive boost pressure to left the poppet valve off its seat and bypass exhaust from the turbo. The upper portion can be open to atmosphere, or it can receive pressure to push the poppet valve closed and raise boost.



When the valve lifts off its seat, it sends exhaust back into the main exhaust stream. Porsche originally used a physical screw to set preload in the spring in order to raise boost pressure. Then they used a manual valve to regulate additional air into the upper portion of the diaphragm chamber.

Wastegates of course are standard now on turbochargers, but at the time it wasn't always the case. The Chevrolet Corvair Monza for example, a turbocharged flat engine, had no wastegate, and neither did a lot of diesels. Not all wastegates were controlled with boost pressure; some were actually controlled by letting exhaust backpressure push the poppet valve open by overcoming the force of the spring. So technology and systems we take for granted today had to be discovered by earlier generations of tinkerers.

5/11/2015 10:58:59 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

These days most new mass production gasoline engine designs with dual overhead cams have phasers on the intake and exhaust cam. This means they can move the valve events earlier or later with continuous variation, rather than a "2 step" on/off system. This type of system can be called things like VVT, VTC, VCT, VANOS, i-VTEC, etc.



The cams default to a locking position, typically with no overlap or very little overlap between intake and exhaust events. Then the exhaust cam can be retarded (moved to the right in the image above) and the intake cam can be advanced (moved to the left in the image above), and in some systems the intake cam can be retarded as well.

Usually the cam phasers have some range of authority, usually anywhere from about 30 to 70 degrees. Newer engines tend to have cam phasers a wider range of authority, and require a piston design that doesn't allow the valves to contact the piston. Fully optimizing for fuel economy requires extensive studies on an engine dyno to measure the brake specific fuel consumption, combustion stability, residual gas fraction, etc.

The thread below, which I mentioned on the previous page, has more explanation about cams and what happens when you move cam centerlines around. http://forums.nasioc.com/forums/showthread.php?t=2687550

So here's an example of a full detailed study ("full factorial" is the jargon) at a specific speed and load point:



Take an engine with a 60 degree intake cam advance authority and a 50 degree exhaust cam authority. The four corners of the map above represent four different cam phasing strategies. At the bottom left is the cam running with the phasers at their locking position. This is equivalent to a fixed cam engine. Whatever the default centerlines, lobe separation angle, or whatever term you want to use, is what you get. Since it's a continuously variable system, you can change the cam setpoint to whatever you want (whether it will hold precisely is another matter), but typically 5 degrees is a good step to move during these kinds of studies. A certain amount of time needs to be averaged at each combination, as this is a steady-state analysis. Now you can start moving cams.



Keep the exhaust cam locked and step the intake cam toward the bottom right corner of this chart and you get primarily an early intake valve closing scenario with higher effective compression. There's also a small amount of residual gas (internal EGR).



Move to the upper right corner by retarding the exhaust cam and you get a high overlap and high residual gas fraction (internal EGR), assuming that scavenging isn't going to come into play under these conditions.



At the upper left corner you have a locked intake cam and a fully retarded exhaust cam. With this particular set of cams you get a small amount of residual gas and minimal overlap. With a different design though you may have more overlap. You also have late exhaust valve opening, which increases expansion ratio. With this particular cam design your default position has a somewhat late intake valve closing.

Note that this image above is somewhat similar to what "dual equal" phasers do, that is variable cam phasing on pushrod engines and SOHC engines (Ford 4.6 3 valve, GM and Chrysler pushrod V8). They can only retard the intake and exhaust cam together, and the overlap is fixed.

If you look back at our cam phasing mapping plan you'll see a bunch of diagonal arrows and numbers. That's one possible way to step through all the combinations when testing on an engine dyno. This can be controlled manually or it could be automated. Now that we've illustrated the 4 corners of the map, let's show what a resulting map might look like:



Each set of numbers represents the brake specific fuel consumption (grams/kwh , lower is better), the location of 50% mass fraction burn (crank angle degrees ATDC firing, ~8ish is considered minimum spark advance for best torque), and the amount of combustion instability (coefficient of variance of indicated mean effective pressure in %, lower is better).

So in the example shown here, the engine seems to like the upper left corner the best in terms of fuel economy (red numbers). This is late intake valve closing, late exhaust valve opening, and could also be considered Atkinson Cycle. Under other conditions, maybe the engine likes the upper right corner better. Notice that the edges of the map, especially the upper left corner in this case, have the highest gray number. The combustion can become unstable from overlap or late intake valve closing. If it's bad enough, a driver could feel vibration in the vehicle. Generally 3% is a rule-of-thumb limit, but it depends on the application. Maybe under other conditions your most unstable point will be the upper right corner of the map rather than the upper left.

The black number is the location of 50% burn. It's between 8 and 9 degrees ATDC here. The reason this moves around is that the combustion speed changes as we change the valve events. The spark timing needs to change with it. A slower burn requires more spark advance, which has to be adjusted by whoever is running the engine or by the automation program running.

5/16/2015 6:01:24 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

Back on page I talked about how twin scroll turbos separate exhaust gas pulses on an I4 engine to reduce residual gas in the engine and give better low end torque:



There are a few drawbacks to twin scroll turbos though. The divider in the middle and overall design of the turbine housing causes more restriction and backpressure at high speed. These housings also must be run at lower exhaust temperatures than monoscroll turbos of similar material, and can be more of a thermal sink to inhibit catalyst light off.

So there's another way to accomplish the same effect without some of the drawbacks of twin scroll turbos. By reducing the exhaust cam duration and retarding the exhaust valve opening event, the blowdown pulses will not flow back into subsequent cylinders. Audi uses this principle on some of their newer I4 turbo engines, and advertise it under the umbrella term of Audi ValveLift System.



The diagram above shows conventional valve timing during overlap. The exhaust pulse from an earlier cylinder firing event travels to this cylinder, causes a rise in exhaust port pressure. When the intake valve opens, the exhaust travels into the combustion chamber and results in hot residual gases remaining in the cylinder. That increases the chance of knocking and reduces volumetric efficiency as fresh air is displaced.



In the image above, the top pane overlays the short duration exhaust cam profile with the standard profile. The short profile opens the valve a lot later and closes later here. In the bottom pane, the short profile's cam results in the exhaust pulse not reaching the subsequent cylinder until after the exhaust valve is closed. The result is similar to a twin scroll turbo at lower engine speeds and high load, but without the drawbacks described above, especially at peak power speed where monoscroll turbines are advantageous due to less backpressure.

5/28/2015 4:22:55 PM

Hiro
All American
4673 Posts
user info
edit post

I miss these posts...

6/1/2016 6:05:31 PM

sumfoo1
soup du hier
41043 Posts
user info
edit post

i'm not going to lie... i really feel that all of the downsides to twin scroll compressors can be made up with other options, turbo blankets shorter exhaust runners etc.

i think fucking with your timing like that would become more of a maintenance issue than anything.

arghx is there any other way we can stalk your intelligence and harass your moving over to the O.E. darkside?

also do you guys do any "engineered to fail" parts.

6/1/2016 8:40:49 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

Quote :
"I miss these posts..."


absence makes the heart grow fonder

Quote :
"i'm not going to lie... i really feel that all of the downsides to twin scroll compressors can be made up with other options, turbo blankets shorter exhaust runners etc."


If you bolt the turbo right to the head on an I4 with integrated exhaust manifold like many/most new turbo engines, you can't shorten the runners, or put much of a turbo blanket on it.

Quote :
"arghx is there any other way we can stalk your intelligence and harass your moving over to the O.E. darkside?"


I don't know if it's the dark side. It's a lot more difficult and challenging in some ways, and much easier in other ways. More difficult because of government regulation, having to adhere to engineering standards that some retired guy came up with in the 80s, that kind of thing. Much easier because if you want to know how a part works, you can literally call the guy. If you need a simulation, you can do it. If you want to know how something is calculated, you open a block diagram. If you want to study something, you run it in a lab with equipment that costs more than your house.

Here's an example. On a port injected engine I've been working on, during warm up the tip-in fuel has to not only drive well but actually pass emissions. On a turbo direct injected engine, some of the features for knock reduction or better fuel economy had to be constrained because it could create noise or vibration and nobody wants to spend the money for something to address that.

Quote :
"also do you guys do any "engineered to fail" parts."


You don't need to intentionally engineer something to fail. That just happens naturally. What if someone told you that you had to design something and it better work great for 10 years on a million cars running around in the real world? All you get is an Excel spreadsheet for quick calculations, maybe one or two rounds of modeling, and if you're lucky some accelerated testing in a lab. By the way though, you won't be able to test it in the real world until a year from now. And if it doesn't meet your requirements, you will encounter extreme resistance to change because it will delay launch and cost millions and millions in discarded tooling.

7/8/2016 8:54:48 PM

sumfoo1
soup du hier
41043 Posts
user info
edit post

it wasn't meant to be accusatory as an engineer myself i know life cycle design.

But sometimes cost cutting things that don't fall in line with the life cycle or the rest of the part or assembly seem to be "engineered to fail"

for example water pumps, oil gerotors etc seem to be made of weaker materials that everything else.

7/10/2016 12:27:42 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

Since I work in gasoline engine performance and emissions system development I'm usually bitching at the hardware design guys for making a piece of shit, so I understand where you are coming from

7/12/2016 11:27:53 PM

sumfoo1
soup du hier
41043 Posts
user info
edit post

ok so can bus, is it pretty much a universal protocol or?

I'm trying to figure out if engine swaps could get easier or since they're tied to effing everything it's just going to be a giant pain in the genitals.

7/29/2016 7:52:33 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

there are different CAN versions, and the network messages vary with manufacturer and sometimes with model year. Despite all the media hype, CAN networks are not fast and easy to reverse engineer, even though in current models they technically don't have any kind of encryption. Those proof of concept stories where one guy made such and such car hit the brakes took a LOT of work and specialized tools & knowledge.

The OEM has the full CAN database and will give pieces of it to suppliers on a need-to-know basis. Then you need a dictionary of what the messages mean and what they do, and how the commands are issued.

All that being said, over time the knowledge will accumulate. Considering the aftermarket can barely handle electronic throttle, which has been in production for about 20 years, it's going to take a while.

8/1/2016 10:55:13 PM

sumfoo1
soup du hier
41043 Posts
user info
edit post

Are speedometers/tachs some digital signal or a simple analog controls voltage/ amperage like the 0-10v or 4-20ma commonly used in building automation

Kinda want to swap a coyote into another obdII ford but am trying to figure out how hard harness splicing will be.

From what i read it's relatively easy with only a couple changes in the pin-out (sohc modular to coyote)

Using the coyote control package so that includes the DBW pedal maf and ecu for the coyote.


[Edited on August 2, 2016 at 12:05 AM. Reason : .]

8/2/2016 12:02:59 AM

Hiro
All American
4673 Posts
user info
edit post

Dakota Digital has plug and play options, but they can be $Texas

8/2/2016 11:42:28 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

I don't know the specifics of the application, but on modern cars the speed comes over CAN message from whatever module(s) is controlling vehicle dynamics (ABS, stability control etc). There are individual wheel speed signals, torque reduction requests, transmission input shaft signals, all that stuff.

Early 90s through early 2000s typically have an analog signal for speedometer. Before that you have a lot of cars using an oldschool speedometer cable.

8/2/2016 9:54:06 PM

sumfoo1
soup du hier
41043 Posts
user info
edit post

Cat before or after turbo for faster spool?

If the cat re ignites the mixture it seems like it would help spool but choke the turbo sooner.


I don't deal well with emissions voodoo halp!

11/1/2016 12:37:28 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

Quote :
"Cat before or after turbo for faster spool? "


After. The cat absorbs thermal energy from the exhaust, and you will have to run richer to protect it, or it will melt and blow up your turbo. Cat before is like the last resort (ie, some Subarus) when the OEM couldn't meet emissions and didn't want to do much redesigning of the engine. You can only do that on an engine with a crossover pipe (usually 2-bank engines) where you can actually fit it in there.


[Edited on November 9, 2016 at 2:17 PM. Reason : .]

11/9/2016 2:11:31 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

Today I want to talk about torque converter torque multiplication. There are gazillions of sites and videos about what a torque converter is and what it does. The torque multiplication effect deserves some extra attention. Here's a Youtube video called "How does a torque converter stator work?"



It shows how the stator, the fixed part in between the spinning engine side and spinning transmission side, directs fluid in such a way as to add torque. This happens when the engine side is spinning faster than the transmission side which ultimately connects to the wheels. To illustrate this relationship, we can use data from a relatively modern but still basic 6 speed front wheel drive auto transmission mostly found in frumpy sedans.

Test Bench



In this case, the transmission is being spun on a test bench where the input and output shaft speeds can be independently controlled, with the converter clutch open (converter is not locked up, top pane of image). The input shaft speed is fixed at 2000rpm, while in 6th gear with a fixed hydraulic line pressure. With the gear and final drive applied we have a fixed difference in speed (speed ratio is 1). Then the test bench can lower the output shaft speed, reducing the speed ratio and raising the multiplication effect. The baseline relationship with closed torque converter can also be measured (bottom pane of image). By running output shaft speeds we can build a map of the torque multiplication relationship.

Data



The bottom x axis is the ratio of speeds between the input and output shaft. Remember, the input shaft speed is held fixed by the test bench, and the output shaft is affected by the gear ratio (around 2.2:1 here). So a 1.0 ratio is not going to have the input and output shaft spinning at the same speed. If the ratio is less than 1, it represents the torque multiplication effect of the converter, not any kind of gearing change, when the torque converter clutch is open.

When the torque converter clutch is closed, the input shaft speed needs to be varied to get the output shaft speed labeled on the top of the X axis.

The Y axis is torque. The test bench dyno motors can apply torque on the input side, and absorb torque on the output side. So looking at 1.0 on the graph, starting with the engine and transmission side spinning at the normal (gear related ratio), we look left at the green line and red lines. That increasing separation shows the multiplication effect. The increase in the red line (input shaft torque) reflects reactionary forces from the output shaft as a result of the torque multiplication effect.

As a reference, look at the blue dotted line and the top X axis. This is just varying the input shaft speed with the converter clutch locked.

Now look at the efficiency plot (lower ratio efficiency can be extrapolated).



This number is a comparison of the power put into the system vs the power put out. The output torque is going up but the output speed is going down. From the perspective of overall efficiency you can see why you would want to run at a speed ratio of 1.0 (torque converter locked), and that's why the newest autotransmissions have a locked torque converter in most areas of operation.

From the perspective of getting more torque to the ground through multiplication effect, you want to expand that gap between the green and red lines in the second image. That's what changing the design of the torque converter stator does, and it's a frequent change used in drag racing applications.

[Edited on November 9, 2016 at 3:03 PM. Reason : .]

11/9/2016 3:03:00 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

the youtube embed works for me if you are having trouble pulling up the torque converter video here is the link

https://youtu.be/vp_tHMkOjB8

11/10/2016 9:47:49 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

per a request via PM about a transmission concern, the inspection procedure for synchros on a manual transmission Acura Integra:



I've disassembled a manual trans before but never rebuilt one myself, although I've had a shop build one. You can see there are clearances and inspection procedures. I'm not sure if having out of spec clearances would correlate to gear grinding or not, but it's certainly food for thought.

[Edited on November 10, 2016 at 10:00 AM. Reason : or maybe they were barely in spec at the time of rebuild and through wear are now out of spec]

11/10/2016 9:56:26 AM

Dr Pepper
All American
3583 Posts
user info
edit post

Quote :
" per a request via PM .... manual transmission Acura Integra:"


C'mon Teg, we know it's you

11/10/2016 11:38:57 AM

Hiro
All American
4673 Posts
user info
edit post

^TROLOLOL CALLED OUT. PWNT!

Quote :
"or maybe they were barely in spec at the time of rebuild and through wear are now out of spec"

Yeah. Definitely this. Synchro rings are trivial in cost vs function/effort later to fix (kits about $200ish for all gears).


[Edited on November 10, 2016 at 3:36 PM. Reason : .]

11/10/2016 3:31:41 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

You know all that nasty black soot stuff that comes out of your tailpipe on these newer gasoline direct injection engines? The newest emission regulations are trying to reduce that, not only in situations that occur in controlled lab tests, but also for real world engine speeds and loads.

One of the ways to do this is to optimize the start of injection timing (SOI), the number of injection events, and the mass split between them. The images below show an optimization test at 1200 rpm, wide open throttle. That exact condition is not too common in the real world, but you can see some interesting trends.

In the example below the engine is run in a lab, running 20 seconds steady state (sitting there at WOT using systems to cool the engine) at each point with a fixed spark at TDC (0 degrees advance). The points in the scatterplot have different injection timings and mass splits. The Y axis is the injection mass split %. 100% would correspond to a single injection event. The other points have two injection events, where 80% on the Y axis would mean 80% of the injection pulsewidth is allocated to the first event, and 20% is in the second event, and 50% would be an even split.

The x axis is the start of injection timing in degrees BTDC firing, the same units we normally associated with spark timing. As you look towards the right, the injection timing is earlier. As you look towards the left, the start of injection timing is later. The black numbers are the brake mean effective pressure (proportional to measured brake torque, higher number = good), the red are the soot (higher number = bad), and the gray are the combustion instability (higher number = bad). Most of those numbers (black, red, gray) should be compared in a relative sense to the other points in the matrix. You want the highest BMEP, the lowest Soot, and the lowest combustion instability, or some compromise among those choices.




So in the upper right corner is a baseline condition, running a single injection event. We know it's a single injection event because the mass split is 100% --> 100% of the injected fuel is in the first event. The injector starts firing at 260 degrees before top dead center. Although the combustion instability of 2.7% is acceptable it is not great. If it's over 3.0% as a general rule you can feel vibration or torque fluctuation in a car. That's a general rule of thumb.

Then in the rest of the matrix you see two injection events. The first injection timing is still fixed at 260 degrees BTDC, while the second is varied from 220 to 80 degrees BTDC. We don't fire the second event earlier than that so the injector has enough time to respond, as this engine doesn't have expensive piezoelectric injectors. When the second event starts much later, we get lower torque, and what's not pictured here are the high exhaust temps as so much combustion is pushed out the exhaust valve (similar to what is done during a cold start). The later injection also has higher soot, as the fuel is impinging on the piston top or cylinder wall.

An even 50/50 split with the second injection firing 60 degrees after the first shows the most optimized point of stable combustion, low soot, and high brake mean effective pressure/brake torque. This is circled in blue.



This chart shows the soot and combustion instability, but also shows the location of 50% burn instead of the brake mean effective pressure. The units are in degrees ATDC firing. Considering these all have a fixed spark at top dead center, a higher number corresponds to a slower burn (which is generally not good) and a lower number corresponds to a faster burn (which is generally a good thing).

We can see from this chart that the highest torque point with the most stable combustion and low soot-producing impingement also is the fastest burning. We know this because the number "22.4" is the lowest among the points, meaning that half the mixture is burned 22.4 degrees after top dead center, with the spark fixed at TDC. In contrast, the upper left corner of that matrix with a low mass late second injection has a 50% burn location of 33.3 degrees. So it's burning much slower. Meanwhile the baseline is 27.8, meaning we've advanced the combustion about 5 degrees (almost like advancing the spark 5 degrees).

12/14/2016 10:13:36 AM

arghx
Deucefest '04
7584 Posts
user info
edit post



If any of you have ever seen the Youtube series Engine Masters, you may have been curious to know more about the engine dyno facilities used at West Tech performance. In the show several guys do before-and-after wide open throttle power and torque runs, usually swapping around parts on an engine.

Engine dyno work is a very "insider" thing as it is not as widely used for home projects and aftermarket performance. It can be a lot more expensive to set up than a new shop trying to find a used Dynojet, setting it above ground and being able to take any customer's car. Doing work on an engine outside of a vehicle is a more narrow application.

You'll see in the Engine Masters series they are always working on older American engines, almost always with a carburetor. This is no accident. Carburetors are cheap and simple, and suit the clientele for that kind of work (drag racers and such). With a carb you don't have to worry about the ECU freaking out because it's not seeing the normal inputs it gets from running down the road in a real car. Pretty much any OBD II engine's stock ECU is going to be unhappy running on a dyno, but as you get into the era of electronic throttle controls (past 10 years) and integration with stability control systems they are even more difficult to get working on a dyno with a stock ECU. So that explains the kind of engines you see on the show.

Now, it still IS possible to make a modern stock ECU work on an engine dyno. However this is done in a few ways that are not straightforward:

1) reprogramming a stock ECU (perhaps modified for additional memory used for testing) with a special tune to turn off all the check engine lights and disable certain self learning functions so that it doesn't freak out. For the most part only the OEM can do this.

2) literally taking the engine out, putting it on the dyno, and extending a bunch of wiring harnesses from the vehicle into the engine dyno test cell so that the ECU still sees all the connections. Then you have to simulate certain other signals such as vehicle speed and G sensors so that the ECU doesn't freak out and go into limp mode. This method is very expensive and time consuming; only a few labs in the world know how to do this method for the newest, most complicated engines. OEMs will pay for this kind of testing when they want a report on their competitors' engines, using their competitors stock ECU.

1/15/2017 11:59:36 PM

arghx
Deucefest '04
7584 Posts
user info
edit post



Now I want to go through the above ^ video discussing the engine dyno facilities. I'm going to give timestamps and insert comments.

0:46 First, I'll make some comments about the engine and equipment shown on the screen. The engine isn't actually hooked up to run, so there are a bunch of loose wires hanging around. Looking on the rack in the middle of the screen, it looks like some 12V power supplies and electronic ignition boxes (the kind of thing you expect on a carb + distributor engine). Also, I find the Autometer style analog needle gauges surprising. What are those there for? Are they just for reading while the engine is running and there is someone in the test cell at low rpm? You can't see them from the control room. Maybe all those measurements are going into the computer system digitally, and the analog needles in the test cell are just there for convenience.

1:32 He's talking about the brand of Dyno (SuperFlow) and explains that it is a water brake dyno rated for a lot of power. The other kind of dyno is an electric motor dyno. That kind of dyno actually generates electricity when the engine is firing and can return that electricity back to the power grid. It is a much more expensive and versatile type of dyno. It can spin an engine all through its rpm range and measure parasitic losses (motoring friction). The engine doesn't require a starter for example, so you don't have to tune the cranking of the engine to get it going. You can also spin the engine without it firing in order to check for leaks. However, electric dynos typically aren't rated for that kind of power. The dyno they have meets their needs.

3:00 he's talking about "the old days" of 30 years ago. Now I wouldn't know firsthand how things were, but you've got to figure that labs for aftermarket use are decades behind what R&D labs are doing. A bunch of guys standing around looking at needles on gauges and writing stuff down, well that's probably what Henry Ford was doing when he was developing the flathead V8. You still had analog chart recorders and other crude methods to record data, starting at least in the 1970s. So what he's describing is a reflection of his experience, but not a reflection of how all engine dyno labs worked in the mid 80s. They were using computers back then.

4:10 the actual dyno load cells and other components can last a really long time if you do the maintenance. However, other stuff in the lab tends to go out more frequently. Anything involving heating and cooling will fail more frequently, and the really sophisticated labs with sealed environmental chambers will nickle and dime you to death with broken air pumps, chillers, etc.

5:50 "We don't actually simulate atmosphere, we correct." That means they didn't have the money to put in a combustion air system. The combustion air system is a separate HVAC system that supplies air directly to the engine inlet via a hose or sort of external manifold. These systems control temperature, humidity, and pressure. So SAE standard temperature is 25C and pressure is 99 kPa. That's basically a June day in Detroit. The link below can calculate what your ambient pressure is at different altitudes.

http://www.mide.com/pages/air-pressure-at-altitude-calculator

Now of course you can "correct" for being non standard, but if you actually read SAE J1349, the correction factor should be within 3%, and there are acceptable tolerances. The STD correction factor he is talking about is from a different document (j607?) and is based on 103kpa atmospheric pressure and 15.5C/60F , so pretty much winter in death valley (very low altitude below sea level which is 101kpa, and thus not a realistic condition). Only lab combustion air can get to 103kpa.

Now only in rare cases do you have a test cell that is completely sealed and can raise and lower the ambient pressure of the whole room. See above comment about how hard those are to maintain. We're only talking about the supply air.

Another thing he doesn't mention is control of water and oil temperature. If you look very closely in some of the dyno videos, you can see their display in the control room showing water temperature at around 160F/71C, which is way cooler than a car is going to run. They're doing this because either they want to reduce engine knock (understandable) or their control system can't maintain a more realistic temperature (at least 180F/82C) without a lot of oscillation, or they haven't taken the time to tune the water temperature control system to something realistic.

The only real torque is the kind you actually measure. Correction factors get real sketchy, real fast. Look at any high altitude runs reported on a Subaru forum.

6:45 "For a lot of the OE's, SAE [correction] is where it's at." Yes, but only if they do a certified witness test can you somewhat trust those numbers. And that's mostly the big 3 who do that. There are a lot of tricks for overrating your engine in an engine dyno environment for advertising purposes that the OEM's use. That's the subject of another post I think.

6:58 This is a pretty old school control panel. Notice the analog needles and the lever at the left to control throttle. Later you will see a single computer screen. Most R&D labs have a small control panel like this and up to 5 computer monitors. 1 will be just for the ECU, 2 will be for the dyno control system, 1 will be for the cylinder pressure/combustion analysis/oscilloscope system, and 1 will be for the emissions measurement system.

7:40 On that screen are basic temperature and pressure measurements, plus measured air fuel ratio and exhaust temperature. There are a few analog looking gauges on there and digital readouts. What you don't see are second-by-second strip charts. Also as I mentioned, there are no measurements from an engine ECU, emissions measurement, or anything from oscilloscopes and combustion pressure. However since they are mostly doing WOT pulls for power you don't need a lot of that stuff.

8:05 He says the computer is taking 100 measurements per second. That depends on what the bottlenecks are for individual measurements. For example, if his wideband oxygen sensor system (the "ECM" box you saw earlier) is sampling at a slower rate than that, he's limited. Much of it will be buffered and averaged as he mentions.

8:30 60 channels (including actual measurement and calculated values) sounds like a lot, but it all depends on the context. Like he said, half of them he doesn't use. Since there is no combustion, emissions, or ECU parameters, yeah 60 a significant amount. However with more equipment hooked up it's more like 200-300 is going to be common on a modern engine, depending on number of cylinders and the nature of the testing plus complexity of engine hardware. If they were doing detailed thermal testing for example it could be an additional 100 or more channels just for the temperature measurements. With that amount of data you need special tools to process everything and get something useful out of it quickly, and it's really easy to get disorganized.

8:45 he mentions trimming cylinders for fuel and spark. I would be interested to see how they have their test cell configured when they do that. Are they running something like a Fast XFI standalone ECU's control software on a different PC and then sending the data to the main dyno control PC?

8:52 he says an "older way" is to look at exhaust temperature. I think he was referring to 10 or 20 years ago when wideband o2 sensors weren't available and you couldn't measure air/fuel ratio cheaply, without an emission measurement system (they dont have one). In those cases a richer mixture will make a cooler exhaust temperature, and a leaner mixture will make a hotter temperature, all things held equal. But exhaust temperatures still matter because you can damage components if you get too hot (valve seat etc).

9:37 I totally agree that being able to acquire all that data on a dyno saves so much time compared to his customers again and again testing on a track without enough measurement. There's a specific kind of customer who wants that though. It's not the broke college student with a 2003 WRX.

11:12 He's talking about safety when the engine is running. Some people are IMO way too reckless about being inside a dyno cell while the engine is running. Union rules in one lab I can think of was that you couldn't be inside the cell if the engine was over something like 1700rpm, and you couldn't be in there if it was at wide open throttle. Whatever is wrong with the engine isn't worth risking your life over, and whoever is pressuring to get the work done can chill the fuck out.

11:30 "We can actually run open headers" That means they didn't spend the money for an exhaust system vent. There could/should be a system in the building suctioning out the exhaust into the atmosphere.

12:00 He's talking about how the cell was designed to run full exhaust systems, which from his perspective is unusual. It's unusual for performance tuning shops maybe, but larger labs at component suppliers, R&D facilities, and OEMs as I mentioned are going to run full exhaust systems all the time and evacuate the gases into the atmosphere. The test cell in the video does have an air handler in the cell moving air out, which is normal. They rely on that exclusively to prevent leaks of carbon monoxide into the rest of the building, rather than a dedicated exhaust vent.

[Edited on January 16, 2017 at 12:04 AM. Reason : .]

1/16/2017 12:00:25 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

13:05 He brings up good points about the physical feedback of being connected to the throttle valve through cable, and how it helps him feel if there is something wrong with the engine. Keep in mind though that you can't get precise control over engine load if you are relying on the operator to physically move the throttle open and closed. On newer engines the dyno computer will send a signal emulating an accelerator pedal input to the ECU. Then control loops can set a specific speed and torque target, so that you can optimize part throttle for driveability, fuel economy, etc.

He refers to a servo motor for a manual throttle cable. Well if you're only doing WOT pulls then yeah I can see how that wouldn't add much value. However if you need to optimize other areas of engine operation using a physical cable is just a hassle. More advanced test cells will let you script a series of actions and optimization tests so that you can press a button, go home for the night and come back to a bunch of data in the morning.


Overall, great video. Hopefully my comments will give more context to how his dyno setup compares to others.

1/16/2017 12:03:56 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

Let's talk about camshafts and low end torque. Bigger isn't always better. We know that, but it's interesting to see it illustrated.

Long duration cams with the intake valve closing long after bottom dead center is not good if your goal is trapping the most air in the cylinder. One way to find the best theoretical valve timing is by measuring the volumetric efficiency of the engine with different cam centerlines and different lift/duration profiles. Below illustrates a fixed intake centerline with varying intake cam duration/lift profiles and a fixed exhaust. Notice that with the different intake profiles, the opening and closing timing of the valve will vary, but the centerline (a straight line through the point of peak lift) doesn't change. Now let's move on to the different combinations of intake centerline and cam profile:



Here is test bench data from a gasoline V8 engine using 7 different cam lift/duration profiles with the intake same centerline, shown here on the Y axis. Exhaust valve timing remains the same in each case. With each profile the intake cam can be at a baseline position optimized for higher rpm or the centerline can be advanced up to almost 50 degrees. Engine speed is 2400rpm, which is realistic for "stump pulling, get off the line" scenarios. The engine is operating as an air pump here; it is not firing, so exhaust gas dynamics (scavenging effects etc) are not a major concern. This method isolates the effect of the intake-side valve timing.



The cylinder filling % is calculated by taking the engine control unit's measured mass airflow (from a hotwire type mass airflow meter) and dividing it by the theoretical pumping according to engine displacement etc.



The peak pressure in the cylinder is calculated by taking the peak pressure measured within a 200 cycle window from a cylinder pressure transducer located in the spark plug.


So what can we learn from these graphs? Well, when the engine is only pumping air, the valve timing of best pumping also has the highest cylinder presssure. At an engine speed like 2400 rpm, the biggest duration with the latest valve timing really sucks for pumping --> compare the 69% volumetric efficiency in the upper left corner with the 93.4 % volumetric efficiency circled in green.

Also, notice that on the far right column with the most advanced cam centerline, the longest duration cam is still too long--the upper right corner is at 93.0%, while the circled green is 93.4% with 16 degrees shorter duration (and thus the closing and opening timing are affected from this despite the same centerline).

In this case, the pumping efficiency (motoring condition) of the engine is more sensitive to change in intake cam centerline than it is to change in profile.

What does that mean to you? Well, there's more to a cam than its lift/duration profile. Those big numbers thrown around for marketing purposes aren't everything. The timing of the valve events has a huge effect, especially the centerline of the cam (or lobe separation angle depending on the context).

2/6/2017 8:26:17 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

If you have a turbo engine in your car, there's probably some thread on a car forum somewhere complaining that the stock intercooler for your car sucks. What I find interesting are the assumed efficiency of the stock intercoolers for the rated horsepower. One of the ways to look at that is by graphing up the intake manifold temperatures (after the intercooler) from the published certification reports. These measurements are taken in a lab although they are supposed to represent real world[ish] conditions.



The X axis represents the engine rpm, and the Y axis represents the temperature after the intercooler. Keep in mind though that the engine speed is ramped very slowly in these tests, like lugging your car in top gear at 1000rpm and flooring it under it redlines. Also, the intercoolers can all be controlled in different ways in the lab, usually using some kind of forced cooling due to insufficient airflow from fans.

So the dotted line is 25C or 77F, which is the temperature of the air at the airbox and the standard condition for power correction calculations. You can see post intercooler temperatures mostly ranging from 35C (95F) to 50C (122F). What's actually interesting and sort of impressive to me are the engines that are making their power with hotter temperatures, because that implies that they don't depend on the stars all aligning to perform well (cold day out).

[Edited on May 13, 2017 at 9:06 PM. Reason : 1000rpm]

5/13/2017 9:04:23 PM

arghx
Deucefest '04
7584 Posts
user info
edit post



These old Aston Martin Lagondas are pretty cool cars. I've had the opportunity to ride in the car above. The host of the video describes some of the "space age" interior features at the time, such as starting the trend of touch switches and digital dashes (early 80s style).



Right around 6:45 he starts describing some of the styling elements that were possible due to the hand made design of the vehicle, such as the fenders. One of the reasons why "cool concept cars" can end up looking so different in later mass production stage is the difficulty of making complex shapes on the body panels.

[Edited on May 20, 2017 at 8:44 AM. Reason : 6:45]

5/20/2017 8:37:04 AM

Hiro
All American
4673 Posts
user info
edit post

That's super cool dude! thanks for sharing

5/22/2017 5:32:57 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

If there are two things in life I hate, it is the following:

1) Going to stores, especially big home improvement stores
2) Maintaining small engines with carbs

I'm not a carburetor guy. That technology's heyday was before my time. I think the only thing I hate more than carbs are points ignition. I've done some minimal work on small engine carbs, with the help of zxappeal. For small engines, the annoying part is that it is a seasonal thing. You may have something sit for months, and then it starts needing an oil change, a carb cleaning, new spark plug, maybe even a carb build. Rather, it doesn't strictly NEED those things at most moments, but we all get busy and lazy. So we may put off the maintenance. And then it doesn't start, or in worst case scenarios with old equipment its lost compression, or spun a bearing, or whatever.

So I've decided to migrate to all electric, lithium Ion powered yard care equipment.



What you're looking at is a 20 inch deck Green Works 40V mower also with an edger/weed wacker. And where else did I get it but Amazon--another nail in the coffin of brick-and-mortar retail. Greenworks was clearing out their old mower design so I got the mower, weed wacker, two 40V batteries and 1 charger for about $400. I also picked up an additional charger so I could have both batteries charging. The earlier production mowers from 2015 had a recall on them but I'm hoping this one holds up pretty well. The new ones with the brushless motor might have some bugs in them.

It has two electric blades and can take two batteries. When the first one is starting to die, the second one takes over. So I run the mower, which might drain the first battery and start on the second, and then pop the second battery into the edger. Then put both batteries on their respective chargers. Here is an animation they posted (no sound apparently).



Now, I have a pretty small yard. This isn't going to work for somebody with a big property, but it meets my needs. The mower isn't self propelled but since it is light, you don't really need that feature. Now I'm thinking about buying the 40V leaf blower and also a 40V snow blower to replace my Husqvarna I got late in 2014. The dumb thing didn't want to start last winter, and I don't want to go through the "oh shit a big storm is coming, I better get this thing working" rush again. I'm hoping the snow blower would work well despite the batteries operating in cold conditions.

6/4/2017 11:36:31 AM

NeuseRvrRat
hello Mr. NSA!
35376 Posts
user info
edit post

there are pros and cons with both options. li ion batteries are not without their troubles. at least when a carb gets clogged, i can usually clean it out. nothing you can do to those batteries but replace them. a replacement carb for a string trimmer is <$40.

and ethanol free fuel has completely solved all of my fuel troubles on small engines, even if i leave them sitting up over the winter.

6/4/2017 12:06:52 PM

gtherman
All American
628 Posts
user info
edit post

yup. find a place near a lot of boats and they will most likely sell ethanol free gas. it doesn't gum up the carbs and doesn't eat away the rubber hoses like the gas most autos use does. the only reason it is not a problem in autos is because we use them so often that they don't have time to sit for several months and gum up.

6/4/2017 7:44:44 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

I'm going to repost (with some editing) here something I put on another forum in a topic about, "what exactly is in a tune file?" The context is aftermarket reprogramming of Ford turbo, direct injected Ecoboost engines. The person starting that thread had some background in software but I presume didn't have any particular in depth knowledge of engine performance.





The factory engine/powertrain control module of the Ecoboost engines are developed by teams of engineers both in Ford and in suppliers. The aftermarket reverse engineers (or in some cases might use leaked information, nobody knows for sure) it. All the different competing aftermarket systems (Cobb, HP Tuners, SCT, etc) are all modifying the existing controls or making ROM patches to add or change functionality.

Nowadays all the engine controls are model-based. This means that the controls use physical properties of the engine to determine air, fuel, and spark, rather than purely using a look up table of how much fuel to inject or when to fire the spark plugs. Somebody decides the philosophy of the control ("we're going to have a model of the turbocharger, and it will require these elements"). Then the controls are put together in commercial software, almost always Matlab with the Simulink/Stateflow and other such tools. From then it is run in simulation such as hardware/software in the loop, run in a lab test bench, and ultimately transferred from prototype code to something that can be put in a car and into mass production.

The files themselves are a bunch of binary, and the actual "tune" or "calibration" meanschanging series of scalars, 2D maps, and 3D maps that all do different things. When you buy these aftermarket reprogrammers they are reflashing the memory inside the PCM and overwriting values inside those scalars, 2D maps, or 3D maps. They may also insert new code for extra functionality.

A lot of the tunes are the result of "hmm, I think changing this map might work to accomplish my goal (change air/fuel/spark), let's try this and see what it does..." Only Ford has full access to all the block diagrams of how the Matlab Simulink models work, and only Ford has full access to internal reference material on what "knobs" (a metaphor for what tables do which thing) to turn to accomplish a particular goal. A lot of stuff is physics-based these days, so there are physical constants based on for example the displacement of the intake manifold, or the length of the exhaust manifold piping. So there is a model of the mass of air entering the engine, and a model of how much torque the engine is making.

If you want to get a taste of what some of the stuff in the ECU does, read the Cobb Ford tuning guide:

https://cobbtuning.atlassian.net/wik...d+Tuning+Guide

At the bottom are references to Ford patents that they used to help understand the controls.

Keep in mind though that there's knowing the engine side, like how the engine works and what needs to happen inside, and there's knowing the software/computer side.

You can know that the engine needs to be richer, but you have to figure out what to change in the control system to make it actually do that. Take fuel controls for example. There are feed forward elements, like the model of the physical properties of the engine (cylinder filling/volumetric efficiency model), and feed back elements (reacting to oxygen sensor reading). The same ideas apply to turbocharger boost control, spark control, electronic throttle, etc.

In the case of the current Ford controls, there is a complicated model of engine cylinder filling to account for the variable intake and exhaust cam centerlines, also known as variable valve timing. There are also temperature models to account for the change in air temperature in the system. In the old days you could have a couple dumb look up tables. So if I have a fixed camshaft and no turbo, I can look up speed and manifold pressure to determine fueling. Now it's way more complicated than that because the different camshaft positions affect the cylinder filling and the amount of residual gas in the combustion chamber.

If you can tune a carburetor and understand engines, you still need to understand software and controls to some extent to get anywhere. If you understand software and controls but don't know much about engines (air,fuel,spark), you're gonna have a bad time.

7/5/2017 10:00:23 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

Tuning Tools - OEM/Commercial vs Performance Aftermarket

Ford uses ATI Vision as their tool to calibrate the PCM. It's like a really expensive commercial version of these aftermarket tools like Cobb or SCT. For ATI Vision and its competing software (ETAS INCA, Dspace ControlDesk, Vector CANape), they have a sort of generic/universal interface to them, because they can be used for engine related control modules or other kinds of control modules (stability control/vehicle dynamics, transmission, etc).

You the user supply a definition file in the format .A2L . The .A2L file is basically a type of XML that defines memory addresses inside the binary which is .HEX, .S19, or others depending on processor instruction set. The .A2L files define specific scalars (single value), 2D tables, and 3D maps that are part of the program control strategy.

Here's a screenshot of a 3D calibration map from ATi Vision website:



You can see the tab "EGR_dmd1" below. That variable name is defined in the .A2L file. From the name we can guess that it is an exhaust gas recirculation demand table of some sort. This is just a generic thing from the screenshot.

Here's the description of a .A2L file from Vector's website:

Quote :
"Description Files for Internal ECU Parameters

Description files let the user access internal ECU parameters by symbolic names. A2L files are elementary components of every measurement and calibration operation, and it should be noted that the CCP and XCP protocols are address-oriented in their operation. The user selects objects by symbolic name, while CANape takes the associated addresses from the A2L and uses them in communication with the ECU.

Contents of the A2L database

An ASAP2 description file (also called A2L) contains all information on the relevant data objects in the ECU such as characteristics (parameters, characteristic curves and maps), real and virtual measurement variables and variant dependencies. Information is needed for each of these objects, such as memory address, storage structure, data type and conversion rules for converting them into physical units. In addition, an A2L also contains the parameters for communication between CANape and the ECU."


The .A2L file is generated by whoever wrote the controls. Again, modern ECUs are pretty much all running controls from Matlab Simulink. So somebody in Ford or one of their suppliers will dump the .A2L definition of each map and memory addresses. They will also release the block diagrams. [B] Even if you have access to every map and control diagram for the control unit, you still have to know what to do with it.[/B] You can understand how the system works and how the inputs and outputs are calculated, but still not know how to make the system do what you want it to do.

Most OEM's now have entire departments whose job it is to figure out how to tune an engine. They don't actually tune the engine; they just tell the other engineers how they think it needs to be tuned, in terms of what order to populate the maps. For example, on an engine with intake and exhaust variable valve timing, usually you tune the engine in a lab with the cams locked at the baseline "0" position first. Once you have it running with a fixed cam configuration, you slowly expand functionality by populating all the volumetric efficiency characterization, backpressure models, combustion speed models, torque models, etc. That usually requires some special tools (usually in Matlab) so that the engine controls model matches (within some acceptable tolerance) measured values in a lab. So if the engine dyno says the engine is making 100 newton meters of torque, your model will hopefully say that it's making somewhere between 90-110 newton meters, so that information can be fed to the transmission, stability control, etc.



So here's an example of a controls block diagram with special "bypass" algorithm inserted, using another uber expensive tool by ATi (other companies have similar tools). When you get an aftermarket "race ROM" or other such special feature, they are basically doing what's shown in this diagram: they are inserting special controls to bypass internal PCM calculations.

For example, Cobb has a special simplified Speed Density mode for the 2.3 Ecoboost. You can find it if you poke around the website from my previous post. What this special mode does is partially bypass the complex volumetric efficiency models in Ford's calculations. When your engine is heavily modified it is very difficult to modify the existing Ford tables and have the engine run right. So you sort of dumb it down to make it more manageable. You turn back the clock to the 90s basically, because 90s controls are still simple enough that one person can make them work.

But Cobb probably wrote their own interface to do these bypass controls, just like they have their own interface for data acquisition and calibration tables. They aren't going to pay one of these commercial vendors a gazillion dollars. And since Cobb or HP Tuners or whoever doesn't have access to a .A2L file from Ford, the memory addresses have to be reverse engineered. So when Ford releases a new PCM update which changes the controls, the memory addresses move around and somebody has to figure out where the spark tables are now.

So that's more details on the back end.

When you the end user have an aftermarket tool, it's actually easier to understand than these expensive generic interfaces from commercial vendors. In most cases the functionality is the same, but there are situations where the commercial tools are more powerful. For example, the commercial tools, with the right equipment, can sample data at much faster rates. They can also change all PCM variables live, without having to reflash a module, because they use expensive prototype memory modules. But for the cost, you get a very large portion of the functionality in these aftermarket tools, and better ease of use.

As I pointed out above though, even if you had access to absolutely all the software information and all the expensive tools (the software side), you still have to know what to do with it (the engine side).

[Edited on July 5, 2017 at 10:01 PM. Reason : .]

7/5/2017 10:01:05 PM

Hiro
All American
4673 Posts
user info
edit post

That's dope. Thank you for sharing

7/7/2017 1:34:23 PM

moron
All American
33692 Posts
user info
edit post

^^ is it possible for the entire control system to be closed loop?

7/10/2017 12:19:28 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

What do you mean the entire control system? Which aspect?

7/13/2017 1:02:23 PM

tchenku
midshipman
18568 Posts
user info
edit post

Quote :
"When your engine is heavily modified it is very difficult to modify the existing Ford tables and have the engine run right. So you sort of dumb it down to make it more manageable. You turn back the clock to the 90s basically"


Dumb look-up tables ftw!

7/14/2017 10:46:10 AM

arghx
Deucefest '04
7584 Posts
user info
edit post



If you've ever wondered how to tune a car and don't know where to start, this video series is a really good introduction. It's for reflashing a stock Mitsubishi Evo IX ECU using ECU Flash and datalogging with Evo Scan. The earlier videos in the series (it's one log session cut up into 16 smaller videos) are more of an introduction to the software and some of the basic tables.

Starting in this video (earlier videos are useful too), the host goes through a log of a wide open throttle pull in an Evo IX. He talks about looking at ignition timing, air fuel ratio, boost pressure, and calculated engine load and then making changes to the relevant maps in the ECU to make more power or reduce knock sensor activity.

The Evo uses a simple control system with "dumb look up tables" that is pretty straightforward for a beginner, compared to something from the Big 3 or a German automaker. Those manufacturers use more sophisticated models of engine operation and tuning them is less intuitive.

7/22/2017 8:31:26 AM

smoothcrim
Universal Magnetic!
18914 Posts
user info
edit post

what's funny is that mopar just spent a ton of money to back port and OBD1 ECU to their 392 hemi in an effort to sell more crate motors because this process has gotten so complicated

7/23/2017 1:13:47 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

^ Got a link?

7/25/2017 9:25:23 AM

smoothcrim
Universal Magnetic!
18914 Posts
user info
edit post

http://www.hotrod.com/articles/factory-plug-n-play-gen-iii-crate-hemi-engines-and-install-kits-from-mopar/

there were several eps on graveyard carz where they built the car to debut this setup at SEMA one year

7/25/2017 2:25:37 PM

smoothcrim
Universal Magnetic!
18914 Posts
user info
edit post

http://moparonlineparts.com/mopar-performance-crate-hemi-engine-wiring-p-7345.html



that is not a modern harness/ecu

7/25/2017 2:36:05 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

That is similar to the product GM offers for LS3 crate motors. It's a production ECU but all the OBD is turned off as well as many other limits. Basically it's not much different than an in house engine dyno calibration. These OEMs have databases with a list of all the stuff to change. Basically you click a button and most of the work is already done in software. Somebody probably went through and massaged it a little.

I talked to a GM engineer about this a couple years ago at their booth display for LS3 kits. So it's probably similar.

That harness is modified to work with basically the fuse box they include. The kit makes it a lot easier to install and wire up.

7/25/2017 9:25:00 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

That IS a topic that deserves a separate posts. How do you take a factory, production PCM calibration & tune and make it work on an engine dyno or in a similar vein, in a swap scenario?

The first thing that has to be done is to turn off the OBD monitors, the stuff that makes the ECU go into failure modes. Usually the easiest thing to do is to turn off the "master switch" bit for each monitor. So, take the monitors that typically show up on a generic scan tool:

Catalyst efficiency monitor (bad cat or not)
evaporative leak monitor (the one for the loose gas cap)
misfire monitor (when your spark plugs or coil are dying for example)

Usually each one of those has a dedicated memory address bit to enable or disable. So you can set a 1 to a 0 and that's it. There are other ways to turn off subroutines in the software. For example, there are usually permission conditions. So for example, if x mode turns on when water temperature reaches some minimum, you just set to like 300F so it never turns. Or there is some kind of authority setting for correction. If you want to turn off some kind of feedback routine, set the authority (max amount it can adjust) to 0. If it's a limit, like a vehicle speed governor, just set the limit to something you would never achieve, like 300 mph or some maximum hexidecimal value "FF". It depends how the software works. Maybe there is a delay timer to enable the mode, and you set it to 999 seconds or whatever so that it doesn't engage.

Getting back to sections of the calibration there are a lot of emissions related things. Evaporative purge solenoid (Charcoal canister related) is one of them. There is probably a fuel pump speed control that needs to be turned off or ignored. There is usually some kind of catalyst protection mode that is turned off. There may be a catalyst temperature model or delay timer to go into enrichment that needs to be disabled. Rear oxygen sensors are mostly used for the catalytic converter and small emission related fuel corrections, so that can be turned off. Any other kind of after treatment like secondary air injection, particulate filter, etc would need to be disabled.

There are probably some kind of torque limiting logic, some kind of look up tables that need to be disabled. If the engine is expecting a transmission signal over the CAN network (from a modern automatic rather than a manual or some 20th century mechanical thing like a TH350), it needs to have that portion turned off so it doesn't freak out when it's expecting something related to a shift.

Maybe some of the temperature or pressure sensors aren't in use for that application. Say the engine has an oil temperature sensor that is deleted; any kind of logic that uses oil temperature needs to be disabled.

Some of the controls may have to be dumbed down or bypassed, as mentioned above, in the sense that they are no longer in a carefully tuned environment that accounts for specific exhaust backpressure etc, but needs to be "close enough" to account for a variety of intake and exhaust setups.


Those are the basic categories of changes made to have a factory control unit on a modern engine work in a way different environment such as an engine dyno or a swap. Again whoever is making these kits have access to some basic list of most of this stuff and can import the changes right in there seamlessly.

7/25/2017 9:45:59 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

So I just picked up a 2017 Challenger R/T with a 6 speed manual. The 5.7L on this car really feels a lot like an LS1, and I'd call this car a more refined version of an LS1 Camaro/Firebird.

Considering the 5.7L was very clearly patterned on the LS1/LS6, it's no surprise that, just like my 2005 CTS-V before it, this car has a 1-->4 "skip shift" feature that is found on LSx powered cars. For those of you who have never experienced it, it basically forces you into 4th gear when you're not really looking to go into 4th gear. So I'm going to use the eliminator, which is just a plug with a resistor.

Here's what I find interesting about skip shift. I'm really surprised the EPA and CARB have allowed it. It's such a blatant workaround to the LA4/Federal Test Procedure cycle and the onboard diagnostic is OBD I level. From the EPA website we can see the basic test cycle on a chassis dyno:



and then the skip shift logic according to the owner's manual:

"Engine coolant (antifreeze) is higher than 106°F (41°C), vehicle speed is greater than 19 mph (30 km/h) but less than 21 mph (34 km/h), and the transmission is in first gear, and the accelerator is at 1/4 throttle or less. The 1–4 Skip Shift indicator message will be displayed during these times."

Looking at the test cycle, it's pretty obvious that by the second or third acceleration the engine will be warmed up enough (it starts at 75F), this skip shift solenoid is going to activate, and the certification driver is going to be forced to put it into 4th every time. And yet, I can pop a resistor in there and the OBD system won't throw a code, just like a pre-1996 OBD I era car, even though it probably has a significant impact on CO2 emissions if it is turned off. Yet if you tried such a resistor on other devices the ECU is going to be smart enough to throw a code anyway through a thing called "Rationality monitors."

I went to the EPA website and did a quick check through the certification application back in 2009 for the original model year, and there's nothing in there about the skip shift solenoid among the control devices. They list important sensors and switches but the skip shift solenoid doesn't seem to be among them. Just type in 2009 Challenger and sort by application. The document you want is "Title: Application for Chrysler LLC 2009 model year test group 9CRXV05.71P0" specifically for the manual transmission 5.7L Challenger. Note that you don't want an "update" document.

https://iaspub.epa.gov/otaqpub/

7/29/2017 8:49:00 PM

Hiro
All American
4673 Posts
user info
edit post

That's cool! I'm assuming that with tuning, you can deactivate the rear 02 sensors as well? I don't have a CEL in my vette, but it's blatantly obvious I'm running a tune (especially at WOT). Im surprised that's allowed as well...

[Edited on July 31, 2017 at 10:56 AM. Reason : .]

7/31/2017 10:56:00 AM

arghx
Deucefest '04
7584 Posts
user info
edit post

If its an aftermarket tune you can turn off all sorts of stuff and still trick the computer to say you are passing emissions, but it depends on the car.

8/1/2017 1:16:29 PM

arghx
Deucefest '04
7584 Posts
user info
edit post

As I write this, the Tesla Model 3 has "entered mass production" and had its first deliveries, which are 30 cars going straight to employees.

This is genius, and here's why.

Tesla promised that they would go into mass production this summer. So what they did is, they called the "pre production" build mass production, and everybody cheered. Wall Street is happy, journalists are happy, and Tesla can keep testing cars until they actually send them out to real customers.

How it's supposed to Work

Let's look at a [highly simplified] development cycle of a vehicle, mostly from the perspective of engineering and testing with less emphasis on styling and manufacturing. I'm going to divide it into four phases: Planning phase, Alpha phase, Beta phase, pre-production phase. You can certainly do it more than one way, but generally speaking, the more "new" the vehicle is, the more phases and the longer each phase may be, unless somebody decided to rush it to market, usually an executive who has a bonus on the line.

Planning phase: This is where you set targets and do things in the virtual realm. So for targets, you decide what kind of vehicle is it, what does it need to be able to do (acceleration, fuel economy/range, towing, safety rating, tailpipe emission if that applies etc). Then you do various forms of simulation, starting out with basic spreadsheets all the way up to complicated computational fluid dynamics. This phase may start 3 to 5 years before the vehicle reaches production, but it could be longer or shorter.

Alpha phase: This is where you start getting into the physical realm. You might have a mule vehicle or proof of concept, where you modified some existing platform to give a rough approximation of what you are trying to do. You may have components running individually on test benches as well. Generally this is a proof of concept phase, and most of the vehicle is pretty ugly and hacked up. Think oil leaks, mismatched body panels, just basically a shitty unfinished project car. This phase might be 2 years before mass production

After a certain amount of time the physical testing needs to be "done" to lock down the design for the next build phase. These prototype parts tend to cost a lot of money. Or you may have a small production run of 3D printed parts, some of which are great and others which are totally unreliable.

Beta phase: This is supposed to be a lot closer to a "real" vehicle. Components come together into a vehicle that is getting close to a mass produced design. The vehicle may not meet all the targets, and it may not drive very well, but it's starting to look like a real car hopefully. Much of the vehicle is hand built with special tooling. The vehicle is some evolved version of the Alpha design and could be significantly different from what was envisioned in the planning stage. At the end of a Beta phase you might be 1 to 2 years before mass production, and components that require a long lead time need to be decided.

Pre production phase: This is when the really expensive tooling has been finished and the car is assembled in small batches. Once the vehicles are built, the engineers start going through a bunch of testing while the manufacturing side is trying to get the production process figured out so there is consistent quality. The main hard points of the vehicle are done, and at this point they are tweaking software to get the customer experience right or to meet a regulation in some market. Then the vehicle goes into continuous production and there may be some small changes during the model year to fix little problems.

Ok, so that's how it's supposed to work.

How it can actually work

So the above paradigm often gets pretty messed up. In the planning phase, you get tons of really optimistic simulations and what I call "magic spreadsheets" that make the engineering targets look a lot more achievable than they actually are.

Once you get into physical prototypes, a lot of corporate stuff happens. You get conflict between the OEM and the supplier in terms of timeline and who is responsible for what and what the real goals and targets are. You get working level people who are under a lot of pressure to, I don't want to say "hide," but certainly minimize all the things that can be going wrong to the decision makers.

Often time a program will get delayed or extended. However, a 3 year development program extended to 5 years is not the same as a program originally intended to be 5 years. Why is that? Well, during the early phases the designs haven't been locked down and long term contracts haven't been signed. So if you have 3 years of planning and Alpha level development, you can churn through a lot of designs. But if you get stuck in pre production and keep delaying, nobody wants to change anything anymore because big changes cost millions of dollars. So you get mediocre product pushed out the door.

In my opinion, there are two main root causes for these "3 years becomes 5 years" programs:

1. Somebody wants to get a product to market fast because to help their career or get a bonus
2. Nobody wants to tell the emperor that he's got no clothes.

The working level people know when it's a bad product. They know when they are making a piece of shit. Eventually though it turns intp the July Crisis (assassination of the Austrian heir) that preceded World War I... everybody gets caught up in a rapidly moving series of events and nobody feels like they can stop the situation.

What Tesla Did

Now that we've covered how things are "supposed" to work and what can "Really" happen, let's use that to understand how Tesla subverted the normal development process. What Tesla has done is called the pre production phase an early form of mass production, when the real volume production won't happen for months. I can guarantee you they are doing all sorts of testing and analysis to these new vehicles behind closed doors, because all the cars went to employees. Since they have telematics in every vehicle they can collect all sorts of data remotely and nobody would even realize they are running tests, and the lack of a combustion engine simplifies things greatly.

Now somebody from Tesla would probably quibble with me and say that it meets all the requirements, the vehicle is in the hands of journalists, etc, and this is a mass production car. And from their perspective, they are right. Personally I'm drinking the Tesla Kool-Aid for the most part, I'm just pointing out how they subverted the industry standard development process to achieve their goals.

Genius I tell you. Somebody pour me another glass of Kool-Aid.

[Edited on August 6, 2017 at 8:10 PM. Reason : .]

8/6/2017 8:09:26 PM

 Message Boards » The Garage » arghx's technical ramblings thread Page 1 2 3 4 5 6 [7] 8 9, Prev Next  
go to top | |
Admin Options : move topic | lock topic

© 2024 by The Wolf Web - All Rights Reserved.
The material located at this site is not endorsed, sponsored or provided by or on behalf of North Carolina State University.
Powered by CrazyWeb v2.38 - our disclaimer.