[info] electronic throttle body control ETB

[AndreyB]

https://svn.code.sf.net/p/rusefi/code/misc/logs/20170529_passat_etb.7z

ETB Red/VIO #3 wire - OutB on the red module
Brown/VIO #5 wire - OutA on the red module

http://rusefi.com/wiki/index.php?title=Electronic_Throttle_Body
http://rusefi.com/wiki/index.php?title=Vehicle:VW_Passat_2002_1.8

[kb1gtt]

Can you do manual duty control and fill in the measured position for the below, starting from top and going to bottom, or vice versa.

Duty%, measured%
0,
20,
40,
60,
80,
100,
80,
60,
40,
20,
0

I expect you will find 20% on top and 20% on bottom are different, caused by hysteresis in the system.

[AndreyB]

duty,position
0,0
10,0
20,0
21,0
22,3
22,8 (15 seconds later)
23,12
23,20 (a while later)
23,23 (later)
24,24
24,28 (a while later)
24,32 (later)
25,37
25,41
25,40
26,58
27,71
27,75
28,79
27,83
27,92
27,100
20,100
15,100
10,44
8,0



See also log file
MegaLogViewer is the software to view
You are interested in TPS, pedal and "debug f1" columns
https://svn.code.sf.net/p/rusefi/code/misc/logs/2017-05-30_21.47.35.msl_etb.7z

[kb1gtt]

I intended for a non PID duty. Basically set the PID output manually to a fixed number then see where the valve goes. I recall for the alternator, you could set the PID to 0, then you could adjust the offset, such that the offset directly controlled the output.

[AndreyB]

My response above has two parts: non-pid table you've asked for followed by a log of the non-pid in the beginning and some pid later.

[stefanst]

It appears that the first table is missing a crucial component (position)

[AndreyB]

It appears that the first table is missing a crucial component (position)

How so?

duty,position
duty=0,position=0
duty=10,position=0
duty=20,position=0
duty=21,position=0
duty=22,position=3
duty=22,position=8

[kb1gtt]

What was this about? I expect the fixed duty to happen very fast.

(15 seconds later)

[AndreyB]

What was this about? I expect the fixed duty to happen very fast.

(15 seconds later)

This is about your expectation not being met. Because your expectation was not met I felt that a log file with additional information (time axis) would be beneficial.

I am manuallally applying duty cycle but the position keeps moving, in a few burst over time. I believe looking at the log file could really provide a lot of info, do you think you would be able to use the MegaLogViewer with the linked log file with the abovementioned three data points?

[kb1gtt]

I can look at it this evening. I would guess you still have a wiring problem. Can you solder your wires? Also how many amps is it consuming?

[AndreyB]

http://rusefi.com/wiki/index.php?title=Manual:Debug_fields

[puff]

Lacks more details. How to save them? How to choose them?

[AndreyB]

Here's a little problem: constant duty cycle input does not produce constant output with these ETBs, at least with the specific Passat unit. I've actually tested two units - same model, similar behavior. 26% duty cycle produces different throttle position.

[Abricos]

Do you have another (stepper driver )?
you can try another driver ...
I KNOW YOU HAVE ...

[puff]

second hand units? those might be worn out?

[AndreyB]

Googling "electronic throttle body control strategy" i guess

[AndreyB]

Some specific numbers in https://pdfs.semanticscholar.org/eefc/39d102845db14d8b6953e587945f383c8201.pdf

Also I-PD is mentioned quite often.

[kb1gtt]

I see mention of dithering to avoid issues with hysteresis. I've mentioned that before, and strongly suggest adding it. As well I suggest either an offset, or even better would be an offset table, such that you manually control the valve, to find an ideal value, which is added to your PIDO, such that PID are only dealing with the error in the system, and your O is doing the heavy lifting.

[AndreyB]

which is added to your PIDO, such that PID are only dealing with the error in the system, and your O is doing the heavy lifting.

I guess my next step is creating a tool to create some O table automatically.

[Flying]

On the subject of throttIe body control, find this exciting:
https://www.facebook.com/Andytronic/videos/10154789793759579/

[AndreyB]

Different approach - https://www.pololu.com/product/1392 Jrk 21v3
Nikita has it, It uses MC33926 driver and something like AVR PIC18F14K50 chip for the logic. No hum or whine while usng 20KHz frequency.

[kb1gtt]

I see the TTL on MC33926 is 3V compatible. As well the chip itself is only $4.88 usd at the below in qty 1.
https://www.digikey.com/product-detail/en/nxp-usa-inc/MC33926PNBR2/MC33926PNBR2CT-ND/3524219

It appears that if we were to attempt direct control over this chip, it could require as little as 1 signal wire. Such that if signal is high, then motor is opening, and if signal is low the motor is closing. Optionally it could contain 2 digital control signals, which would allow for both sides of the H-Bridge to be put to 0V an this causes the motor to go into a braking operation. Also optional is a return signal for analog amps of the motor.

If this chip is used, I would likely suggest powering it via P902 on the Frankenso, as that would limit the voltage spikes which could damage this chip if the input voltage exceeds 28V.

Soldering a QFN package is harder than soldering flying leads. Soldering this chip reliably is harder than what we currently do. As well thermal expansion of a QFN is more delicate, as the flying leads are not there to compensate for thermal expansion stresses.

[puff]

yey! ten times less than the actual pololu board!
but what is the nature of that sound anyway? why is this chip so unique that it doesn't produce this noise? increased switching speed?
as for soldering, I've seen people do it putting QFN chip upside down (for prototyping). Though, if it is intended to get hot during its operation, this method probably won't do.

[kb1gtt]

Yes increased switching speed, the 20KHz is hard to hear, and is determined by the external MCU processor. If we direct drive we can bring that noise back. Many consider that high pitch noise a handy diagnostic tool. The high pitch noise is hard to hear when the hood is closed.

This chip will make significant heat, and will need the center slug connected to a heat sink.

[AndreyB]

pololu board overall is a nice intermediate solution - I can play with it as a baseline of what would be nice to repeat with rusEfi own firmware.

[mzk]

If external controller should be implemented for the ETB, why not servo signal - 50 Hz frequency (may be higher) with duty cycle from 1 do 2 ms, meaning, 1 ms = 0 degrees, 1.5 ms = 90 degrees, 2 ms = 180 degrees?

I got lost in the different ETB types - servo, stepper, ...

Also I stand up for only driving the valve position, not to change the whole calculation based on the accelerator pedal position. At least for street use.

Is there a wiki page dedicated to the ETB?

[AndreyB]

Is there a wiki page dedicated to the ETB?

There is a very limited set of notes https://rusefi.com/wiki/index.php?title=Electronic_Throttle_Body

[AndreyB]

[kb1gtt]

What happens if you push lightly with your finger? AKA I think you need the I term, not when the engine is off, but when you have variations on force on the butter fly, caused by moving air.

[andreika]

Could you please make a video of rusEFI-controlled ETB with P=1,I=0,D=0,offset=0? Or, even better, with two ETBs side by side?