MS3 pre-1.4 alpha 1 boost control issue

[Suprazz]

Hi,

Here is my problem. I would like to understand why do I have boost control duty spikes when I'm over the target boost control.

My setup is "special".

I'm running a Toyota 2JZ-GTE with an holset HE351VE VGT turbo and a wastegate. I'm controlling the vane of the turbo with the stock control board over CAN with an other board. This conversion board accept 0-5v input and send CAN messages to move the vane depending on the voltage. 0v = valve closed, 5 = valve opened. I'm using the boost control of the megasquirt to control the vane with a PWM output that I pass in a RC filter so I convert the pwm output in 1-5v signal. When the valve is fully opened, the wastegate open with an independant boost controller.

So the goal is to close the vavle (to around 1v, 65% duty cycle) and wait untill I get the desired boost and then I start to open the valve to end a 5v (0% duty cycle).

But I dont understand why, the boost control, glitch few times at high revs and this make the valve close.

So sensor01 is the voltage after the RC filter.

vgt1.png

settings3.png

Thanks!

[Matt Cramer]

Can you attach the complete MSQ?

[Suprazz]

yes!

[jsmcortina]

Looks like your boost control is now massively over-sensitive. If you look at "boost duty" you'll see that it jumps between 65% and 0%, corr

What is the CAN control setup?
I'm wondering if there's a way to build a "generic CAN broadcasting" feature that would allow you to command it directly from the MS3, taking out the PWM -> analogue -> CAN convertor and the compromises that are entailed with that.

James

[Suprazz]

Hi James,

Yes I noticed that. My boost control setup is always jumping from 0 to 65%. Sometimes but rarely, it jump to ~30%.

I dont really know why. Maybe because of the D parameter. But this is the only way I was able to have a good reaction from the turbo when it spool. If not, it react too slowly.

About the CAN from the MS directly to the turbo, yes it would be the best thing to do but the problem is the protocol to control the turbo is proprietary and there is no information anywhere. I could try to do it but I dont have the equipement to.

So do you think my problem is a boost control issue? I'm over the target boost so it should stay to 0 and not try to oscillate even if the boost drop a little bit?

[jsmcortina]

Have you tuned the bias table for the typical duty required that boost level vs RPM ?
I'd certainly try reducing the sensitivity slider significantly - I'm almost certain that that excess sensitivity is causing the steps from 0 to 65%.

James

[Suprazz]

No, really. I'll try to find tune it a little bit.

I'll also try for the slider. Will see how it affect the boost control...

Thanks

[elaw]

I agree with James that the sensitivity is too high.

But you may have another problem too... your R/C filter is introducing quite a bit of delay on the boost control output. Look at the first time in your log that the boost duty transitions from 65% to 0%... sensor1 takes around half a second to get close to its final voltage.

I'd suggest making the boost control output frequency as high as possible (1021 Hz?) and the "R" and "C" in your R/C filter as small as possible in order to reduce control lag. That lag will tend to make the boost control oscillate and will make tuning much harder.

[Mr_Random]

I am very interested in using this turbo, I got to reading online and found a small google drive folder containing the CAN messages from stock ECU to TCU (they are calling it Turbo Control Unit), the guy that made this used Arduino to communicate with it, from I snagged all the files from the cummins forum... I'll find the link if I am allowed to post, but it is searchable quite easily. I'm hoping this can simplify the control and implementation of this very nice turbo My 22re should spool it nicely and achieve great mileage.

In any case, if ms2/3 could control the beast with simple code and a couple wires, I would be very happy to help figure it out. I have been wanting to learn a little more coding...


I have also heard that this turbo is kinda loud.


WHOOOOSH

[jsmcortina]

Go ahead and post a link.

James

[Suprazz]

I reduced the RC filter and increased the frequency. I also moved the slider for the boost control. Now I have a way better pid but it's still not perfect.

I also found documentation about the CAN messages for the HE351VE controller.

This company also made controllers for this turbos: http://www.axiomatic.com/electronic-val ... oller.html

[jsmcortina]

I looked at the spreadsheet, but I can't make any sense of it. Is position measured in centimetres or ???

James

[Suprazz]

it's the exhaust AR size.

from 3cm² to 25 cm². Equivalent of around 0.18 to 1.83.

The turbo is not moving in a linear way. There is 16 steps where you can move the exhaust vane to.

[Suprazz]

and the can protocol is J1939.

[Mr_Random]

https://drive.google.com/folderview?id= ... sp=sharing

should be all the info... hope it helps!

the original forum topic:

http://www.cumminsforum.com/forum/94-98 ... ntrol.html

[jsmcortina]

That topic makes it sound easy.

If I added support to broadcast those CAN packets directly, would anyone be ready to test it?

----------------------
Technically, from the linked forum topic
29bit CAN message 0x0cffc600
8 bytes
D0 = 0 - 0xff for how far open the vanes should be. 0xff is all the way open.
D1 = 0x02
D2 = 0x01
D3 = 0xff
D4 = 0xff
D5 = 0xff
D6 = 0xff
D7 = 0xff
----------------------
29bit messages could easily clash with the non-standard Al-CAN that Megasquirt uses, but in this case, that header translates to:
29bit CAN header = 0xcffc600
var_blk = 0x0
to_id = 0xc
from_id = 0x8
msg_type = 0x7
var_offset = 0x33f

So, provided there is no device on CANid 12, it should work ok.
The 8bit vane position matches up with the 8bit boost duty in MS2/3 so ought to be easy to output.

So, I think I know what needs doing and it shouldn't be a big deal to add it, but I'm holding fire until someone will actually use it.

James

[Suprazz]

I'll try for sure if you do it.

[Mr_Random]

Buying a turbo this weekend hopefully I'm using ms2 currently.

[Mr_Random]

I have my turbo and will be buying/building parts to mount it in the coming month. For now I can help implement control through bench testing as soon as the CAN protocol is set up! Some basic code to just cycle Vane positions would be great to start... Turbo is in good shape but would be useless to test beyond that if the Vanes are stuck. I'll pull the turbine housing (wanna clean and clock it anyways) in a couple hours, hopefully I can tell how the vanes are like this, if not I'll have to wait for you to test.

Getting Excited! I need to read up more on boost control theory now!
Just pulled it apart and seem to have gotten lucky with the price I paid! This thing is really young... The turbine came off with minimal issue to reveal some decently shiny blades and fins, very little soot buildup. I'll clean the housing tomorrow, but so far I think we are good to go with it!

[jsmcortina]

I've added this broadcasting to the development code ready for the next alpha. If anyone wants it sooner, please PM me your email address.

James