Holset VGT Electronic Controller Project

[Mistaken]

Yah dont ruin a good turbo Whatadsm, I'll have a spare actuator to dissassemble early next week.

BTW did it move to a home position at all when you powered it up? I read that there is a safe mode it goes into when it doesnt pick up a valid signal that moves it to a specified location.

Also i dont think you could go into production with a control unit if it wasnt CAN based, unless you sell the turbo and controller together since it will most likely have to be opened up and modified. So CAN still might be the only route.

Whatadsm...any chance you have access to a CAN controller to see what messages it is pulsing to the ECU?

[blownhemi]

Actually, frequency might be the key... if the ST is receiving the PWM with a quick A/D unit, too low a frequency will either charge the sample and hold capacitor in the A/D to the high or the low level. Frequency would have to be magnitudes higher than the A/D samping time to get the mean value of the signal.And if it's using a timer to measure the PWM signal, it could overflow during the period of no voltage level change in the signal, if the frequency is too low. I'd try higher frequencies, I don't think it can hurt the pins. Keep in mind, even if you transmission with the bitrate of CAN, you COULD make CAN bits, but definitely not a CAN frame, with valid fields, and a mathing CRC.

Did you send in a differential signal, or just a normal signal on CAN-hi?

I really hoped something would come of this, although I thought all along that the easier solution for Delphi would have been to have the PWM lines seperate, and easier for Holset not to lead a wire out for those pins.

As a side note, how do you know, which are the power, ground, can-hi and can-lo wires? Is it written on the housing somewhere (sorry, I haven't yet seen this turbo up close and personal) ?

[Mistaken]

There is a wiring diagram floating around a few forums for it. Outside pins are power and ground, inside are CAN Hi and CAN low.

[blownhemi]

Are you sure about that? By this diagram, it seems more like first two connectors are power and ground, second two are CAN-Hi, CAN-Lo. The wiring diagram numbering isn't in the same order as the depiction of the connector front face below...

[WhatsADSM]

Yah dont ruin a good turbo Whatadsm, I'll have a spare actuator to dissassemble early next week.

BTW did it move to a home position at all when you powered it up? I read that there is a safe mode it goes into when it doesnt pick up a valid signal that moves it to a specified location.

Also i dont think you could go into production with a control unit if it wasnt CAN based, unless you sell the turbo and controller together since it will most likely have to be opened up and modified. So CAN still might be the only route.

Whatadsm...any chance you have access to a CAN controller to see what messages it is pulsing to the ECU?

That is not exactly true. It very common for all mechanical systems which do not have an absolute encoder to home to torque or travel limit or something. That way you have a known home position along with an encoder count. Basic premise... that is totally normal operation even if the unit DID connect.

Ultimately CAN IS the best route, although it will likely be the most difficult.

I am in the middle of doing some CAN disection work on it now. I have some data on it, but will know more later today... just stuck at work now.

Actually, frequency might be the key... if the ST is receiving the PWM with a quick A/D unit, too low a frequency will either charge the sample and hold capacitor in the A/D to the high or the low level. Frequency would have to be magnitudes higher than the A/D samping time to get the mean value of the signal.And if it's using a timer to measure the PWM signal, it could overflow during the period of no voltage level change in the signal, if the frequency is too low. I'd try higher frequencies, I don't think it can hurt the pins. Keep in mind, even if you transmission with the bitrate of CAN, you COULD make CAN bits, but definitely not a CAN frame, with valid fields, and a mathing CRC.

Did you send in a differential signal, or just a normal signal on CAN-hi?

I really hoped something would come of this, although I thought all along that the easier solution for Delphi would have been to have the PWM lines seperate, and easier for Holset not to lead a wire out for those pins.

As a side note, how do you know, which are the power, ground, can-hi and can-lo wires? Is it written on the housing somewhere (sorry, I haven't yet seen this turbo up close and personal) ?

Like I said before I put the correct differential CAN dominant high signal on the bus.

IMHO I'm not going to spend my time trying every PWM frequency to try to get it to work. If it was being sampled through a RC and ADC I'm not sure how it's own can messages wouldn't get in the way... or other messages on the bus for that matter. ?! when would it know to sample? The more I think about it, although I would love for it to be a simple PWM signal I just don't think it is in the cards. It might however be in the firmware and just not brought out to the hardware. Again I am not willing to look at that quite yet until I am sure the CAN won't lead anywhere.

I am also POSITIVE i have it wired correctly. It is talking, and I am sure of it.

There is a wiring diagram floating around a few forums for it. Outside pins are power and ground, inside are CAN Hi and CAN low.

Can't see the picture at work but the CAN lines are NOT in the middle with power and ground on the outside. Can lines are next to each other on one side with power and ground next to each other on the other side.

[Mistaken]

Yah i didnt look at the Diagram before i posted it, its not a pinout just a color reference. So your right on the pintout. When you get home and can see the picture it tells you which wires are what colors just to be 100% sure.

[WhatsADSM]

Yah i didnt look at the Diagram before i posted it, its not a pinout just a color reference. So your right on the pintout. When you get home and can see the picture it tells you which wires are what colors just to be 100% sure.

Yea its probably the same one I have... If so there should be pin numbers by the connector blocks.

[Mistaken]

Yah i see em.

At the rate your going your gonna have it working before i even get my turbo...

[WhatsADSM]

Yah i see em.

At the rate your going your gonna have it working before i even get my turbo...

Oh... I dunno about all that. I am going to give it a shot over J1939. However from what I have seen thus far it might not be too easy or standard. Maybe I can try to either give holset or delphi an email and ask nicely if they will divulge any information on their J1939 application layer... Otherwise I will have to find a local with a truck and do some sniffing. What model dodges do these things come on?

[Mistaken]

the 07-09 dodge 2500 and 3500 i believe, with the 6.7l cummins.

[blownhemi]

What are you using for a J1939 controller? I wonder what info it is sending out over CAN. According to the delphi site, it could be the position feedback. Which is probably proprietary... in that case one of the PGN ID's around it, +1 or -1, might be the position command.Do you have a copy of the J1939-71 SAE paper?

http://www.rx7club.com/showthread.php?t=810685 - post #9, seems someone already asked them, but that shouldn't keep you from trying. If you ask them, could you somehow sneak in a question whether or not they use the same wires for CAN and PWM, like you need it for, say, connector design/choice? CAN is of course the cleaner and more fit-for-an-engineer solution, but just in case...

[Mistaken]

If you guys find a contact over at delphi or holset, I can always play the "I'm an engineering student working on a class project" card.

I'll ask around at work if anyone has a friend/contact at either of the 2 that will talk to me. ITW is a big company, someone has to know someone.

I agree also that CAN is the most fit for an engineering solution, but it may not be the most practical unless we can find someone with a truck. I tried looking around for a used ECU/wiring harness but they are impossible to find. The only motorset I found for one was nearly 8000, so that option is out.

We could always try to make a simple inline box that would monitor the signal between the ECU and Turbo, run a program such as CANalyzer on it or something to record the traffic.

[WhatsADSM]

I played with it a bit yesterday and yea they are proprietary commands. It ignores the standard stuff from J1939/71.

That actuator is a frustrating little bastard. Funny enough I asked it if it supported some of the proprietary stuff that I see it spitting out on the wire and it said no... ?! But when I asked it its address then it gave it to me happily. Almost seemed like it didn't want to tell me too much about what it knew.

I wrote a basic dissector for the J1939 stuff last night so moral of the story is I should just need a truck now and do some logging. Hopefully those trucks don't have too much else on the CAN bus, otherwise that would definately make my life harder.

Hmm, hemi that's not a half bad idea... I guess without a truck I could give it a shot to write something to the write to PGNs around its proprietary stuff... Or to scan all proprietarys. Who knows maybe the brute force method would actually yield some results?!

FWIW I emailed delphi/holset, although I fully expect to get nothing back in return.

[l0nestar]

*subscribed*

[Mistaken]

What kind of hardware are you using to do the CAN testing?

What if you were to find a dealer that would let you take a truck home for the night to "show your wife"/"see if it fits in the garage" ... and either back probe the CAN lines from the ecu to the turbo, or make an adapter to plug inline and interrupt it.

I personally dont know anyone with a diesel truck that new. The only person i know with a cummins drives one thats pre-VGT.

[blownhemi]

Did it send a lot of different proprietary PGNs, or were there just a few that it kept repeating?Yeah, it's completely up to the implementer what parts of the standard they support, nothing is mandatory. IIRC not even the address claiming.

Mistaken, the fake buyer idea is actually not bad... feel like expanding your horizon into social engineering as well? The worst that can happen is, they say no.

[Mistaken]

The only problem is that im 20 years old. Most car dealerships wouldnt take me seriously. Also I dont have any equipment to log the CAN communications even if I was able to get a truck.

Maybe a quick post on the Diesel forums will find us someone close to either myself or WHATADSM a little north from me. I'm sure one of the guys would be willing to let you back probe into the wiring as long as you dont damage anything, even if I have to throw him a few dollars for his time.

[Mistaken]

On a side note the spare actuator i bought shipped out this morning. It's coming from texas so i should hopefully have it early next week.

[WhatsADSM]

Yea I am a member of a local car club with a fair amount of vehicles. I made a post there, and hopefully someone will be able to help, but no luck thus far.

I am honestly not *that* attached to project to actually screw around with a dealer to see if I could borrow the truck for a day and then say I am not interested. I just remembered though that I have an old friend that I thought bought like an 05 dodge diesel a year or two ago. Maybe if I am lucky I'll call him up and it will be the right model, although it sounds like it might need to be a newer model. Otherwise the guy that I am investigating the turbo for works at a roofing company and I kind of wonder if they have any of those trucks, again probably not too likely since it is so new.

[Mistaken]

I'm working on a couple big leads for trucks right now. I'll let you know in a few days. The main thing is I just need to figure out how to get the information needed once I get a truck.

Will hooking up a simple J1939 CAN microcontroller in line with the turbo and ECU be enough to recieve the commands? Or is there a special type of controller/program I will need?

I really am new to this aspect of programming, being that I am a ME, my only background is Parker Hannifan plc programming language and some ladder logic programming.

This whole project is gonna be a huge learning experience from me, and is a slightly daunting task, which is why I appreciate everyones help with it.

[WhatsADSM]

I'm working on a couple big leads for trucks right now. I'll let you know in a few days. The main thing is I just need to figure out how to get the information needed once I get a truck.

Will hooking up a simple J1939 CAN microcontroller in line with the turbo and ECU be enough to recieve the commands? Or is there a special type of controller/program I will need?

I really am new to this aspect of programming, being that I am a ME, my only background is Parker Hannifan plc programming language and some ladder logic programming.

This whole project is gonna be a huge learning experience from me, and is a slightly daunting task, which is why I appreciate everyones help with it.

To be honest it sounds like you picked the wrong major. It seems daunting for that reason.

Embedded networking type stuff like this ideally you would have a solid background in embedded firmware and a decent background in electrical to make sure you know what you are doing. I am a firmware engineer by trade (with a lot of my focus on real-time networks) , but I also have a decent background in hardware. So that is why it seems like I am moving along rather quickly.

Electrically to connect you will star off the CAN bus. (read installs like a "T"). You can't just go buy a CAN micro and hook it up to the bus though. There is some interfacing you will need to connect directly on to the CAN bus, not to mention all the rest of the hardware needed to get a "from scratch" micro up and going.

I would suggest buying a pre-built CAN converter MODULE of some sort. If you have cash you can probably get one that will have some J1939 software support as well and just concentrate on the software/firmware for now. IMO trying right now to start from scratch with a microcontroller is a bad idea.

[WhatsADSM]

P.S. if you can find a lead in or near Milwaukee please let me know.

[Mistaken]

Yah, thankfully i have the help of a few very good electrical engineers at work. They will be helping me a great deal once i get the turbo and spare actuator in.

I'm personally gonna work on getting PWM to work on my own just as a good learning experience because i know if i can figure out the pinout I can personally do the programming. The EE's at work are going to be helping me with the CAN communication aspect.

We have a few leads on getting a truck to do the research we need. Although I have no guarantees yet. If someone comes through with a truck and I can get it up to Milwaukee I will let you know.

[RustspecS13]

Wow its nice to see some people collaborating on a VGT controller!

Ive been eyeing these VGT's for a few months, and if I did it myself Id probably modify a giant RC sail servo. Basically put it on steroids. And its easily controllable with a PWM signal. 0-5v The only issue would be an automatic controller but with simple pwm that would be much easier.

I'm also considering a compound turbo setup, taking another page from the diesel guys! Check out this 2jz+ compound turbos in a z31 :http://www.youtube.com/watch?v=JbFr_BW_278

~Alex

[WhatsADSM]

Wow its nice to see some people collaborating on a VGT controller!

Ive been eyeing these VGT's for a few months, and if I did it myself Id probably modify a giant RC sail servo. Basically put it on steroids. And its easily controllable with a PWM signal. 0-5v The only issue would be an automatic controller but with simple pwm that would be much easier.

I'm also considering a compound turbo setup, taking another page from the diesel guys! Check out this 2jz+ compound turbos in a z31 :http://www.youtube.com/watch?v=JbFr_BW_278

~Alex

The problem with putting on your own servo is the packaging, mechanics, and heat issue. Where would you mount your huge servo? How would you mount your servo? If in the engine bay how would you avoid heat? How would you gear it down to provide the holding torque needed? What about an encoder? Would your whole motor/encoder system be waterproof? etc... But yea it is possible, just not very likely.

CAN using stock hardware > modified stock hardware > complete electromechanical redesign.

Did a little "brute force" CAN work on the actuator last night with still no luck. Pretty sure I will need a truck here to get some data off of.

[RustspecS13]

The servo would be pretty easy to package as its not that big,relative to the turbo anyway. Especially the small control board would be plenty easy to isolate from heat and water. DC motors can run perfectly OK submerged in water, its AC motors that short out in water so it would just need a shield.

As far as gearing it down that would be pretty easy since the servo would have a built in gear box or if I had to use a bigger motor there's plenty of cheap R/C parts to make gear boxes out of.

Like I said Ive been thinking about this for a while and I just wanted to get the best possible VGT actuation as fast as possible, and I have no idea how to do any software writing for the can.

I do agree the more stock stuff used the better as it is designed to work with the under hood conditions.

~Alex

[Mistaken]

Unless you could remote mount the Servo, I dont believe any R/C servo would stand up to the heat of a Turbo. The main advantage of retaining the factory equipment is that the electronics are designed to take some pretty serious heat and are also water cooled on top of that. Even then there is still some skepticism that it will hold up to the heat of a gas engine, although no one has had a failure yet.

Not trying to bash your idea because its good to see someone trying to find an innovative solution to a fairly difficult problem, i just thing you might need to get either really creative with your mounting solution, or find yourself something that can stand up to the heat.

[WhatsADSM]

Still looking for a truck. Turns out my one buddy has an 07 but it is the early 07 with the 5.9.

I made a post up in the Cummins forums in the wisconsin section. Hopefully there is someone that is willing to let me collect some data from their truck. If I don't get any takers, I'll give it a shot in the IL section of the forums... Hopefully I find someone!

[Mistaken]

I am waiting to hear back from a few leads on trucks.

On a side note my spare actuator arrived today, and I am picking up my turbo from 240z4u later tonight (thanks again). I am going to partially dissassemble the spare actuator tonight or tomorrow to see if I can find the I/O pin for PWM, and also see if I can get part numbers off the microcontroller and motor controller for the CAN side of the project.

[240z4u]

Glad to help, see you tonight man! Thank god I have blizzaks on my S14.. else I wouldn't be looking forward to the drive.

Evan