HE351VE Controller Project

[1990AWD]

I personally would be interested and even if the controller is more expensive than a used Holset turbo, it still would be a fraction of a big new Garrett plus you get the benefit of the VGT.

 

[nuclearr]

If you don't already have Link, would you buy Link because it supported an electronically controlled VTG turbo (i.e. could tune VTG maps, bidirectionally communicate with ECU, etc.)?[/I]

I would.

 

[casuprock]

seeing how good used vgt holset can be had for around $250, I doubt many people would like to pay more for a controller than the actual turbo.

I really hope this will be available by next summer

as far as buying link, personally that would be a no. for some the 98 99 flashable ecu is a cheaper option

Should be ready by next summer, at least in testing.

Maybe the price of the HE351VE will go up after there's a controller people can use for the darn things. At least it would give them more value. This turbo represents so much untapped awesomeness. Right now the market for them is zilch plus the repair/rebuild market for compatible truck engines.

It's amazing what can be done with those 98/99 ECUs! I've done a little bit with ceddymod on the flashable ECUs with OBDII monitors/inspect.

 

[ed1380]

Should be ready by next summer, at least in testing.

Maybe the price of the HE351VE will go up after there's a controller people can use for the darn things. At least it would give them more value. This turbo represents so much untapped awesomeness. Right now the market for them is zilch plus the repair/rebuild market for compatible truck engines.

It's amazing what can be done with those 98/99 ECUs! I've done a little bit with ceddymod on the flashable ECUs with OBDII monitors/inspect.

+1 for an underrated turbo.

 

[knochgoon24]

I was just thinking about this the other night. I don't have all the electrical engineering skills you do, so I was going to try using an Arduino Mega 2560 and go from there. I figured out that it would take about $800 to get the turbo on my car and running, and I just don't have that kind of spare cash right now.

I'm glad to see you're making progress.

 

[turbotalon1g]

I'd be interested depending on price of course. Nice work

 

[bastarddsm]

You do know that 1g guys can use the factory ecu to control the original stepper motor?

 

[casuprock]

You do know that 1g guys can use the factory ecu to control the original stepper motor?

Well I don't have a 1G for starters... Don't you think the stock ecu's transistors might be too wimpy for the current required to drive this motor? When I was driving the Holset motor wired as a bldc on the bench from 14.5V, the current was around 1.2A rms.

I would like to have a controller that makes use of the VR sensor like Holset intended. Just for fun, I'd like to have more control over what's going on. There's a lot of room to optimize given the opportunity and effort.

Hey aren't you the guy that told me to throw my engineering books away?

 

[knochgoon24]

I'm surprised it's drawing 1.2A. And that's not even fighting exhaust pressure.

So will you use a RPM vs TPS map to pick a target boost? Then from there, have the vanes adjust to provide as much boost as it can without choking the flow?

I don't really understand why you need position feedback. The vanes either need to stay put, open, or close from the current position. Unless you want to set max and min positions. But even then, why limit yourself?

Half these questions are just to get discussion going. There were a few people working on controllers over on homemadeturbo.com. No one has one working yet. Holset VGT HE351VE Controller - HomemadeTurbo - DIY Turbo Forum

This will be the next big Holset break though. Let's make sure the DSM guys stay on top.

 

[casuprock]

I'm surprised it's drawing 1.2A. And that's not even fighting exhaust pressure.

So will you use a RPM vs TPS map to pick a target boost? Then from there, have the vanes adjust to provide as much boost as it can without choking the flow?

I don't really understand why you need position feedback. The vanes either need to stay put, open, or close from the current position. Unless you want to set max and min positions. But even then, why limit yourself?

Half these questions are just to get discussion going. There were a few people working on controllers over on homemadeturbo.com. No one has one working yet. Holset VGT HE351VE Controller - HomemadeTurbo - DIY Turbo Forum

This will be the next big Holset break though. Let's make sure the DSM guys stay on top.

Good points. Here's the 'evil plan', feel free to suggest more plan or less evil at your leisure.

I won't be controlling target boost directly, that will be done with a traditional external waste gate. I got a precision PW66 for that. The goal of this project is to reach target boost as fast as possible while keeping the compressor efficiency up.

Step 1:
Know input and output temps from the compressor along with pressure ratio to calculate the compressor efficiency. I'll log the pressure ratio, calculated efficiency, and shaft speed from the turbo controller and CFM from my MAF. Voila... compressor map just from driving around logging!

Step 2:
Once the compressor maps are understood, the shaft speed associated with a specific PR and compressor efficiency can be made into a look-up table on the turbo controller. (Note shaft speed has significant slope across the compressor map, especially at increasing Pr.)

Step 3:
Implement the look-up table from Step 2 in a PID control. Here's where vane position feedback comes into play- required for proportional feedback so I know where I'm at and where I can quickly move to in the blade stroke. That gives a faster acting correction versus an incremental correction. Shaft speed is sensitive to changes in exhaust flow, so the controller will 'naturally' prevent excessive back pressure (because it too much back pressure leads to excessive shaft speed).

As long as the turbo time constant is faster or equal to the engine's, the shaft speed (hence controller) should respond alongside back pressure and not choke the engine.

The real answers will come once the system is completed and measurements can be made. Let it be judged by it's performance!

 

[knochgoon24]

Ah. Makes sense now. A bit complicated, but should yeild better results. Everything I've read also suggests that a wastegate is needed no matter what, because the car will still build boost with the vanes all the way open.

 

[casuprock]

Ah. Makes sense now. A bit complicated, but should yeild better results. Everything I've read also suggests that a wastegate is needed no matter what, because the car will still build boost with the vanes all the way open.

Yeah, I think that's what would happen. I'm not entirely sure it does need a wastegate, but better safe than sorry.

I figured I'd try do the design right, making the prototype electronics capable of handling the worst case scenario (and complicated control). Then I can figure out what's really necessary to do a good job without having to rebuild the prototype. It's easier to back off than go the other way.

I think I'm about 2 weeks from ordering the first PCB revision, so stay tuned! I'll be sure to post pictures once I have it assembled.

 

[casuprock]

Update: The schematic is done. Here's a picture of the recent board layout.

Some of the key features/additions since the original specs:
- 32 bit Atmel UC3B processor which will easily deal with all the control, communication, and some math
- High speed USB connection for data logging, uploading/downloading maps
- Two type K thermocouple inputs (for immediately before & after the compressor inlet and outlet). Note these exist because temp has to be time aligned with pressure to calculate efficiency.
- High resolution analog pressure sensor input (for GM press sensor or otherwise)
- Variable reluctance sensor input for turbo shaft speed
- Bipolar stepper motor driver (for either 2 or 4 coil steppers)
- Ambient temp & Barometer on board
- Four LEDs for status indication, fifth led for power status
- Reverse input protection
- Buck converter regulator
- Tactile switches for manually incrementing or decrementing the stepper motor (for test, etc.)
- Screw terminal connectors for easy hookup

You must be logged in to view this image or video.

 

[Long Duk Dong]

Quick question - Are you using a scripted text or ladder logic to do your programming?

Kudo's on both the idea and your efforts. If it weren't for people such as yourself there would be allot of us stuck in the 70's.

Jim C.

 

[97gst_spyder]

Looks good.

 

[97gst_spyder]

Looks good.

 

[TexasTurbo]

Will it be able to drive the motor that's in the stock control system attached to the turbo along with being able to read the stock shaft speed sensor?

I'm very excited for this product and will be buying one from you when it's completed. Leave it to a DSM owner to come up with the best aftermarket control device for a VGT. Those diesel guys have knobs that manually adjust their turbo on their aftermarket solutions (from what I've seen). This is much more robust and would be closer to the factory control by far.

 

[casuprock]

Quick question - Are you using a scripted text or ladder logic to do your programming?

Kudo's on both the idea and your efforts. If it weren't for people such as yourself there would be allot of us stuck in the 70's.

Jim C.

The program is being written in AVR32 Studio in C. PC software will most likely be written in Python.

Thanks for the encouragement!

Will it be able to drive the motor that's in the stock control system attached to the turbo along with being able to read the stock shaft speed sensor?

I'm very excited for this product and will be buying one from you when it's completed. Leave it to a DSM owner to come up with the best aftermarket control device for a VGT. Those diesel guys have knobs that manually adjust their turbo on their aftermarket solutions (from what I've seen). This is much more robust and would be closer to the factory control by far.

I had considered using the stock motor because it can be wired up as a 3 phase BLDC. After some testing, I found it's not designed to change the blades fast enough to respond to rapid changes in engine flow rate e.g. engine RPM (it's designed for a slow revving diesel). The internal gearbox can be adapted to a more appropriate stepper motor with a wider range of RPM and torque capability. Position control and stall detection is also significantly more complicated on a BLDC running BEMF without hall effect sensors.

I had also considered using Holset's board and reverse engineering the program on their microprocessor (an ST Micro ST72F561J9TC). However, it would still need the external sensor inputs which Holset's board lacks since it was designed to talk to the Cummins ECU...

Yes I'm using the stock turbo shaft speed sensor with no modifications to the turbo core.

What I'll end up with is the ability to unbolt Holset's motor box and attach this aftermarket unit designed to bolt-up without requiring any physical modification of the turbo.

The incarnation of any future "kit" would be based around the following:
- Controller board
- Stepper motor
- Flange plate & coupler (to mount stepper to Holset's little gearbox)

 

[casuprock]

Some people have been wondering what application I am using to design this PCB. I am using Altium Designer 10. "How do you draft the electrical parts?" The parts libraries are all custom and have been developed by my company over thousands of hours working on different projects.

Each part in this design has been chosen by it's specific properties. Every IC, diode, transistor, capacitor, resistor, and inductor gets chosen to satisfy its purpose. On a design this size, it literally means reading a hundred data sheets and going through calculations for each one. Some of the time is reduced because I have used the parts before.

"Will this be a DIY thing with parts from Radio Shack." Sorry. A lot of these components are surface mount size down to 0402 (the size of pepper flakes). It takes special equipment to assemble this kind of circuit board. SMT components are far more cost effective for compact designs. It will make the end result much cheaper!

For those interested in electrical design, download PCB Eagle. It's free and there are a ton of resources out there. I started out working with Eagle. It isn't as advanced as Altium or the other professional packages, but it can do most things.

Hope this answers those questions.

This is what Altium Designer board routing looks like:

You must be logged in to view this image or video.

 

[dexterholland04]

Who are you going to use as your board PCB fabricators? From experience, it can get very expensive to run even a small lot of boards.

 

[kthackst]

I am going to be all over this!

Looking good nice to see some progress Chris.

 

[casuprock]

Who are you going to use as your board PCB fabricators? From experience, it can get very expensive to run even a small lot of boards.

I do this for a living... got the hookups. The PCBs are the cheap part, the parts are the costly bit. I'm using the best quality parts available since that's not something to compromise on.

I am going to be all over this!

Looking good nice to see some progress Chris.

Thanks man, it's a ton of work. You'll be the first 2.0 running one of these so it's up to us to show people how it's done!

 

[TexasTurbo]

Would it make any sense to have a lookup table of values hard coded in memory for stepper motor positioning vs. doing a bunch of realtime calculations?

 

[casuprock]

Would it make any sense to have a lookup table of values hard coded in memory for stepper motor positioning vs. doing a bunch of realtime calculations?

Yes that makes good sense for a specific "turbo tune".

Setup and generating those tables is where the equations come in to play. The final operating mode after setup and some tuning may have tables of data generated for the turbo. See this post for the three steps.

The process can be summed up; Learn, Set Conditions/Populate Tables, Monitor Feedback

The processor is fast enough to run the math in real time. Once I start programming, we might find that the best way to handle things is a combination of tables and compressor efficiency equations. The equations might be the primary control point, with tables serving as a "reality-check" to make sure a sensor isn't giving borked readings. Regardless, the calculation results will always be available for logging purposes.

 

[92awdgsx]

I am very curious about this setup. Kevin has filled me in on how everything is going to work. Sounds like an awesome setup. Can't wait.