HE351VE Controller Project

[knochgoon24]

You should be able to eventually get a good SD tune dialed in, even with VGT, if the controller is good and controls the vanes in a repeatable way. I wouldn't want to be rough tuning the vane control while on an SD setup.

 

[wicked93GS]

Well...I picked up one of these turbos...I was thinking...if you have a SD car and MAP isnt the best option...what about a backpressure sensor...install a MAP sensor on the exhaust side and use a reading from that to control the vanes...this way.you could fool the ECU into thinking the engine has a fairly consistent VE

 

[TexasTurbo]

Are there any updates on this?

 

[knochgoon24]

Are there any updates on this?

I don't know about his unit, but I ran into a problem with the motor I chose to use. It was too weak. So I'm going to try and source another whole electronic assembly (with the OEM electric motor) and go from there.

 

[kthackst]

I am getting my car ready for the turbo to be installed in the next few months. After that will be A LOT of tuning of the fin movement. We had a goal of making it to the shootout but unfortunately I had a few other things come up that required my time and money. So we will see but my GSX is officially retired and no longer daily driven. Now for the hard work.

 

[TexasTurbo]

Someone suggested to me that you could remote mount a servo attached by cable to the stock actuator with a little fabrication.

 

[GST with PSI]

Question: I'm building a mechanical system to actuate the vanes on a Holset HE351vgt I picked up. From my understanding, this turbo will largely decrease spool time while still performing like a larger turbo, given vane position is actuated correctly. I want to actuate vane position based largely on boost pressure, as well as RPM and TPS (more specifically WOT) signals all combined. The problem is, idk at what boost pressure and RPM would be a good start point? Is this pretty much going to be a guess and check thing, or can one of you guys take a stab at what I'm getting at. Or, is this a shitty idea all together? And, if this is a shitty idea all together, what are your suggestions on what signals to use to best actuate the vanes with a mechanical system? This system could be made using a vacuum wastegate actuator, a few Hobbs pressure switches and TPS and RPM trigger switches. I'm not at all trying to discredit whats going on here. This electronic controller when finished will be far better than any other system available for our cars, be it mechanical or electronic. I'm just looking for a more simple mechanical solution for the time being. The info here is the best I have found, and the members here seem to have the most knowledge on this subject. Can one of you guys help me out?


I have been doing a lot of research on how the control system works. The problem is, Most of the info I am finding is for VGT operation on diesel engines. The way they were designed to work on the Diesel engines they are equip on, and they way they work for us have some differences from what I'm reading. For example, incorrect vane position on a diesel engine could mess up fuel economy and emissions control, where as on our smaller displacement gas engines this isn't really an issue. The diesel engines also use the vane position as a way to limit boost, where we would still have to run external wastegates. These are just a few examples of what the differences are between engines. I know the basics, and also know that not only does optimal vane position vary by each particular setup, but there are many factors that determine the optimal vane position for a given time. I'm not looking for a lesson in how VGT turbos work, but more of a simple breakdown on what largest factors that contribute to optimal vane position on OUR small 4cyl engine are, and how they can be figured out. From my understanding, the largest benefit to running a VGT turbo on our engines is a quick spool up time. If that's the case, couldn't you just make a simple mechanical system that uses boost as a signal to determine when the vanes start to open? You could also incorporate a RPM and TPS signal into this to further fine tune when the vanes open, but it would still be pretty simple. I appreciate any help you guys can give me. Thanks in advance.

 

[dan catalyst]

i hear some guys run a wg actuator for the veins. this car has that done holset he351 VGT dsm eclipse turbo start up - YouTube

 

[dsatchell]

Roughly, what kind of price range are you looking at for this; $200, $400, $600 ???

I'm interested but if the price is too high then I will need to consider other options.

 

[casuprock]

This project got technically deep very quickly. I'll go through a list of updates. We are thoroughly testing the controller's inputs, hardware, and writing software.

1. Custom thermocouple. I had to design a custom thermocouple with a sufficiently low time constant to measure the desired changes in outlet temperature. The outlet air temp changes very rapidly and we are also sampling at high rates, requiring a special sensor. This thermocouple is a welded exposed tip Type K, was manufactured with stainless steel strain relief, stainless cable braid, and protective sheath. The sheath that enters the air flow is 1" long / 0.040" diameter with the 0.005" thermocouple ball located at its tip. It is pictured below mounted in a Forced Performance 68HTA. Yes, this TC will be available in the kit we are preparing.

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

2. Data streams. We are streaming data directly from the controller through USB. These aren't your typical streaming rates... Thus they capture neat things that one would otherwise never see from a normal log. I am providing a link to a raw data stream (you can see we have the VTG stuff in there for stepper control). These data streams are focused on temperature testing. The turbo being logged is a 68HTA on my WRX.

http://www.suprocktech.com/files/DSM/turbo_controller_data/20120826/VTG445-20120826-1501.xlsx
http://www.suprocktech.com/files/DSM/turbo_controller_data/20120826/VTG445-20120826-1503.xlsx

3. Cool results from data streams. I will leave you with a teaser... because of how fast our controller is sampling the compressor temperature, we think we are seeing the turbo stuffing air at a closed throttle plate before the BOV opens up. The temperature value spikes as the compressor approaches surge (mass flow drops rapidly) and compressor efficiency momentarily drops. Compressibility and the vacuum line pressure drop on the way to the BOV may be partly to blame. The particular BOV is a TurboXS RFL (yeah... ). We will do more logs with the pressure sensor/maf signals streaming to confirm.

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

One last thing I've been informed of already (i.e. so hold it )... What's the difference between a Subaru and a cactus? A cactus has pricks on the outside.

 

[casuprock]

Pressure and temperature time aligned:

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

A pull showing pressure behavior... note how the pressure gets "fuzzy" as the turbo gets up to speed and boost starts to build. :

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

The "fuzz" is actually cool information. Here is a detail of pressure at boost, showing that we can resolve pressure spikes due to exhaust pulsation. This is very spiffy because everybody knows this is happening but now we can actually resolve it in a live data stream. This might be valuable for tuning.

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

As usual, here are the logs these plots were generated from:
http://www.suprocktech.com/files/DSM/turbo_controller_data/20120828/VTG445-20120828-1909.csv
www.suprocktech.com/files/DSM/turbo_controller_data/20120828/VTG445-20120828-1911.csv

 

[knochgoon24]

The rise in temperature before the BOV opens is also because of the pressure increase, not just because of the drop in in efficiency. You could probably do some quick math to see how much each contributes, just for kicks.

Cool results. It making me reconsider method for measuring intake temps. I might use thermocouples now.

 

[ed1380]

adiabatic heating pretty nice resolution

 

[casuprock]

adiabatic heating pretty nice resolution

"Adiabatic... now there's a word I've not heard in a long time."

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

Hey, guess what... We adjusted the PGA gains before the ADC on the hardware and voila, 10x better noise performance! It helps when you take advantage of the hardware built into your design. I'll get some new data plots up soon. Next set of plots I will see if I can get the MAF signal as well. In theory, we'll be able to plot compressor efficiency with that infomation.

 

[TexasTurbo]

I cannot even begin to tell you how awesome it is that you are doing this. That data is really cool. The sensors must have a really high resolution.

 

[casuprock]

I cannot even begin to tell you how awesome it is that you are doing this. That data is really cool. The sensors must have a really high resolution.

Thank you for the encouragement!

 

[casuprock]

I have a few questions for everyone following this thread:
1. Who wants to see a second pressure sensor for measuring exhaust backpressure with similar resolution and speed to the charge pressure we are already testing?
2. Who wants to see a second thermocouple input for measuring exhaust temperature with similar resolution and speed to the charge temp sensor we are already testing?
3. Who wants to see a CAN bus interface for talking with external motor controllers and *possibly* talking to stock Holset VTG turbos without physically modifying the oem Holset turbo motor?

 

[knochgoon24]

I'm sure you've seen my thread, but I personally am looking at 6 sensor locations for temperature and pressure.

-pre compressor
-post compressor / pre intercooler
-post intercooler / pre throttle body
-post throttle body in the intake manifold
-pre turbine
-post turbine

Now that you've shown how nice the thermocouples work, I'm ditching my thermistors for those. And for the pressure, I'm now looking at the same sensor as my build thread, but a different package.

What chip are you using for the thermocouples? I've looked at the Maxim 31588 and TI's TMP513.


And just something I was thinking of the other day. A lot of guys running SD have problems with the IAT sensor heat soaking at startup. Since the airflow is so low at idle, a lot of people have idle issues with hot starts. I wonder if there is market for a plug-and-play thermocouple based sensor that would work in the place of the GM IAT sensor.

 

[TexasTurbo]

Couldn't you use a microcontroller with the thermocouples to feed the stock sensor input a signal that is proper?

Casuprock:

1. yes
2. yes
3. yes

 

[knochgoon24]

Couldn't you use a microcontroller with the thermocouples to feed the stock sensor input a signal that is proper?

That's what I'm going to do, but that's only because I have already have the uC and I have plenty of I/O ports still open.

 

[casuprock]

Be careful before jumping into TCs. There's a reason it took months to develop a good sensor and interface (and it's what I do for a living). Thermocouples are more involved than thermistors to get good noise performance because the voltages are much lower. They have to be designed properly to work well. When correctly implemented, they pay dividends for response time. That is the holy grail in this case, and it came at the cost of my time and money. The next major leap will be making one that tolerates the violence and heat of the exhaust side while retaining a fast time constant.

We built our own differential 24bit interface for the thermocouples because the standard options like you mention did not have good noise performance at satisfactory sampling rates. The bottom few bits will be noise on those converters at an acceptable sampling rate. Even with the assumption of zero noise, the best case resolution on a signed 12bit is 5V/2^11 =2.4mV which still may not be acceptable for a type k TC. Consider you also need the same resolution on your junction compensation.

I'm not being terse, just trying to keep you out of the rough. Your thermistors aren't a technically bad idea, send me a PM if you want to have a more detailed discussion about those. Also let me know if you need TC sensors.

Could you combine your pre-throttle body and post throttle body temp measurements? Air is moving at up to hundreds of meters per second depending on the RPM... hopefully manifold air temp and post-IC air temp are very similar.

Finally, I think you caught me red handed with the IAT sensor. That's a direct application we are working on using hardware from the turbo controller to output an analog signal instead of only to USB.

I'm sure you've seen my thread, but I personally am looking at 6 sensor locations for temperature and pressure.

-pre compressor
-post compressor / pre intercooler
-post intercooler / pre throttle body
-post throttle body in the intake manifold
-pre turbine
-post turbine

Now that you've shown how nice the thermocouples work, I'm ditching my thermistors for those. And for the pressure, I'm now looking at the same sensor as my build thread, but a different package.

What chip are you using for the thermocouples? I've looked at the Maxim 31588 and TI's TMP513.


And just something I was thinking of the other day. A lot of guys running SD have problems with the IAT sensor heat soaking at startup. Since the airflow is so low at idle, a lot of people have idle issues with hot starts. I wonder if there is market for a plug-and-play thermocouple based sensor that would work in the place of the GM IAT sensor.

 

[knochgoon24]

The Maxim 31855 (not 31588) claims a 2°C accuracy and 0.25°C resolution since it's 14-bit. I'm curious to see the output. I burnt up the first two chips I had when I inadvertently gave them 5V instead of 3.3. I should be getting some new ones soon.

I received some samples of the TMP513 from TI on Saturday. It is only 12-bit, and wasn't designed for use with thermocouples, but it does have the ability to do cold-junction compensation. I'll hopefully have something up and running with those soon, just to see what I get with my ungrounded thermocouples.

 

[casuprock]

The Maxim 31855 (not 31588) claims a 2°C accuracy and 0.25°C resolution since it's 14-bit. I'm curious to see the output. I burnt up the first two chips I had when I inadvertently gave them 5V instead of 3.3. I should be getting some new ones soon.

I received some samples of the TMP513 from TI on Saturday. It is only 12-bit, and wasn't designed for use with thermocouples, but it does have the ability to do cold-junction compensation. I'll hopefully have something up and running with those soon, just to see what I get with my ungrounded thermocouples.

One thing to do would be to use a ceramic capacitor between the leads of the thermocouple to reduce noise with the Maxim part. Keep in mind this will increase your time constant (slower response) but could help if you encounter noise issues with higher sampling rates. Noise increases with faster sampling rate and you won't be able to do a whole lot about the internal noise of the ADC that is related to settling time of the converter.

We run most of our designs from 3.3 or 2.5V... 1.8 is starting to appear more often in parts now. The days of 5V digital systems is over, since transistors are happy at lower gate voltages now.

 

[knochgoon24]

I'm not sure who you went through for thermocouples, but my search for exposed thermocouples that may live in hot exhaust lead me to two interesting probes from Nanmac. It seems that they make a "right angle ribbon probe" (no price) with claimed 3200°F and 10,000psi limits.

Right-Angle Ribbon Thermocouple for Air, Gas, Liquid, Blind-Hole and Immersion Temperature Measurement

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

The other is an eroding tip probe (again, no price) that slowly wears to provide tiny junction at the very tip for quick response time.

Pencil Probe Eroding Thermocouple


If I get a lot of noise, a small capacitor may help. I'll have to see just how bad the noise is.
And the only reason I'm still running 5V stuff is because the Arduino likes 5V. I need a level shifter just to communicate with the Maxim chip. That is another part I'm waiting on.

 

[casuprock]

Here is a very interesting plot of pressure during two pulls. Since we are sampling so fast and have uber resolution, we can see frequency components in the pressure as outlet pressure fluctuates with engine breathing. This kind of plot is called a spectrogram and shows the frequency content of the signal over time. During each pull, you can see the frequency components of pressure respond as the RPM of the engine increases. Note the mean (0 Hz) component is offset because this is an absolute pressure sensor.

Everybody knows that this is happening, but now we can start to quantify the information.

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

Source data:
http://www.suprocktech.com/files/dsm/turbo_controller_data/20120907/20120907_Pressure_Tests.zip