Since I requested this section :P

[hakcenter]

They work together, one takes your SPI outputs to the other, last one sends CAN on the way out.

I never looked up much, but you can check the seeed schematic.
Wiki : CAN-BUS Shield - Wiki
Eagle File : http://www.seeedstudio.com/wiki/images/7/75/CAN-BUS_Shield_v0.9b_20120621.zip

[Rx7man]

ah, OK.. and is this a bidirectional thing? can you read the position of the VGT through it?

this is a steep learning curve for me, I'm doing lots and lots of research on it though.

How do you etch boards?.. more importantly, how do you print the mask and drill the holes?

[hakcenter]

I've never actually etched my own boards, just engineer something in eagle and proto it.

I probably should make this a wiki so that people can understand the chain of command on the turbo. The Holset VGT's have 2 plugs, 1 4pin & 1 2pin (in a 3 pin housing).

The 4 pin is the main connector, it is power / ground / can low / can high
The 2 pin is the VR sensor connector, it is VR+ and VR-

When talking to the turbo, it doesn't do anything on its own, except when it hasn't received a canbus command for a vein position, it will default to vein position 0 if it hasn't received a message in 5-10ms.

You can only talk to the turbo through canbus, and there is some other communication information, Steed knows some info I don't know if he will join.

Only 2 things I know that you can tell the turbo to do is,
Vein position
Re-calibrate position (this one is wonky sometimes the turbo shuts off afterwards sometimes nothing)

From what Steed told me, what the turbo can tell you is
Feedback from backpressure (some type of measurement the motor force used to maintain position of sorts)
Some turbos have a temperature probe ??

So in a sense, just talking to the turbo its only 1 direction, but CANBus is basically like your old vampire coax networks, where everything is connected, has addresses, but its more modern in that, you designate what you want to talk to, and everything hears it but only the appropriate device will respond given that the command you send is good.


The VR sensor is a plug all on its own, completely independent of the controller housing. I imagine some manufactures use it, but most likely a lot have a harness connection that goes no where.

[Rx7man]

OK, good info!.. if I could get some sort of info about pressure back from the turbo it would save me a big chunk of change, even if the calibration isn't known or great (can always put a mechanical gauge on it and figure it out), and same thing for EGT if it has it.
Wow, it wants updates every 5-10 ms??? that's darned demanding!.. and the defaulting to position 0 is kinda stupid.. should just stay put!

[hakcenter]

Safety factor, to default open good sir.

The back pressure was the first thing I asked Steed about, and he said the information was really only helpful in finding if there was soot buildup not for actual backpressure.

I'm not sure where the temp probe, if 351's have it on the board. When I asked Steed he made it more about engine coolant temp, but when I had my board open I never saw a thermistor so maybe different boxes.. might have one No idea.

Yep, its a demanding controller. But you don't have to update position, you can continuously send the original over and over, and it just won't move. Which is where my code has honestly taken me. I don't see the need to micromanage the turbo, certainly going to blow the controller faster by making it constantly move.

I was working on top end code, to try to move away from my staging, but I can't find a reliable way to keep it under 130k rpm, unless its already in the 400 range.

[Rx7man]

Oh, 0 is fully open? alright.   I will update the CAN bus every in the Void loop, but only update the variable every once in a while.

Is there a good PID library around?  I made one in VB but it was way too complex to be used on a microcontroller

[hakcenter]

Haven't really looked or used one at all

[Rx7man]

I found one, may or may not have what I want/need, but I might hack it apart and make it do what I want it to... I think it would be the best way to control the VGT / boost level.

I was also thinking in the spirit of keeping it responsive, just set a minimum step size for the VGT but update as often as you loop through it.. with a correctly spaced step size it will naturally not move often... this is especially true in the closing direction, you want it to open responsively at all times though.

[Rx7man]

Have you ever hooked up an ammeter to the power wire of the VGT to see what it draws for current?  I was thinking about that fried board on CF and thinking perhaps it would be a good idea to put a fuse in there that's not rated too high.. perhaps a 5 or a 7.5 amp?.. better to blow the fuse than fry the board.. them things ain't cheap!

[hakcenter]

Around ~10, pushing from one extreme to the other, (0 > 1000 | 1000 > 0)

I have mine on a bosch fuel relay 20amp.

[Rx7man]

I have a pile of those.. certainly a good idea to run one.. Some models of those relays have 87 and 87a connected both as normally open, but not tied together.

[Rx7man]

I just looked into the LM2907, it doesn't need preconditioning to be run on a VR sensor (YAY!!).. and here I was, I had already had a MAX9927 already drawn up.. Oh well.. if nothing else I'm getting the hang of Eagle

I gotta say though.. what are the people smoking when they're drawing up these libraries??  They seem to try and deliberately make it impossible to have a nice schematic!..   Looking at the MAX9927, it has 2 VR sensor inputs.. the two positive inputs were above the two negative inputs, so it's impossible to not have criss crossed wires.. rather than have the first positive over the first negative, followed by the second pair which would just look nice.

I also see that Texas Instruments will offer free samples.. might be something to test out :>

As far as finding out the component values I may need a bit of a hand.. not all datasheets give you good info

[hakcenter]

9924 for a single input.

[Rx7man]

yes, but the 9924 is only a signal conditioner... the 2907 is a frequency to voltage convertor, which suits our application well, and it doesn't need nearly as many passive components around it than the 9924 does as far as I can tell.

I went with the 9927 because it may be handy to use with the Vehicle speed sensor at a later date, granted it'll probably need a different set of passive components to deal with the lower frequency

[hakcenter]

In A2, it needs 2x 10k resistors input and a cap between them. Power filters if you want. If you can afford the gate speed on the ADC multiplexers then ya go for it.