T-leg Piggyback ECU R&D

[vrg3]

Sweet, a free tuner chip!

Actually, I was asking about discounts on OmniSoC though.

The buffered sampler thing makes a lot of sense... Basically a poor man's oscilloscope. Nice idea.

I'm not sure I understand what you mean about implementing a closed loop for the oxygen sensor... Do you mean this chip would take over the responsibility of performing fuel trims at light load?

Hey! That's a good idea! For multiple reasons! Yes. It's all making sense. When you install this chip, you should disconnect the oxygen sensor from the ECU. That way the ECU will always run in open loop so it'll run predictably. Ignition timing should also be more predictable. Then this chip can do its own closed-loop fueling under conditions specified by the tuner (basically a manifold pressure threshold and a TPS threshold, I'd think).

Ooooh. This gets better and better. You could have the switchover voltage user-specified too. Then, with a wideband sensor, you could run in closed loop all the time, not just at light loads -- you could have another 16x16 table containing desired air/fuel ratios.

Mmm. Definitely something to think about for the future.

Yes, the values in the MAF conversion table must be divided by 256 to get the airflow in grams per second. It may be useful and sensible, though, to first divide the number by engine speed, computing grams per revolution rather than grams per second.

[free5ty1e]

Indeed, thats a great idea - have this chip take over the O2 reading, an ECU that isn't adjusting is one that's easier to compensate for! heh... sweet!

I've got tons of room for static lookup tables, by the way. User-adjustable tables, well, we've already got two 16x16s. Absolute max number of 16x16 tables that would actually fit into SRAM would be 6, leaving one RAM page for the stack and one for other variables (and the smaller 8x8 table). So as far as adjustable tables, we've still got room for 4 more large 16x16... definately food for thought. The possibilities are friggin endless!!

First thing I'll do is display the MAF reading in grams per second on the LCD, along with the MAP reading in mmHg, O2 voltage, and coolant temp. in degrees F. That's a realistic goal for something to have operational and tested by this weekend.

[free5ty1e]

(OK, 10% discounts for all Legacy board members for the OmniSoC products now too.)

[evolutionmovement]

This wouldn't control timing, though, right? With the o2 disconnected, I think the ECU would go into a slight retard on the timing as when my o2 was bad. Once I replaced it and reset the ECU I got my power back. It does run advanced timing when cold, though ... Maybe fooling it into running 'cold' mode and just relying on the knock sensor for safety? I feel like the dumb kid in class, so sorry if these questions have been answered, but quite honestly, I just skimmed most of this as it's over my head.

Steve

[free5ty1e]

No worries Steve - yes this has a 16x16 table to allow the user to adjust timing trims. It will intercept the spark signals to the ignitor and can delay them to correct for unnecessary advance from scaling the MAF signal. It should be possible to advance timing as well with this system, it will be worked out at some point.

Hmm thats something we haven't talked about yet - should I monitor the knock sensor? What sort of signal does that produce, another voltage...? Isn't it just a microphone? So many sensors, so little time lol

[heelhooker]

"What has a response time on the order of seconds, now? Nothing I'm doing here will be too slow, I will see to that... "

I'm not saying anything you're doing will be too slow. I'm saying the termperature sensor itself is too slow.



"What do RTD and TC stand for? Are you talking about temperature sensors? They do make thermistors with much faster response times. Also keep in mind that air is flowing very fast over the probe, so circumstances are much more favorable than the typical "in still air" measurement. But, yes, you will end up running rich for a short period when boost starts. It's a phenomenon exhibited by most speed density supercharged engines, including OEM ones. It typically isn't so bad that the engine bogs down though. "

Yes, RTD=resistance temperature device, TC=thermocouple. Do you know what the time constant for these high speed thermistors is?

Yeah, I know that the convection coefficient is strongly dependant on flow velocity, as well as the geometric dimensions of the probe, among other things (there's actually a bunch of non-dimentional parameters used to relate these properties)

Anyway, I'd be interested to see what kind of temp sensor most speed density systems use. Reading about the megasquirt, it seems there is no compensation for this.

[vrg3]

Chris - Is it possible to store the tables in some kind of nonvolatile memory? I just realized that if they're stored in SRAM then they'll disappear when power disappears.

Sweet. I wanna learn to use the PSoC. Does it have to be programmed in C, or are there assemblers? What Intel instruction set does it use?

Yes, the knock sensor is basically a piezoelectric microphone. You could act on a voltage threshold, but my understanding is that the only really useful knock sensing stuff is based on DSP and a priori knowledge of the motor's acoustics. If you do want to interface with it, you need to use a high-impedance amplifier to connect to it.

It might be cool to integrate such a high-impedance amplifier and connect it to a headphones jack next to the LCD. Then you can let the wetware deal with the processing.

And maybe it would make sense to also run the ECU without the knock sensor, so it stops playing with timing. One of the big difficulties with piggybacks on Subarus is the large amount of leeway the ECU has with spark.

Steve - Yeah, like Chris says, this device should be able to control timing, but only in one direction. It'll be able to retard timing fairly precisely.

heelhooker - I don't know any of the time constants for the automotive sensors like the ones used by MegaSquirt. The basic idea there is that if they're good enough for GM to use as OEM, they're good enough for us. I can tell you from experience that the enleanment delay as the temperature rises on boost isn't really noticeable from the driver's seat.

Let's also note that this type of piggyback doesn't necessarily need an IAT sensor.

[free5ty1e]

True, IAT sensor won't actually be used for anything but unless I need that input for something else that's what that input will be for. Either way, if something new and cool to monitor comes up to replace that input, an upgrade chip will take care of ya. Once completed, the PCB layout will be pretty much all-purpose for this application. (Yes, this product will have a printed circuitboard with an informative silkscreen... and the in-circuit programming interface for the PSoC so it becomes a programming socket if you ever get a burner and want to experiment with the code)

Yes, one of the menu selections will be for saving all the settings to flash memory. Upon powerup the chip will read the saved settings back into SRAM and it's all good again. This way if you make some tuning adjustments that land you worse than you started, you just don't save 'em.

You can program with intel-based assembly (not sure exactly which version but what I mean is the commands are "mov" and "jmp" instead of the Motorola "sto" and "bra" mnemonics) with the free PSoC Designer IDE software, available from http://www.cypressmicro.com (I believe they just changed their page but you can find it from there). To program in C you have to purchase an Imagecraft C compiler license specifically designed for the PSoC instruction set. Interrupt service routines can even be done in C with a little template editing...

Might not be a bad idea to just amplify for the knock sensor and provide a mini headphone jack for future use, or hey if you ever want to hear your knock sensor.....

[vrg3]

Oh, okay, that's sensible... Semipermanent stuff in flash and new stuff in SRAM.

Heh, cool, I cut my teeth on MOV and JMP.

Yeah, people should want to listen to their knock sensor. Many professional tuners do this while tuning; they say that detonation can be inaudible from the outside of a closed-deck Subaru motor even when it's obvious when listening through an earphone.

[free5ty1e]

... well then. What if I were to just straight transmit the knock sensor's signal as an audio source via FM to your radio, just tune in when ya want to listen? Would that be worth somethin to ya then?

(headphones ARE illegal when driving, although not for the passenger. Then again, catalytic converters are kinda required by law too...)

[vrg3]

You wouldn't be listening all the time while driving; you'd be listening while tuning, which would ideally be happening on a dynamometer. It's not like you can stare at the other instruments while driving either.

I built a little doohickey to transmit the knock sensor signal via FM, but it was close to useless. It may have just been my crappy circuits or my stereo setup, but it seems the bandwidth of FM radio is insufficient. Or maybe my ears are just useless.

[stipro]

This sounds awesome if it all pans out, a piggyback for our Legacy turbos, Finally.

Iam interested. Let us know when its ready

[evolutionmovement]

I would buy one for sure. This saves me a lot of money on stand alone and would bring my project very close to reality (that and the fact that Prodrive seems like they'll never release their active suspension which leaves me with much cheaper alternatives).

Steve

[jake15]

I will definitley want one. what is the ETA on the finished product?

[vrg3]

Don't get too antsy... Give Chris time to make it happen.

Oh, sorry about the thread hijack, Jake... But now that it's been done, do you think maybe you could change the title of the thread to something that reflects this piggyback development discussion?

[jake15]

Don't get too antsy... Give Chris time to make it happen.

Oh, sorry about the thread hijack, Jake... But now that it's been done, do you think maybe you could change the title of the thread to something that reflects this piggyback development discussion?

i'm in no hurry, i just need to know when to start saving my pennys. lol

and the thread title is changed to something more appropriate. this is the best thread hi-jack ever i dont mind at all.

[vrg3]

Hehe, whoops, I meant to put a smiley after that first sentence. I wasn't trying to scold you or anything.

[jake15]

i understand, expressing any emotion other than anger is hard to do on the internet, thank god for smileys

[free5ty1e]

heh... no worries. I'd estimate in the next two months I'd have an operational unit I could hand build copies of for sale. Shouldn't be a week or two after that when I have the circuitboard made. You'd be surprised with how much is in the works already. I plan to test the first burn of the chip this weekend, it should read and display all the analog signals' measurements from the lookup tables on the LCD and allow the user to select which to monitor. I'm also going to test the signal sampler this weekend, which will evolve into a multichannel datalogger to record runs for detailed tuning outside the car.

The topic seems somehow... more appropriate to the actual discussion content now. Sorry about the threadjacking lol we did a number on this one!

I'm starting an official list of people who'd like to be notified when this product is ready for sale... just let me know here or PM me or email me. But either way this thread should always have the latest status.

[stipro]

I pm'd you. Just imagine the possibilitis.........
Im so giddy right now. I want one.

[THAWA]

How would this interface with the car? Would you have to do some splicing or could you have it setup with a plug and play system? Like how Vikash did with his FCD. I'm not sure if you could do that resonably for everyone.

[Kelly]

I personally have no problem with a bit O cut n solder, and chalk me in as well for a buy. Im really excited about this.

[legacy92ej22t]

Mmm, this sounds tight! I'm definitely ineterested too.

I've been trying to keep up with every thing but damn, you guys are going off. The brains on you two must be smokin'! This is exciting.

[BAC5.2]

I'm down, let me know when it's ready.

I need something to control 440cc injectors without having to dyno tune every time the weather changes. I want 16psi to be more reliable damnit!

[free5ty1e]

BAC - Yes, unfortunately this won't be as plug-and-play as Vikash's fuel cut defender circuit. The majority of the signals for this thing can be found right in your cab at the ECU pinouts, but you'll have to run a few wires through the firewall.

I actually envision offering base trim tables for certain configurations that I've already tuned for. Once I've got a setup ready I can upload it to my laptop and have it available on my website or something. So can other users. But either way it will take some time to make runs and adjust the trims.

Once I'm further in and can maybe experiment with a wideband O2 sensor, I think there will be an auto-tune mode that lets you run in closed-loop all the time....