New dual VR board V2.0

[Joe Perez]

Gratuitous shot of the new board:



The input goes to a 1/8" stereo headphone jack. From there, it's about a 10 foot run of shielded two-conductor audio cable (Belden 9451) to the sensor. The shield is carried through the headphone jack and grounded at the board, and left floating at the sensor side. The output has a 1k pullup, and goes directly to the MS's Tsel pad, which takes it to IRQ1 at the processor. +5 and GND come directly from the proto area on the MS, by way of another piece of shielded audio cable and a molex connector. (This board lives in a small box external to the MS)

As you can see, I decided to use a little adapter board from Proto-Advantage to sit the chip on. These turned out to be pretty inexpensive, and made the project a lot easier. Fortunately, we have a nice Leica scope and Metcal SMC soldering station here at work.

The two resistors on the left are the 10k series limiters on the input. The cap for them is an 0805 part on the backside. The resistor on the right is the output pullup. For the three power supply filter caps that they call for in the datasheet, I again used 0805 parts, and just stacked all three on top of one another and stuck them directly to the backside of the carrier board.

[racingmini_mtl]

Good to have some independent validation of the circuit. For those who don't know, the 9924 is the single VR conditioner version of the 9926 I'm using which has 2 conditioners.

And as you said, the differential input is a very good feature. What I also like is that it works equally well with a square wave input (0-5V or 0-12V) which makes it a universal conditioner for all sensor types without any tweaking of some pots (it will require a pull up on Hall sensors but it's the same as other circuits).

Jean

[Joe Perez]

As promised, the scope captures.

This first one is interesting. I am waving a screwdriver in front of the sensor at quite some distance, producing a waveform that's just barely distinguishable from the noiseon the line. (I had a fluorescent desk lamp switched on, and I believe that's where the noise was coming from.) Despite this, the circuit properly triggers on all three pulses:





Now, this next one is even more impressive to me, as this is a situation where the 1815 never quite worked right when I was experimenting with it. Note how all of the input pulses are of vastly different amplitude. The input event in the middle of the screen is about 1.3v p-k, and the one that follows it is maybe 250mv p-p. Despite this, the circuit catches every one of them properly:





And of course, the proverbial money shot, the circuit running on the actual engine at idle:





No question in my mind, this is THE decoder IC to be using. And I hadn't previously given any thought to it, but as you pointed out, its suitability for use with 0-5v square wave signals will greatly simplify the Rev 4.0 schematic, if and when such a thing ever happens.


As an aside, it looks like Digikey finally has some of these in stock. When I got mine just a month and a half ago, none of the usual distributors (Digikey, Mouser, Newark) even had them in the database, much less on the shelves. Had to sample 'em direct from the factory in Taiwan.

[Joe Perez]

What I also like is that it works equally well with a square wave input (0-5V or 0-12V) which makes it a universal conditioner for all sensor types without any tweaking of some pots (it will require a pull up on Hall sensors but it's the same as other circuits).

A question about this, since you're running in mode A2 also.

Did you have to bias the IN- pin of the input stage upwards in order to achieve detectable zero-crossings when using a 0-5v input, or is the automatic threshold detector working as advertised even with a single-ended input where IN+ is never < IN-?

[racingmini_mtl]

For 0-5V (or 0-12V) inputs, you simply connect it single-ended and the threshold is automatically set to 2.5V. And it works.

Jean

[Mad Professor]

Are these boards in stock, how much are they, and how can I order?

[racingmini_mtl]

Yes they are available. There is some pricing information here and you need to contact me to get one at this time (like you did) since I haven't done the web page yet.

You also need to tell me which version you want: the potted version with pigtails or the simple assembled board.

Jean

[vttinman]

For 0-5V (or 0-12V) inputs, you simply connect it single-ended and the threshold is automatically set to 2.5V. And it works.

Jean

How do you wire it up for single-ended mode? I'm guessing you wire the hall output to In+, but should In- be left floating, or wired to ground?

Scott

[racingmini_mtl]

For a Hall sensor, you leave IN- floating which will set the threshold at 2.5V. If you ground it, the signal will not cross the threshold so there wouldn't be any output. For a single ended VR sensor, you'd ground IN- since you want a zero-crossing trigger.

Jean

[vttinman]

Thanks for the info Jean. Would you recommend the MAX9926 over the traditional Optoisolater circuit for a cam in signal via a hall effect?

Scott

[24c]

Looks very nice Jean, is the website link up yet?
Nope, I just checked and it is showing the older 1.1 version. I'd like one please, can you PM/email me, you should have my contact stuff, and I'll send the coin by return.
Thanks Mike

PS Potted version with pigtails please.

[racingmini_mtl]

As you saw, the web site is not up yet...

I need to build a new batch of potted boards this weekend. I'll send you PM right after I post this.

Jean

[cocovi]

Looks very nice . Useful !

[dhallilama]

dying to get my hands on one of these

are they currently available (internal, not potted version)?

[racingmini_mtl]

Yes they are. I've sent you an email with the details.

Jean

[24c]

As promised, the scope captures....

Joe, just seen these, very impressive and useful info, as I am about to restart the cam sensor detection side of the GTS1000 project again, using Jeans potted pigtail version, so I can go sequential. I am confident this will work now with the original cased MicroSquirt, which was so problematic on VR2 detection at cranking speeds, plot sample below.
The pale blue peak to peak trace wasn't enough to trigger the VR2 circuitry on the cased MIcrosquirt.

[chense]

I've spent the better part of the weekend installing the potted version of the board.

Executive summary
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
I was successful , but there were (and still are) issues to work out .

This chip is so insanely adaptive and sensitive that it will bite you in the ass if you're not careful.

The long story
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Setup: 36-1 crank (originally connected to LM1815) + 1 tooth cam wheel (originally connected to V3 onboard conditioner).

Went on to condition both VRs via Jeans board (tore out the old setup). Spliced the board in the harness near cam VR, added a longer cable to the crank VR, power and ground from 5V TPS lines. The shielded cables Jean's using are amazing, I haven't seen a shield this thick in a long time, kudos!

First problem - cam tooth comes once per engine cycle. That means that on cranking and low revs the MAX9926 watchdog doesn't see a tooth for longer than 85ms, and resets the adaptive threshold level to V_MIN_THRESH, default min threshold (min 4mV - see MAX9924-MAX9927.pdf pages 3 and 13). That is below noise level for my particular VR, and I get false triggers due to noise, observe:

cam_out.png

After some experimenting, I've put two 27k in-line resistors on VR+ and VR- lines, and that cured false triggers, and amazingly enough the real teeth were still there .

After fixing the cam and starting the engine I've had a strange episode. Start the engine, let it warm up a bit, than I revved it a bit, but the engine died, fuel pump kept running, comms dropped out, VR board/MAX9926 appears dead on consecutive cranking (no signal from either channel, even when cranking with the spark plugs out for 10 -15 secs composite logger showed nothing ). MS tested fine on stim. Gave it some time, reconnected the VR board to fused +12V (was on 5V TPS), added 47k in-line resistors on crank VR inputs and it's alive again ?

for Jean: I believe in-line resistors did the trick, and I'm utterly confused about maximum input voltage, datasheet say max is VCC+0.3. But that probably refers to some inner clamping mechanism of the chip, since even with added in-line resistors the voltage is higher than that? My multi says I have 30VAC on crank VR output on idle, but there could be spikes higher than that, and I have no way to check. And I'm a little worried about being left stranded on account of this.

Second problem - I still get occasional sync losses (and eventual death). After some investigation it seems that MAX9926 does not like tooth number #3 on the crank wheel. Observe:

crank.png

Crank trigger is set to rising edge (yes, I'm aware, I wired the VR wrong), and it's newer sampled when low (except on cam tooth). This is intermittent, but when it happens it's always on the same tooth, with the same pattern/times. So there's probably something wrong with tooth #3, the bastard changes height in the middle. Will take it off and check it. LM1815 didn't have problems (at least not this visible) with it.

Polarity checking was on, but couldn't cope with this.

for James: with MS2 (3.0.3s, t, and RC1), is it normal that fuel pump stays running (engine dies) after sync loss? TS losses comms, so I can't really verify if MS sees any RPM, nor what was the lost sync reason.
And as the board outputs stay high, it's probably normal to see 4ms "ticks" in composite logger?

And Jean, I'm changing my mind.I want two more of these boards (not potted this time), I have an idea LOL. So count me in for the next batch (if there'll be one).

--
Cheers,
Tom

[racingmini_mtl]

Tom,

I'll study this a bit more when I'm back home and I'll get back to you. And I'll check to see if I have 2 assembled boards left from the previous batch. If not, the next batch should be assembled this weekend.

As for power, the board has a 5V power supply on it with an LDO voltage regulator. So powering it from 5V will work but not as intended. It needs a 12V supply and the regulator will withstand the battery fluctuations.

And I now put higher value input resistors to help with what you're seeing but that may still not be enough in some cases, especially for a single tooth cam signal (due to the 85ms time out and the high sensitivity, as you mention). I could probably still go higher and not have problems picking up very small signals.

Jean

[Peter Florance]

I've done some testing with V1.04 MS2Sequencer (IIRC uses same family of IC as Jean's board)
High output Bosch VR sensor.

Let me dig through my notes...

[racingmini_mtl]

I've done some testing with V1.04 MS2Sequencer (IIRC uses same family of IC as Jean's board)

Actually, it's the exact same chip: MAX9926.

Jean