Still losing tach; looking for new theories (Solved!)

[sportage4x4]

C1 is going to cause issues with dropping tach events if you choose it as per their suggestions. i normally use a 0.01uF cap. that works fine up to the 11,000rpm rev limit of the bikes, with a 36-1 wheel. on the bench, with a 60-2 wheel, its fine up to 8000rpm (as fast as my test rig will spin it)

[sportage4x4]

just to add to this, i generally bypass the opto with this setup as well. not much point in using the opto, as you can select rising or falling edge as the trigger.

[SQLGUY]

I've ordered NCV's and other parts from Digi-Key. May have them tomorrow. I'll update this post with the results once I've got the alternate VR conditioner installed (I plan to tack it directly onto the uS board).

Thanks for the help,
Paul

[SQLGUY]

Got my parts today. I installed the NCV with a 15K input resistor, a .01uF filter cap, and 16K for the tuning value (my pickups measure 600 Ohms). I installed it directly on the uS main board with the NCV output going to CPU pin 1.

It sort of almost works. Basically, I get tach during cranking, but as soon as it starts to fire up, trigger +/- decrements (log attached). During cranking, I'm only getting about 10V P-P from the pickup. As soon as the engine "catches", this jumps to about 30-40V P-P, and the CPU loses tach.

I'm going to try tacking in a pot in place of the input resistor to see if I can get some better behavior out of this. Any other ideas?

Update: the new data log was done with 33K resistors added in series with the input and adjustment resistor. This allowed the engine to sort of run with the throttle left at idle... it would bounce up and down with the resynch acting as a sort of off-idle rev limiter.

Update 2: by moving the pickup further back from the trigger wheel, I can get things to the point that I can rev to 4500 RPM or so, but, this also makes it much harder to get initial synch. This makes me think that I do in fact need more than .01uF for the input filter. On the scope, what I'm seeing at idle is about 10V P-P for the main teeth, spiking to about 18V P-P at the missing tooth. Loss of tach occurs at about 40V P-P for the missing tooth. This is measured at the input to uS.

Update 3: Getting a lot closer. I am not following the rules exactly, but I think what they're trying to explain in the datasheet is that 250mV is the minimum trigger voltage of the NCV at its input. They're using the input resistor and the cap as a 6dB / octave attenuator to keep the input voltage between 290mv (a little higher than 250) and the 5V clamp limit. Try as I might with different cap values and load resistors I was not coming up with anything that seemed to match the amplitude vs. RPM response of my VR pickup well enough to satisfy the NCV. So instead I tried putting in a pair of clipper 1N4148's between the NCV input and ground. They are reversed to each other and in parallel, to ensure that we don't get much over .7V at the input. That helped quite a bit. I can now start reliably and rev to about 4K. I'm going to play around with the cap now to see whether the circuit really wants more than the current .01uF, or less. The one bit of good news is that I kept the engine at 3K RPM for a couple of minutes, with no loss of tach... so it's looking like Sportage's theory about the VR conditioner is correct... now if I can just get this one to cleanly rev to redline....

Update 4: I think I've got it. Adding another .01uF in parallel definitely made things worse, so I removed it and the original cap, leaving just the side-by-side 1N4148's. Between that and a maximal pickup gap, things seem good. It's a little late to do much red line revving, but she revved cleanly to 5K and I have a good feeling about the setup at this point. I'll update tomorrow after seeing how well she does going to red line. Funny that neither the stock VR circuit nor the NCV would work with any capacitance on my bike. I also ran again for a couple of minutes at 3K... no glitches!

After all the adding and removing parts for testing, though, there's a bit of a bird's nest next to the NCV. If I'm really bored I may go back and redo it, but I think I can live with it for now.

Thanks,
Paul

[sportage4x4]

i think i still have the particulars of the last install i did where i used the NCV. i'll check monday and let you know what i ended up with. from memory it was a 2.5" 36-1 wheel, a VR pickup off a carbed R6 mounted on the points plate of the GPz kawi engine. i used the regular DIY Autotune supplied shielded tach cable, shield terminated at the MS. the MS was a full size unit, v2.2 board, with the C64 MS2 daughtercard.

[SQLGUY]

Well, better, but not there yet. With no capacitance and the side-by-side diodes, I can rev to 6K or so, but that's about it. Next thing, I guess, will be to try the same setup but with more resistance, or with some load resistance parallel to the diodes, since these types of things helped when I was doing some of the other testing. I also need to take another look with the scope to see what the input waveform looks like at 6K and see if I can get an idea of what's causing the loss at that point.

Maybe a REALLY small cap would help? Something that will put the corner point of the filter around 3KHz = 5000 RPM? (should be about 3.5nF at that point)

Sportage, thanks for the help. One note here is that this trigger wheel is more like 3.125".

OK, update on this one: I went to scope the input to compare what's happening at the point where tach gets lost. But, when I went to start it, I was getting weird sounds and a bunch of backfiring. Pulled the plugs, and they were totally fouled/flooded. Installed new plugs - mostly the same thing. Just to check, removed the scope probe from the input, and it fired right up. So, with the current setup the input is actually so sensitive that the capacitance of the scope probe is enought to throw it off. Makes it kind of hard to troubleshoot.

Update 2: Apparently I can't tell the difference between 1 and 2. I just noticed, while looking at the data sheet again, that I transposed the Adj and IN1 pins when I was designing my circuit. I guess the amazing thing is that this worked at all. Either later tonight, or tomorrow, I'm going to retry this with the original circuit and correct hookup. If that doesn't work, I'll swap in the other NCV that I bought, since I wouldn't be surprised if I blew the - side of the comparator on this one.

Cheers,
Paul

[sportage4x4]

i was going to ask to have a look at your circuitry because this shouldnt be that hard

hopefully you didnt toast anything.

[SQLGUY]

Well, I just tried putting the original (15K, 16K, .01uF) circuit back in place, with correct wiring, and, while pulses were making it through to the CPU (as evidenced by the IAC activating), it wasn't getting enough of a signal to synch. As I was writing this post, though, I remembered that I had found that I originally had to set uS to rising edge in order to get synch. Since I actually had had the inverting and non-inverting inputs of the NCV switched initially, I just went back and tried setting it to falling edge. Sure enough, it fired right up... but still won't rev past 6K.

Tomorrow I'll see if I have any better luck scoping the signal, and I'll probably try another of the (15K, 16K, diodes) circuits, to see if that gets me any further.

Cheers,
Paul

[racingmini_mtl]

Maybe you could try an LM1815 since it has an adaptive mode which could help ignore noise at higher RPM. I know trying yet another circuit is not ideal but if you have a spare lying around or get some samples it could be worth a try.

Jean

[SQLGUY]

Looks like it's good. I put back the 15K , 16K, diodes network and I can now rev cleanly to red line with no loss of synch (trigger +/- stays at 0).

Here's the result: NCV tacked piggy-back onto the MC34072, with pins 8 and 4 bent down and in to pick up 5V and ground. The results are a bit rough because of all the parts swaps for troubleshooting purposes. I had damaged the pad of C30 back when I was troubleshooting my original rev limit problems, and when I went to remove the load resistor I had being trying in there last, it finally gave up the ghost. Oh well, it's not needed anymore anyway, nor are R26 and Q2. The output (pin 7) is jumpered to the pad that used to hold the collector of Q2.



Believe it or not, the bigger resistor is a 1/4W carbon film, and the smaller ones are 1/2W metal films. I had to double-check the packaging to make sure they'd sent me the right parts! Cool modern technology.

I do still need to do some longer-term testing to make sure I'm not losing synch at any of the trouble points (1K, 3K, 6K), but since it was clean at 3K when I had it miswired, I'm feeling pretty good about that being solved. Will update when it's a bit warmer. (when I first tried this latest configuration, the first rev test was kind of iffy, but I think that was because it's about 0 degrees Fahrenheit here and I was being a bit mean to the bike asking it to rev right off like that... after a (slight) bit of warmup, it was hitting red line with no problem).

At least I got to verify that COLD start works really well. Stock, with carbs, most guys find it impossible to start one of these bikes when they're this cold without starter fluid. Mine fired right up after about 1 second of cranking! ... not that I have any reason to be riding when it's this cold, but it's good to know the bike would be up for it, at least. :-O