Electronic Distributor Adapter

[90TurboVan]

My schematic program doesn't change where the box is when you use a male or female connector, even though it is actually different. I changed it in paint so that 1 is at the bottom. The readme was right and the schematic is now correct.

I have 1k input resistors on the inputs to prevent damage if the inputs get more than 5 volts. Since the HEP outputs are 8 volts, they're needed with hybrid adapters, but the 2.4 sensors run at 5 volts. I removed the input resistors as well as tied the reset pin to 5 volts which should eliminate resets as a possible problem. I'll test it out tomorrow, nice and warm outside now...

Maybe I'll de-pot a Neon computer and see how the cam/crank sensor inputs are handled.

[lametec]

The 1k resistors may keep the internal protection diodes in the PIC from burning up, but won't keep transients from latching it up. A latch up doesn't damage the PIC in any way, just causes it to stop working properly until power is removed or the watchdog timer resets it.

The internal protection diodes are basically there for ESD purposes. The maximum/minimum input voltage with reference to Vdd and Vss is +/- 0.3V. Even though the sensors run on 5V, there's probably a lot of noise on that 5V, causing it to go above 5V constantly. One little spike over 5.3V, and the PIC can get all kinds of confused. And when the input is grounded, anything below -0.3V will confuse it the same way. Both these scenarios can cause latch ups.

So do some filtering/limiting on your inputs to make sure they stay within the maximum/minimum voltages of the PIC and you'll be better off. Even if it works 999 out 1000 times without it, the one time it doesn't your device might just cause a serious accident.

Just some advice from someone who's been there and done it.

[90TurboVan]

I understand what you're saying, but the datasheet claims maximum clamp current at +-20ma. The 1k's adhere to that up to +25 -20 volts on the inputs to the module. Your mention of the watchdog made me check my program, I forgot to disable the watchdog in the config bits when I changed over to a new source file. The 5 volt supply to the sensors comes right from the computer's internal 5 volt supply. I highly doubt that it's all that noisy, certainly not with spikes up to 25 volts. Chrysler uses a SMPS to supply the computer, so it should be immune to input variations (I'm just tieing into the sensor outputs on my Neon for testing).

I know that there are areas that the circuit can be improved, but I have boards made. If I use all the ones I have, I'll incorparate several modifications to new ones. Of course, then the price goes up since there will be more parts/bigger board...

Sounds like you know MCU's, what were you working on?

[lametec]

It's not a matter of too much current through the protection diodes. It's a matter of the input voltages being too high or too low. Anything above 5.3V on the input when running at 5V can cause unwanted behavior or chip damage. Same for anything lower than -0.3V. The clamp diodes don't do squat until your 0.6V over/under max/min. They're basically there for ESD protection.

I'm not saying input limiting is your problem, but it's something you should consider in your circuit design. Running the inputs through some op-amps is one way of doing it, and should add only pennies worth of components to your project.

I've used PICs in a few projects, but the "biggest" one is a 4 channel EGT display/logging gauge.

[90TurboVan]

I see, the protection diodes have the standard ~0.7 voltage drop, didn't factor that in. I'll have to put some Shottkys on the inputs then, and I really like the opamp method.

[lametec]

If you use op-amps you don't need the Schottky diodes. An op-amp powered by the same 5V supply that powers the PIC pretty much guarantees that that PIC won't see anything above 5V on the inputs.

[turbo.shaft]

Any new results from retesting? I would like to get this module soon if possible when debugged for my omni for some spring fun.

[90TurboVan]

That's why I liked the opamp method, no discrete components besides the opamps, however, after reading some datasheets and testing, you usually can't operate an opamp with input voltages higher than the supply.

Voltage at the PIC inputs with input voltages above 5 volts are around 5.8 volts, so I'll put Schottkys or Zeners on them. I don't think that's the source of this particular problem though. The 5 volt supply for the sensors comes from the ECU's internal supply, so it should be clean enough.

I updated the rar with the newest Non-NGC code, if you want to take a look at it lametec or anyone else and tell me what I'm doing wrong, I'd be eternally grateful.

[90TurboVan]

Put a Zener clamp on both sides of the regulator as well as a reverse polarity protection diode. Went to NPN transistor pulldowns on the inputs since I don't have any Schottky's laying around. Just inverts the inputs, won't let them go over Vcc no matter what the input. I'll test it out tomorrow when I can get it in the warm garage.

[turbo.shaft]

any test results with the latest changes?

[90TurboVan]

Still doesn't work.

[turbo.shaft]

I think that not having hard points is screwing up the emulator. Try the NGC motor. I think fact that all the hard trigger points coincide to the old pickup points it may be easier to fire the coils rev/properly. Do you have access to a NGC vehicle for testing?

Considering that using the stock pickup firing the neon coils is fine you know the problem is confined to the input side of the equation at least.

A solution for pre-ngc controllers may be to use the same crank trigger from a mazda/ford early 90's setup.

[90TurboVan]

I'm actually not even emulating the HEP yet, just trying to get it to synch to the cam and crank signals. Megasquirt can do it, and I'm using basically the same method for detecting TDC's. I made up a program that just outputs a pulse on every rising or falling edge on the crank input, it should show any noise that's getting in and narrow it down to a hardware or software problem. I'm still going with software...

I should note that the HEP outputs that the hybrid adapter uses are different from the cam/crank sensor outputs. That the hybrid one works doesn't necessarily mean that the 2.4 one should.

I do have an NGC vehicle (several), but I'll probably end up with the same problems since both are based on timing teeth to find TDC. NGC just has a different tooth pattern.

[90TurboVan]

Well, it appears that there are random falling edges being detected. That's interesting since there are no corresponding rising edges...

There is no apparent noise on the inputs, they both are clamped to Vdd and Gnd. The 5 volt output from the regulator has no noise and less than 10mv p-p, same for the ground lines.

I'm going to assume that it's a code problem with false Interrupt On Change interrupts being generated. Checking it out now...

[WVRampage]

Any updates on the new design,let me know when its right and Ill take one.