PCM Hammer Suite development

[antus]

try putting one of them between vpw and gnd. if it doesnt work try smaller values. It will filter some frequencies out of the waveform and most definitely will change the signal. In a helpful way? Dont know.

[MudDuck514]

Hi all;

If any of you with the AllPro also have a Y-cable, what happens if you connect the cable and leave the other connector empty?
If it IS a capacitance related issue, that should be a quick way to test!

Mike

[antus]

I dont think its all, allpro. Mine works fine as it is. I believe NSFW said that just leaving the other connector empty was enough for his.

[NSFW]

No, I need something attached to the other side of it. The Y harness by itself isn't enough. At 1x speeds everything is fine, but 4x is crap... kernel upload never succeeds, which is kinda good because it never gets far enough to erase anything.

But with the Y harness plus another interface, the AllPro is completely reliable at both 1x and 4x speeds.

At first, I thought I'd broken PCM Hammer somehow. Then I thought I'd broken my AllPro, so I bought a second one. Then I hought that some code changes that should be (and are) meaningless were really critical. Then I realized the AllPro just worked some days and sucked some days. Then I realized that the key was to have something else plugged into the other side of the Y. There was some frustration involved.

[Gampy]

I shall try it ...

Antus, That begs me to ask about your bench harness (loom), what gauge, length of wire and connections do you have between DLC Pin 2 and PCM Red Pin 58 ??

I'd guess my DLC pigtail is 20ga and PCM pigtail is 18ga, total length is about 18 inches, they are soldered together.

[antus]

i couldnt even tell you, I just keep plugging what I need in to it. So it probably is longer than a direct connection. All I can say is I have not seen a problem. But I cant say when the last time I used a short direct connection was, if ever. I think probably the allpro to a P01 was actually the first thing on the bus when I was looking to develop the 4x firmware for the allpro an before we had any flash software, but that was quite a while before NSFW and youself joined in. viewtopic.php?f=31&t=4632

[Gampy]

Antus, I suspected you didn't have a simple bench harness.

Ok, tried it, definitely seems to help, but ultimately still fails after a few minutes. I was able to do multiple actions involving kernel uploads, not typically doable. So it does appear to help.
So, I doubled up the caps, two connected the same way, didn't change a thing.
Do I need to series the caps ??

I'll see if I can find some different values after a nap ...

[edit]
Ok, haven't made it to nap yet, this cap in Series thing was (<- note was) bugging me.
It just completed it's 11th Test Write, Starting 12 now, kernels up, processing ...

I would have to say it is working far better, I don't remember ever getting this much use out of it without another device on the bus.

Hmmm, the 11th had one retry, 12 is done with 3 retries ... hope this is not a symptom to failure!
Starting #13, kernel is up, 13 done, 3 retries.

Whoops, 14th Test Write failed: Kernel upload 87% complete.

[edit2]
The caps may help, but they do not solve the problem ... At least what I've tried.
It's like it over heats ... Works good at first, then degrades to failure. Let it set a couple hours and again works good at first then degrades to failure.
I dunno know ...

[NSFW]

FWIW, my wiring is around 18" total, 16 or 18 gauge I think, and I used wire nuts.

Putting caps in parallel gives the sum of the individual values, so if you have more, add more, and see what happens.

[antus]

come to think of it, the other option would be widening the timing values in the firmware of the allpro. you should be able to recompile with nxp mcu xpresso (free) and flash it in with a standard usb serial interface. https://github.com/ObdDiag-Net/allpro/b ... bd/j1850.h https://www.nxp.com/design/software/dev ... presso-IDE

Also I notice there is only 1 set of VPW timing variables on the official github code. In my implementation I hard coded values for 4x and I opened some of the timeing up a fraction at that speed because I found it was needed for reliability. Maybe this is a firmware regression.

[Gampy]

I just don't know ...

I do feel it's a hardware issue, if it was a software issue it would not degrade over time unless it was a memory leak or the like, then a reboot would bring it back to life, it does not, it takes time off, to get it to work again.
How long, I don't know, definitely longer then a few minutes, I have never timed how long it takes.

I do feel there is a high probability there is a fairly simple solution ...