[rusEfi] Smart 452 Roadster with "Bigblock" Engine #33

[nighthunter]

You might get a pinch better signal from your op-amp board if you put a weak 2.5V on one of the op-amp inputs. I think the 9926 has that on the negative input. AKA keeps your signal at about 2.5V instead of allowing the overall potential to wander near the op-amp rails.

Sweet, is that a STM32 running rusEFI firmware? Very nice.

If you want my comments / review of the PCB layout and such, I'd be happy to comment either publicly or privately.

Yes its there, the VR1 is TL431 so it is adding 2.5 as a decision level. And yes its STM what else... The plans will be free to everyone, just wait i bit. I am bussy at work now. It needs a little work to tidy the plans. When i have some more time i will hook the plans here. But if you want, ill send you board files, but be warned, it is draft, so there are no real res/cap values added and so on. Do you want the drafts? Hmmm i think yes, or? PM me your mail.

[Rhinoman]

By the way do not try to use LM1815 for VRs.

Why not? they were used in many ECUs.

[kb1gtt]

The LM1815 is weaker in terms of common mode noise rejection. So it's more likely to have false edge detection caused by electrical noise. As well the LM1815's auto gain control is less optimized, which also causes incorrect edge detection at specific RPM's. The MAX chip is enough of a pain to deal with, the LM1815 is more of a pain to deal with. They both work, but the LM1815 requires more validation and more effort to ensure you do not have problem. There was a post about this the LM1815 at the below thread.
http://rusefi.com/forum/viewtopic.php?f=4&t=227&p=914

Looks like the internet is broken, as the forum pictures are now broken. If we can get the scope capture back, it would be handy. It showed the weakness of the LM1815 pretty clearly.

[AndreyB]

forum pictures are now broken. If we can get the scope capture back, it would be handy.

please be more specific, which exact post?

[nighthunter]

With LM1815 i have a lot of missdetections. It is very poorly designed. You must have very big signal in compare with noise for this IC, it is not differential.

[puff]

What was the trigger wheel and what do you mean by 'a lot of'? How do you measure these misdetections?
LM1815 suggests several various configs and might need some tuning.
So far I used it with a 36/1 wheel for my diy tacho, and it provided somewhat decent signal during my half an hour trip) Not sure though if this signal would be suitable for the ecu...

[nighthunter]

I have a 60/2 trigger wheel, i measured in with scope and compared with the output signal.

[Rhinoman]

Problems with the LM1815 are usually down to poor PCB layout, insufficient decoupling or too much current into the inputs.

[puff]

I have a 60/2 trigger wheel, i measured in with scope and compared with the output signal.

so you calculated the number of errors per 60 pulses? per 600 pulses? per 6 mln.pulses?

insufficient decoupling or too much current into the inputs.

decoupling - means capacitance at the output of the circuit? or at the input?
too much current - a wrong resistance at the input?

[nighthunter]

Ive compared the signal (output from LM1815) to the output on original ECU. (Inside there is CJ910/CJ911). At low rpms and deceleration 1-3 tooths were not identified.

[puff]

1-3 teeth per each revolution?