[info] VR + knock module

[Smoczy]

obraz.png
It look like it will be good even for single wire VR sensor.

As I show here, it works on emulating, but i not test it i real.

[Smoczy]

Or to be better than the existing discrete VR designs that don't use an expensive chip like the MAX992x.

At the moment the $2 MAX9924 wins by removing the requirement for skilled threshold tuning.
Tuning is only need i begining, for C2 + R10 ONLY,

My priority list is
1-2) same SINGLE pin VR/Hall input
1-2) auto-threshold
3) price does not matter as long as under $10 per channel

Single pin is done, only need to add/remove components, on the same layout, and make shorts, to change from 2 wire VR to HALL thru 1 wire VR.
And the price is less than half (or 1/10 if 10$ is the limit).

[kb1gtt]

As I show here, it works on emulating, but i not test it i real.

Without posted source files it's difficult to comment on the emulation accuracy.

Selection of components is critical for understanding if a schematic will work. The opamp selection is critical.

[mck1117]

Looked over your schematic.

I have a couple questions/concerns:

1. Some of the triggers that the max992x has trouble with are "single tooth per cam rev". That means they could be as slow as 4hz input at engine idle speed. Your peak detection circuit doesn't work below idle speed on a 60-2 (60-2 frequency is ~RPM because of 60 teeth). There might even be difficulty getting the engine to start as the 330k/100n RC will decay too fast during cranking. However - the tradeoff of a slower RC is that if you drop RPM rapidly (redline shift from 1->2?) and the voltage drops significantly, you'll start losing teeth as the sensor voltage could drop faster than the threshold circuit.

2. This only triggers on zero crossing in one direction, right? That means with a single-wire sensor you might end up with edges on the imprecise edge instead of the precise edge because you can't flip the polarity, even in the PCB design.

The opamp selection is critical.

The opamps are actually not critical. It doesn't have to operate close to the rails, and there's no huge bandwidth requirement. This is practically DC as far as the vast majority of opamps are concerned.

[Smoczy]

As I show here, it works on emulating, but i not test it i real.

Without posted source files it's difficult to comment on the emulation accuracy.

Selection of components is critical for understanding if a schematic will work. The opamp selection is critical.

Source files are posted as VR.zip

[kb1gtt]

I get this error when I try the simulation. Where did you get this lib from?

Error: Could not find include file D:/Elektronika/Spice/Models/Comparators/LM393A.lib

I did not find this lib on my C:\ on my KC7 WIN10 64bit install.

I got a TI SPICE from below, then updated the links for both chips to point to this new file. Don't worry about the .lib it will figure out the extention and work with the TI file from below.
http://www.ti.com/product/LM393A/toolssoftware?dcmp=dsproject&hqs=sw&#desKits

It now lets me probe, graph, etc.

I wonder how well QSPICE might work for this.

In the past with QUCS I was dubbing with using mutual inductance to emulate the VR sensor. That would likely provide a more realistic simulation. However with QUCS I couldn't get stable operation. I also tried LTSpice, but again stable issues. I wonder if QSPICE would potentially work better.

[puff]

donttrustit.jpg (25.9 KiB) Viewed 32443 times

[Smoczy]

I change R10=100k, R11=100k.
All measurments i made. Comment, please

[puff]

I'd suggest to check how it works with low voltage cranking scenario.

[kb1gtt]

I recall Honeywell has a good paper which details a bunch of math for VR. I also recall we have a copy on the wiki. That paper might help you. I once made a spreadsheet with that math in it. However that may have not been migrated when the wiki migrated to GitHub. See notes in this thread
https://rusefi.com/forum/viewtopic.php?f=4&t=994&p=20111

If that paper doesn't provide enough data / information to predict a voltage / energy level, from memory I think you want at least down to like 10mv to 50mv detection. Lower is better.