cold start issue, voltage related

[dave]

So I THINK I got it. All of my testing tonight was with the car warm and in running condition so I will see in the morning.
The short of it is that one of you answering "turn R56 fully CW" when I asked "things to try" would have saved me a few hours in the cold.

So bear with me while I have the same revelations as anyone else who's scoped the VR circuit.

In observing the VR signal I noticed that it's amplitude is HIGHLY dependent on SPEED, and at low cranking speeds can be lower than 5V despite normal running producing a 20V signal.

Here it is while running at idle. what's that, 30Vpk-pk? nearly 50 if you count the irregular waves the missing tooth causes. That's quite a bit.. It seems I forgot to save a pic of it while cranking. It's substantially less.

The "zero crossing detector" circuit does not always put out 5V signal. instead of it converting it to a square wave it more just clips it off at 5V and cleans off the noise. Not sure if signal is divided down or just clipped. For some reason I assumed the transistor was there to boost signal but I think a FET would be more appropriate since we're dealing with nearly zero current after an RC snubber network. Anyways this means that our signal headed for the op-amp could have a peak of only 2-3V. At idle speed or higher it looks like a solid 5V square wave but cranking is a whole other story.
In fact, here it is while cranking.


where it goes to solid 5V square wave is when the engine started.

Turning R56 CCW RAISES the op-amp reference threshold at pin 3, when it's fully CCW it's up over 4V! Much low-speed signal is lost. Why should it EVER need to be this high? Even 6-8 turns in as suggested it's much higher than the low signal we get when the motor first starts cranking.

Most of those signals are ignored until they reach >3.5V

To get it down under 3V I had to go FULLY CW on R56 then 6-8 turns in on R52.

Now I get a signal as long as VR can get over 2.3V.


This is Tsel compared to VR input (TachIn). Hopefully that looks right. Seems like it's running fine w/ it.

And sidenote, Q23 seems totally unnecessary. And R42+R55 being 1K seems like a lot of current for that divider. Actually I might do quite a bit differently given what I observed tonight. I realize this is sort of a one-size-fits-all circuit but man. It wouldn't take much rework to get this reading a 1V signal off one of those OMG HOLD YOUR BREATH AND CROSS YOUR FINGERS CAUSE THIS THING CAN BARELY TURN OVER kind of starts.

Anyways I'll see in the morning when things are cold and slow if this fixes it.. Feeling lucky though. It's reading a much lower V signal now.

[john.p.clegg]

Looking Good..Keep it up.

John

[DaveEFI]

Sounds like your VR sensor is producing more volts than most. Highest voltage I've ever seen is about 10v. But I've not seen them all.

[elaw]

Glad to hear you got it working!

I think the reason nobody suggested turning R56 clockwise is you either have the wrong part in there, or it's installed backwards! As designed, turning R56 *counterclockwise* reduces the input voltage threshold, and of course clockwise increases it. So something's wonky with your particular system.

And Q23 is actually there for a very good reason... well a fairly good reason.

If you look at Q23 along with Q22, you'll see they're doing the same job... they're level shifters. U7, which is called out on the schematic as an opamp but really should be a comparator (and I think later BOM's specify a different chip) doesn't work well with input levels very close to ground. So Q22 shifts the level of the input signal to be about 0.6 volt above ground, and Q23 does the same for the "reference level" - the voltage coming from R56.

And I'm with ya on R42 and R55... theoretically they should be 100K. But the lower resistance doesn't really affect anything.

[dave]

So. It's 22º and it started fine this morning.

Thinking about it a bit this morning, I think there should be a resistor inline w/ R52, as giving it 100% feedback is a bit more than just hysteresis and kind of brute-forces the R56 adjustment right out the window. I probably should have adjusted that one first so that R56 had some better wiggle room.

But yeah if what you're saying is true then both these pots are operating backwards. how strange. But the suggestion of maxing out R52 instead of turning it down to a short makes way more sense, as well as turning the reference down.

[dave]

If you look at Q23 along with Q22, you'll see they're doing the same job... they're level shifters. U7, which is called out on the schematic as an opamp but really should be a comparator (and I think later BOM's specify a different chip) doesn't work well with input levels very close to ground. So Q22 shifts the level of the input signal to be about 0.6 volt above ground, and Q23 does the same for the "reference level" - the voltage coming from R56.

And I'm with ya on R42 and R55... theoretically they should be 100K. But the lower resistance doesn't really affect anything.

That makes sense. I looked at it like they were trying to make the adjusted reference super stiff which wouldn't have been necessary. Small level shifts like that, why not just use diodes? Less parts that way.

And yeah nobody is going to miss that 1/10 W of power. When I get into that engineering mode nothing is safe.. ;)

[DaveEFI]

My feeling is there needs to be a rather more comprehensive section in the manual on adjusting those pots. (If it does exist, I've not found it) Generally, it seems to imply they will work ok on their default fully anti-clock position. But even on my on my JimStim, they won't. I get no tach signal at all until R56 is adjusted. And it needs to be checked at high revs too. This on several V3 boards - not just one. But I've never had to adjust R52 from the default.

I don't claim to have anything like the experience of many on here - but have found the setting of these pots (or lack of it) have been the cause of several very dissatisfied users.

[dave]

agreed.
I think even a simple paragraph about how they work or how to check them might have tipped me off about them operating backwards from default. "turn them both fully CCW" is a TOO simple of a setup for this.
Most people probably don't have a scope but it's easy to turn on the power and check where the reference voltage is set.

[DaveEFI]

It might be possible to set them near enough by using a DVM set to AC connected to Tsel. I've found if you get an approx 5 volt pulse there you'll be ok. But since I do have a scope, haven't tried it. But will do next time I have one on the bench.