VSS resetting to 0

[tobmag]

Lance,

When you reffere to your bench, what exactly are you using for testing your self, how is your GPIO built?
According to what spec, what Bias resistor do you use for the vss/iss?

What do you use as in/outputs, is it just the trans stim?

What code do you run your self at these tests?

Just to know that we really are comparing apple to apple?

I have now a new function generator on my bench, unfortunatly it can only put out max 1,6v but should be enough,
Good thing is that I can adjust the frequence and voltage stepless....

BR//Tobmag

[Bernard Fife]

Tobmag,

I have been using 4.127 with the MSQs you and gui67 have provided. I also tried 4.125 and 4.126.

My bench uses the trans stim (with a 12V square wave for the VSS) as well as a 40111A BK Precision function generator (usually at the 5K setting) with a sine wave output (though I regularly test square and saw waves too).

I use the default 4l60e (http://www.msgpio.com/manuals/mshift/assembly.html) build for all the circuits on a standard GPIO board.

I'll compile some new code in the next day or so that bypasses ALL the filtering, smoothing, and averaging on the VSS if the VSS input masking is set to zero. That should help narrow things down.

Lance.

[tobmag]

Lance.

Ok then we have that sorted even though I thought it was about that way.

Did some more testing today with the new function generator I borrowed.
I can replicate the problem at 1300hz(+-100hz) everytime, I have made a new log that is attached making a sweep over 1300 and back a few times.

With 1300hz I mean what I put in at the pigtail, actaul VSS frquence in TS is usually of by 40-60hz(always less)
I attached the scope image of the signal I put in at the pigtail and also from the VR1 at the 25pin header.

[tobmag]

And then the log file

[tobmag]

And then a second run today, the perfect run

What I did was setting the VSS input masking to 0 (zero) and then the oscillation at 1300hz is gone.....you can se some minor dips, the speedo is just slighly irratic but nothing compared to before, pleasse see the attached log file.
So I guess this will norrow things down?!

BUT

I also noted that the load is ignored in the gear shift table.
It will only change gear as to 20Kpa....
Im running the GPIO as stand alone, using the 0-5v output in the trans stim as load
Load is changing as it should but if I put in 100% load its still changing gear according to 20Kpa.....

Also attached the current .msq


BR//Tobmag

[Bernard Fife]

Tobmag,

I tried your latest MSQ, and I can't generate the sort of behavior you are seeing.

Your 5V square wave screenshot shows a range from -1.75 volts to +3.06 volts at the 25x2 header. Is that correct? The processor PT0 I/O pin should not be seeing voltages lower than zero (with -0.3V as the absolute minimum according to the datasheet). This suggests to me that something isn't right in your test set-up or build.

Is the scope grounded properly to the same ground as the GPIO board? Is the signal generator also grounded to that spot?

If this was mine, I would also look over the VR1 circuit very, very carefully. It is easy to get the transistors in backwards, or bridge the pins with solder or flux.

Lance.

[tobmag]

well if thesignal coming into theprocessor is DC I had the scope in the wrong coupling, had it set for AC.

Below in DC coulping at the 25pin header.

[tobmag]

Yes all grounds to the same place, did a quick test and used ISS as VSS input just jumber VR3 and PT0

Same thing going on at 1300hz if the filter is on.....if I turn it off its (almost) 100% ok.

Re-checked all components and orientation of the transistors, only thing that is not according the build instruction are the transistors that are named 2N3904-A47
Instructions says 2N3904FS-ND

BR//Tobmag

[Bernard Fife]

Tobmag,

Ah, that explains it. I am focusing on the scope shot for a few reasons - one of which is that the trace is a sample, so it might not be capturing any signal problems as they happen. I had an issue recently (not with the MShift/GPIO) where my signals (generated with an AVR uP and custom code) looked perfect on a scope, with a variation of <<1% over a full cycle (60 teeth in this case). But I couldn't easily see on the scope that this signal had a considerable but gradual variation within a cycle that messed with the filters' ability to do their job on individual teeth.

Does the VSS also clean up if you turn off the filtering?

Lance.

[tobmag]

Does the VSS also clean up if you turn off the filtering?

Sorry Lance but what do you mean?

If you look on the last log file thats the one with the filter off and compare to the log in the post earlier today, thats with the filter set to 50%

Any idea what happend with the load index in the gear shift table?

BR//Tobmag