double ramps - where to start troubleshooting.

[holtm]

After much pre-Christmas testing and all was well, I now have a strange issue....
I seems like when I ramp from 0 to 100% in vixen the lights ramp 0 to 100 % twice. It's late, I'm having a hard time describing it.. here is some data points.
Vixen at 20% lights at 20%
Vixen at 40% lights at 80%
Vixen at 49% lights at 100%
Vixen at 50% lights at 0%
Vixen at 90% lights at 90%
Vixen at 100% light at at 100%

Anybody have any idea where to start? I'm thinking zero cross is off? I have my zero cross transformer about 50 feet from my first Renard8 and a long way from my last one, however they all act the same. Could it be a diode gone bad in the AC-> unfiltered DC board?
Any ideas where to start would be appreciated.
Thanks.

[P. Short]

At first thought, any one of those could be possibilities, the diode thing is what I'd check first.

--

Phil

[xmus]

check your oscillator... and check your renard code .....
the Renard code uses osc cycled to decide when to launch a pulse to trigger the lights.... if the osc was running 2x to fast, you might get that behavior.
Or if the FW was setup incorrectly relative to the 18.432MHz osc, it might do that.... you sure your using the correct firmware?

[holtm]

I've done some more checking. Looks like I'm getting a zero cross issue when certain controllers are plugged in. I don't know if it is because of the number of controllers or if something is wrong with the controller (diode, cap ???).
I'm not using an oscillator, just the 8MHz internal clock. I'm pretty sure it's not the software or firmware as some work fine as long a some others are not plugged in.
I'm going to do some checking on what boxes are causing the problems and see if it's quantity or specific boxes. It's too cold this tonight, I'll post back in the morning after it warms up and I can try some things.
Thanks.

[xmus]

In my experience the built in oscillators are often quite unreliable and fluctuate wildly, having said that, I've done a great deal of experimentation with the Renard code using the internal oscillator and have never personally had a problem, the problems listed above are related to PIC's in general, not our specific setup, which seems to work fine for me.

Can you grab an oscope trace of the ZC with the other devices plugged in, and unplugged?

[P. Short]

Are you using Ren-T, or some other source of zero-crossing?

One thing that may not be obvious ahead of time is the need for a resistor to ground on the zero-crossing signal. This makes sure that cable capacitance, etc, doesn't prevent the signal from reaching ground (or close enough to turn off the transistor on the Ren8). I don't remember what resistor value is best, but I think that something on the order of 4.7K should be fine. This resistor is present on the Ren-T board, FYI.

--

Phil

[holtm]

I might have to resort to that, my scope is older than dirt and I'm not very skilled at using it. If I was able to see the signal I would have to draw/describe it. I'll also thinking about getting another transformer and dividing the zc signal up a bit. Of course they would share a ground, so maybe it wouldn't work any better. Thanks for the input. I made boards with a place for the osc, but didn't populate it as my channel count is low enough for 19200. I may have to backup and put the osc's on.

[holtm]

Thanks for all the tips. I have it working now. The double ramp was caused by a bad power supply on one of my 40 channel controllers. It must have been feeding a spike into the zero cross. I changed the way I power the board and it works now. I had several other problems; from bad cat5 ends to a PIC that wouldn't talk (reprogrammed now ok).
One interesting issue.... I'm using two wires for the zero-cross and two wires for ground in a cat 5. In one connection one of the zero-cross wires was not connected. It caused a problem at the controller where it was connected even though the bad end was going to the next controller where it didn't cause a problem. I'm guessing inductance was to blame.
Thanks again.