The clean test would be - run this thing the way you did it before, when it tripped, but see the actual measured DC at the output through the integrating RC at the same time.
This way, if it trips again, you will see what was the offset.
There again, measure voltage with the filter installed. All my meters read the same with it connected. Maybe your input transistors have enough mismatch to cause offset to reach that level before the servo began correcting.
Valery,Jeff ,yes i will but i try to find an easy way to disable the servo and see what happened😉
The same thing is measuring here on the protection board
Attachments
Last edited:
Then just short the servo filtering capacitor (0.1uF), removing the 100R between this cap and the servo output, if you don't want to abuse the OpAmp.
That will show the natural misbalance of the circuit with no servo involved.
Yes, this is the right place to measure.
Your protection input filter if tuned for much higher cut-off frequency, than ours.
Yours ~ 1.6Hz
Ours ~ 0.07Hz
I'm now testing with an output stage but no load. I think the VZ-X4 front end would benefit from some means to tune out DC offset without the servo connected. If I play irregular high frequency signals, I can still get DC offset to drift well above 1V. The Vertical CFA is staying perfect under 1mV no matter what I throw at it.
Your DC detecting circuit is roughly 20 times faster than ours.
So, in some cases, it may recognise some slow low-frequency AC waveform component as DC.
I managed to trip our DC protection with some high frequency hard clipping with the Vertical CFA. I've got it tuned to trip around +/- 1.2V. I'll try the VZ-X4 next.
I couldn't get the protection to trip with the VZ-X4 until I was thrashing it around 30kHZ. I've got another one built as per schematic. I'll see what it does next.
I was mistaken on which input I was testing. I tested the unchanged VZ-X4 first. The one I modified performed mush worse. It's tripping right at clipping at anything above 15kHZ. I guess this is showing the difference in input transistors. The untouched one must be a better match.
Well, some DC at hard clipping is fine - neither NFB, nor servo are working properly in deep saturation conditions. That's one of the dangers of hard clipping and one of the reasons to have a good protection system 😉
I changed everything back to original, then changed the integrator cap to 1uF. I also butchered a 1M resistor in series with pin 2 of the servo, but that made no difference.
I actually redesigned the DC detection circuit to trip at lower voltage when I did the redesign to optically isolate the digital side. These are much more sensitive than the previous design. I was concerned at first they may false trigger more often, but it hasn't been an issue yet. The SMT version is a little more hair trigger than the through hole version though.