I rebuilt a mono UT for analysis in this thread below and found numerous ways to induce
oscillations:
SWTPC Universal Tiger Rebuild For Analysis
I began to make minor changes such as moving grounds around and thought that it
would be good to have another one that could be left without changes as a reference
since I have 3 of the mono block UTs. The second one did not oscillate in any of the
ways that the first one did.
Started by looking for visual differences:
1. First (UTM#1) is on an aluminum chassis, second one is on steel (UTM#2).
2. Capacitor brands are different
3. UTM#2 has a lot of corrosion, probably from sitting in a damp environment
for many years. I should probably clean the grounds.
4. UTM#2 has nearly every solder joint with about twice as much solder as there should be.
a joint could be cold, but who knows?
Obviously semiconductors have wide variations in parameters and I don't have a curve
tracer to test each one. I suppose the VAS is a likely suspect since one with more internal
capacitance might improve stability acting as a Cdom cap.
I know for sure that many of these oscillated from past experience, yet a few builders
report no problems at all.
I wanted to keep the speculation on this issue in another thread just for clarity - so here it is.
I suppose that I should try UTM#3 to get another data point.
oscillations:
SWTPC Universal Tiger Rebuild For Analysis
I began to make minor changes such as moving grounds around and thought that it
would be good to have another one that could be left without changes as a reference
since I have 3 of the mono block UTs. The second one did not oscillate in any of the
ways that the first one did.
Started by looking for visual differences:
1. First (UTM#1) is on an aluminum chassis, second one is on steel (UTM#2).
2. Capacitor brands are different
3. UTM#2 has a lot of corrosion, probably from sitting in a damp environment
for many years. I should probably clean the grounds.
4. UTM#2 has nearly every solder joint with about twice as much solder as there should be.
a joint could be cold, but who knows?
Obviously semiconductors have wide variations in parameters and I don't have a curve
tracer to test each one. I suppose the VAS is a likely suspect since one with more internal
capacitance might improve stability acting as a Cdom cap.
I know for sure that many of these oscillated from past experience, yet a few builders
report no problems at all.
I wanted to keep the speculation on this issue in another thread just for clarity - so here it is.
I suppose that I should try UTM#3 to get another data point.
Last edited:
I'm reviving this because a member here has a failed UT that he is currently working on.
My basic thoughts are that stabilizing this amp was just wishful thinking in the design phase.
The logical thing to do is to measure the open loop frequency response to try and get an idea of what's going on.
Have people figured out the best way to do this on the bench?
I tested both of these amps with a function generator I could have just switched to a square wave to check the overshoot.
The unstable one goes off just looking at it wrong, so this might make measuring with a square wave difficult.
I suppose that I could cut the gain in half or quarter in the amp that works and see how the square wave response
changes and if it is still stable.
Anyone know of a good writeup on the best way to measure open loop frequency response of an amp on the bench?
Remove the input filter, measure ahead of the output inductor, and somehow shunt the AC part of the feedback.
If a big cap is used the response will be contaminated by the complex impedance at HF.
These amps are minimum phase, I believe, so the phase response can be computed based on the FR.
I can go to 11 MHz from the function generator.
I suspect that the first UT has a bad or failing component due to having over a volt of output offset, none of the devices
are matched so perhaps they are just very poorly matched the amp does work. The other UTM#2 has about .3V of output
offset, as noted in the other thread shorting the input drops both to the mV range as it should be.
My basic thoughts are that stabilizing this amp was just wishful thinking in the design phase.
The logical thing to do is to measure the open loop frequency response to try and get an idea of what's going on.
Have people figured out the best way to do this on the bench?
I tested both of these amps with a function generator I could have just switched to a square wave to check the overshoot.
The unstable one goes off just looking at it wrong, so this might make measuring with a square wave difficult.
I suppose that I could cut the gain in half or quarter in the amp that works and see how the square wave response
changes and if it is still stable.
Anyone know of a good writeup on the best way to measure open loop frequency response of an amp on the bench?
Remove the input filter, measure ahead of the output inductor, and somehow shunt the AC part of the feedback.
If a big cap is used the response will be contaminated by the complex impedance at HF.
These amps are minimum phase, I believe, so the phase response can be computed based on the FR.
I can go to 11 MHz from the function generator.
I suspect that the first UT has a bad or failing component due to having over a volt of output offset, none of the devices
are matched so perhaps they are just very poorly matched the amp does work. The other UTM#2 has about .3V of output
offset, as noted in the other thread shorting the input drops both to the mV range as it should be.
I wanted to mention that UTM#1 has all new components on the driver board. That board and its mate came
from Tubelab (George) on here and not all the components were installed so it could not have ever been powered up.
I believe that the outputs were used, but I do have new old stock if I need to try them.
UTM#2 and 3 are both used and according to the previous owner never failed. I wonder if driving them hard, don't
know if he did or not, partially damages the driver transistors to make the amp MORE stable?
My family's UT stereo version worked for a year or more before blowing up on the bench, we did not drive it into
clipping often. It blew up on the bench in the 70s, I repaired it, not knowing about the undersized resistors, and
then it was fine for years but blew up in the 80s when driven into clipping during a party. I was in another room
and informed that a speaker seemed to be nearly on fire with so much smoke coming out of it.
from Tubelab (George) on here and not all the components were installed so it could not have ever been powered up.
I believe that the outputs were used, but I do have new old stock if I need to try them.
UTM#2 and 3 are both used and according to the previous owner never failed. I wonder if driving them hard, don't
know if he did or not, partially damages the driver transistors to make the amp MORE stable?
My family's UT stereo version worked for a year or more before blowing up on the bench, we did not drive it into
clipping often. It blew up on the bench in the 70s, I repaired it, not knowing about the undersized resistors, and
then it was fine for years but blew up in the 80s when driven into clipping during a party. I was in another room
and informed that a speaker seemed to be nearly on fire with so much smoke coming out of it.
The grounding and layout are obviously quite bad on the UT amplifiers and I was recently thinking that
there's room around the perimeter of the board to place the output transistor emitter resistors, the zobel,
and get it all more compact with better grounding.
I was also given Plastic Tiger boards from another friend, which are essentially the same design with lower
supply voltage and all those components moved onto the board. I went through the boards and one has
massive burns and melted feedback caps, so it seems that the more compact layout does not help the situation.
Plastic Tiger compensation is nearly the same as the UT - wishful thinking.
there's room around the perimeter of the board to place the output transistor emitter resistors, the zobel,
and get it all more compact with better grounding.
I was also given Plastic Tiger boards from another friend, which are essentially the same design with lower
supply voltage and all those components moved onto the board. I went through the boards and one has
massive burns and melted feedback caps, so it seems that the more compact layout does not help the situation.
Plastic Tiger compensation is nearly the same as the UT - wishful thinking.
Of course what we really want to measure is the loop gain with regard to stability, but with a simple amp like
this I've always felt that the open loop gain will provide some idea of what's going on. I'm going to consider
if there might be a way to measure the loop gain on the bench that's not too complicated. I don't believe that
accuracy below 1 KHz is important for this amp, but I'm sure that others might offer a different opinion.
Two concerns are measuring the loop gain as built, and second in simulation to
experiment with improvements.
This is an excellent video on how to do it in simulation with LTSpice:
This is more advanced and includes the minor loop(s):
experiment with improvements.
This is an excellent video on how to do it in simulation with LTSpice:
This is more advanced and includes the minor loop(s):