I've written here several times about the pain and frustration caused by the infamous Universal Tiger that I built
and repaired many times as a young kid. I believe that there are multiple issues with this design but the major
ones are poor high frequency compensation that was later corrected by the company, mutual conduction, and an
issue where in clipping the reverse Vbe rating of the driver transistors is exceeded.
I first reported about the reverse Vbe issue in the Universal Tiger here:
http://www.diyaudio.com/forums/soli...-tiger-improved-simulation-2.html#post2585028
And the larger Tigersaurus here:
http://www.diyaudio.com/forums/solid-state/189970-swtpc-tigersaurus-250w-amp-simulation.html
I've also pointed out how surprised I was to see that the Bryston 3B is nearly a copy of the Tigersaurus, or
Tiger .01 front end. The Bryston output stage is more similar to the Tiger .01 schematic linked here in .pdf
http://baselaudiolabs.googlepages.com/TIGER-01-207B.pdf
Link to the Bryston schematics, I used the first one for this simulation:
http://www.bryston.com/PDF/Schematics/3B-8BST_SCHEMATICS.pdf
Similar to the Tiger .01, the output devices employ a darlington connected driver. The Bryston takes that output
stage and builds on it making it even more complex where both NPN AND PNP output types are used on the positive
and negative rails and the EF is a triple. I'm going to name this the "can't make up our mind" output stage. But
otherwise it is very similar to the Tiger .01 output stage.
I remember reading years ago that the Brystons are subjected to a very extensive torture test and so I wondered
what they might have done to correct the reverse Vbe issue. One possibility is that the reverse Vbe current is much
lower because the feedback resistor values are scaled up by a factor of 10. The simulation showed that it does
have large reverse Vbe (10V) when clipping into 4 ohms. The simulation is attached. I simply want to show that
this design also has the reverse Vbe issue and have not done any distortion or other sims but obviously this is an
option for anyone who is interested. If the rigorous testing by Bryston includes deep clipping, it might be
confirmation that the lower reverse Vbe current does not damage the junctions.
and repaired many times as a young kid. I believe that there are multiple issues with this design but the major
ones are poor high frequency compensation that was later corrected by the company, mutual conduction, and an
issue where in clipping the reverse Vbe rating of the driver transistors is exceeded.
I first reported about the reverse Vbe issue in the Universal Tiger here:
http://www.diyaudio.com/forums/soli...-tiger-improved-simulation-2.html#post2585028
And the larger Tigersaurus here:
http://www.diyaudio.com/forums/solid-state/189970-swtpc-tigersaurus-250w-amp-simulation.html
I've also pointed out how surprised I was to see that the Bryston 3B is nearly a copy of the Tigersaurus, or
Tiger .01 front end. The Bryston output stage is more similar to the Tiger .01 schematic linked here in .pdf
http://baselaudiolabs.googlepages.com/TIGER-01-207B.pdf
Link to the Bryston schematics, I used the first one for this simulation:
http://www.bryston.com/PDF/Schematics/3B-8BST_SCHEMATICS.pdf
Similar to the Tiger .01, the output devices employ a darlington connected driver. The Bryston takes that output
stage and builds on it making it even more complex where both NPN AND PNP output types are used on the positive
and negative rails and the EF is a triple. I'm going to name this the "can't make up our mind" output stage. But
otherwise it is very similar to the Tiger .01 output stage.
I remember reading years ago that the Brystons are subjected to a very extensive torture test and so I wondered
what they might have done to correct the reverse Vbe issue. One possibility is that the reverse Vbe current is much
lower because the feedback resistor values are scaled up by a factor of 10. The simulation showed that it does
have large reverse Vbe (10V) when clipping into 4 ohms. The simulation is attached. I simply want to show that
this design also has the reverse Vbe issue and have not done any distortion or other sims but obviously this is an
option for anyone who is interested. If the rigorous testing by Bryston includes deep clipping, it might be
confirmation that the lower reverse Vbe current does not damage the junctions.
Last edited:
Member
Joined 2009
Paid Member
> large reverse Vbe (10V)
Reverse Vbe is not destructive unless the Power is excessive. If current is not-large, it stays together.
I agree Dan was pushing envelopes. Both good and bad push. It's much too long ago for me to re-visit today.
If Bryston used larger resistors for the size of the drivers the current may be "safe".
Thumb-counting: looks like 0.6W in breakdown which is a lot for a small part, though it will be half-wave at worst. May be more going on.
Reverse breakdown "does" cause permanent shift in forward Vbe and hiss. The effects are small, but threaten calibration in microvolt input stages, and sometimes drives up Noise Figure in microphone inputs (subjected to line level).
Reverse Vbe is not destructive unless the Power is excessive. If current is not-large, it stays together.
I agree Dan was pushing envelopes. Both good and bad push. It's much too long ago for me to re-visit today.
If Bryston used larger resistors for the size of the drivers the current may be "safe".
Thumb-counting: looks like 0.6W in breakdown which is a lot for a small part, though it will be half-wave at worst. May be more going on.
Reverse breakdown "does" cause permanent shift in forward Vbe and hiss. The effects are small, but threaten calibration in microvolt input stages, and sometimes drives up Noise Figure in microphone inputs (subjected to line level).
Member
Joined 2009
Paid Member
http://www.diyaudio.com/forums/solid-state/245619-tgm8-amplifier-based-rod-elliot-p3a.htmlWhich one was that ?
Remember, this is judging by my ears
I ran this and first which schematics did you use for the output protection, it is very
different from what I have.
Also, interesting that you added zener diodes because I thought I tried that on the Tiger
and it didn't work from memory, I could be wrong.
Did you notice that the zeners are not conducting in their zener mode but rather in the
forward direction? So, regular diodes would work.
Very interesting.
Bryston 3B III complete SPICE Simulation with mods.
Try this link for the last updated prints that I know of for the 3/5/7/8b and 370.
https://www.manualslib.com/manual/219483/Bryston-3b.html?page=2
I just updated the model to last 1997 prints. Looks like they changed the input
impedance and tweeted some of the RC networks. The output is the same as the 3b III board (Bryston design). The link has all but maybe not the earliest prints.
You right I was in hurry and did not save my last changes so reverse the 2 diodes. The voltage was picked for the best Vbe of the 2 VAS driver transistors.
Just a academic exercise to model a complete amp and see what it has or not. I know in the day that the Bryston was a much better choice than the Tigers but I was broke college grad at the time.
Try this link for the last updated prints that I know of for the 3/5/7/8b and 370.
https://www.manualslib.com/manual/219483/Bryston-3b.html?page=2
I just updated the model to last 1997 prints. Looks like they changed the input
impedance and tweeted some of the RC networks. The output is the same as the 3b III board (Bryston design). The link has all but maybe not the earliest prints.
You right I was in hurry and did not save my last changes so reverse the 2 diodes. The voltage was picked for the best Vbe of the 2 VAS driver transistors.
Just a academic exercise to model a complete amp and see what it has or not. I know in the day that the Bryston was a much better choice than the Tigers but I was broke college grad at the time.
Zeners are, roughly, like regular diodes with a breakdown voltage equal to the zener voltage.
So, if you reverse them they will forward conduct along with the driver transistors and rob current
from them. You could put a regular diode in series to conduct in the zener direction but be open
in the zeners forward direction.
But again, I thought I tried this, I might have jumped to the conclusion that they would add
distortion due to the nonlinearity forgetting that they conduct on the side that is turned off
so should have little if any impact on distortion. Just thinking out loud, so more thought or
testing is required.
So, if you reverse them they will forward conduct along with the driver transistors and rob current
from them. You could put a regular diode in series to conduct in the zener direction but be open
in the zeners forward direction.
But again, I thought I tried this, I might have jumped to the conclusion that they would add
distortion due to the nonlinearity forgetting that they conduct on the side that is turned off
so should have little if any impact on distortion. Just thinking out loud, so more thought or
testing is required.
Last edited:
Bryston 3B III complete SPICE Simulation with more mods.
Opps... Pete you are correct I forgot to add the diodes in LTSpice. I was doing all my modeling in PSpice and trying to look at the distortion. The zeners may need to be knock down a notch for the added voltage drop. I do not use LTSpice that much and have not set up a big lib of parts. Let me know if you see anything else. Thanks for your input. Fred
Opps... Pete you are correct I forgot to add the diodes in LTSpice. I was doing all my modeling in PSpice and trying to look at the distortion. The zeners may need to be knock down a notch for the added voltage drop. I do not use LTSpice that much and have not set up a big lib of parts. Let me know if you see anything else. Thanks for your input. Fred
Thanks for your input Fred, this is looking like a workable solution but why knock
down the zener when we can just use a regular silicon diode and get well below
the usual rated reverse Vbe? Maybe two diode drops.
Perhaps, like me you were thinking of minimal disturbance to the circuit. I'd
probably go with 2 to 4 volt zeners if I choose that route.
The thing with the Tigers was that the gain setting resistors are 100 ohms and the
diode current would be very high, still it seems to work so perhaps fast diodes with
a decent current rating.
I also kick the input voltage up to 6V while testing this.
down the zener when we can just use a regular silicon diode and get well below
the usual rated reverse Vbe? Maybe two diode drops.
Perhaps, like me you were thinking of minimal disturbance to the circuit. I'd
probably go with 2 to 4 volt zeners if I choose that route.
The thing with the Tigers was that the gain setting resistors are 100 ohms and the
diode current would be very high, still it seems to work so perhaps fast diodes with
a decent current rating.
I also kick the input voltage up to 6V while testing this.
Here is my current reasoning, the two resistors that provide the reverse voltage are one
to ground and one to the very low impedance output therefore clamping should not cause
and increase in distortion in that regard. But the clamp current on the other side of the
diode has to be driven by the VAS and that probably _will_ increase distortion. But the
reverse current is high when in clipping where distortion is already high so perhaps it
doesn't matter.
to ground and one to the very low impedance output therefore clamping should not cause
and increase in distortion in that regard. But the clamp current on the other side of the
diode has to be driven by the VAS and that probably _will_ increase distortion. But the
reverse current is high when in clipping where distortion is already high so perhaps it
doesn't matter.
Bryston 3B III complete SPICE Simulation with more mods.
Well you nailed it. That was my line of thinking more of a clamp using fast diodes with "minimal disturbance to the circuit". I think the Vbe of the bottom transistor is 5 volts so a 1 volt drop plus 3-4 would allow the amp to it's limit before kicking in to save the drivers. Nothing like beeting this old amp to death. I repaired a few "Flaming Tigers" in the day. I can still smell the smoke. Thanks again Pete
Well you nailed it. That was my line of thinking more of a clamp using fast diodes with "minimal disturbance to the circuit". I think the Vbe of the bottom transistor is 5 volts so a 1 volt drop plus 3-4 would allow the amp to it's limit before kicking in to save the drivers. Nothing like beeting this old amp to death. I repaired a few "Flaming Tigers" in the day. I can still smell the smoke. Thanks again Pete
Bryston 3B III complete SPICE Simulation with more mods.
I saw that and assumed it was a model issue. Maybe I will pull out some of PSpice models and use the correct parts that was in the real amps. I gave up all my Tigers and burnt parts at a hamfest many years ago. I never found a burnt Bryston they may not had a real life problem. Even the prices of the old 3Bs today is up there. I always want one to try. For now I just look at it in spice. Oh well back to building tube models for the next amp project. Let me know how that works if you do Mod/Tame the Tiger. Thanks, Fred.
I saw that and assumed it was a model issue. Maybe I will pull out some of PSpice models and use the correct parts that was in the real amps. I gave up all my Tigers and burnt parts at a hamfest many years ago. I never found a burnt Bryston they may not had a real life problem. Even the prices of the old 3Bs today is up there. I always want one to try. For now I just look at it in spice. Oh well back to building tube models for the next amp project. Let me know how that works if you do Mod/Tame the Tiger. Thanks, Fred.
Bryston 3B III complete SPICE Simulation with more mods.
Just FYI, good for a laugh;
Only if we had spice back then. There would have been less head scratching, no solder burns, and maybe no "Genie Smoke". And a LM3876 (Old chip) is a (Tiger .01) on a chip but with a THD+N 0.06% and IMD 0.004% at full output.
Fred
Just FYI, good for a laugh;
Only if we had spice back then. There would have been less head scratching, no solder burns, and maybe no "Genie Smoke". And a LM3876 (Old chip) is a (Tiger .01) on a chip but with a THD+N 0.06% and IMD 0.004% at full output.
Fred
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.