Hey Bear, well the idea of dominant pole compensation is to have one and only one stage with the "Miller" cap, best for it to be the stage with the most voltage gain. If I had to use dominant pole compensation, which I view as the sledgehammer approach to solving the problem, I'd also remove or reduce C8 which I believe further hurts stability.
I believe that they pushed the limits of this design too much, no "miller" comp, wide bandwidth, high feedback to lower distortion even at high frequencies. I think it was rated at over 100W into 4 ohms and the rails were +/- 42 V idling. The safe operating area curve breaks from constant power of 200W at a Vce of 25V, to the secondary breakdown slope where they only handle 1 amp or 50W with a Vce of 50V. This is steady state pulsed secondary breakdown current is much higher and one could argue that it's safe but I believe the devices are over stressed with inductive loads. Really, load analysis should be done for the output devices. It ran VERY hot when driven hard. A thermal cutout might help.
The amp also does not have flyback diodes, the output device data sheet shows them internally however I don't believe it's good to rely on them, external devices are better.
I believe that the Tiger amps had more than one problem, the amp that lasted had 1W resistors for R12,14,15,16 and was rarely driven hard. It did fail one time just by being driven hard with 6 ohm loads before I went to 1W resistors.
I believe that they pushed the limits of this design too much, no "miller" comp, wide bandwidth, high feedback to lower distortion even at high frequencies. I think it was rated at over 100W into 4 ohms and the rails were +/- 42 V idling. The safe operating area curve breaks from constant power of 200W at a Vce of 25V, to the secondary breakdown slope where they only handle 1 amp or 50W with a Vce of 50V. This is steady state pulsed secondary breakdown current is much higher and one could argue that it's safe but I believe the devices are over stressed with inductive loads. Really, load analysis should be done for the output devices. It ran VERY hot when driven hard. A thermal cutout might help.
The amp also does not have flyback diodes, the output device data sheet shows them internally however I don't believe it's good to rely on them, external devices are better.
I believe that the Tiger amps had more than one problem, the amp that lasted had 1W resistors for R12,14,15,16 and was rarely driven hard. It did fail one time just by being driven hard with 6 ohm loads before I went to 1W resistors.
PB2 said:
Don't have the schematic out.
Think I started with the idea of compensating the stage with the most voltage gain... and rapidly went to the idea of reducing the speed of the amp by any and all means... That's going back a long time now and for sure I was shooting in the dark at that time.
Added the external diodes to the outputs, increased the size of
the resistors that blow up the most...
nada. Pfffft! <frying sounds> power level no the issue.
Clearly, at least I did not have the thing stablized...
If I was doing it now, I'd probably degenerate some stages substantially, maybe one by one, and reduce the feedback levels, and see if that calmed things down a bit, before selectively jacking up the gains and seeing what was getting squrrely. I do recall that I didn't *see* any signs of parasitics on a dummy load, but back then I didn't try to see what happens with a capacitive load pack...
_-_-bear
I've been lurking on this thread for a while but finally decided to look at my Tiger folder to see what all the fuss/smoke/love was all about.
I have Dan's original 1970 Popular Electronics article and wouldn't touch that schematic.
PB2 suggests using 1watt resistors in various places. Yep, this was suggested in 1972 in an Audio Amateur.
Later that year, Dan publishes 'Optimizing the Tiger 175-A' in Audio Amateur issue #3 This is much different than the original tiger (and hopefully more stable).
Digging deeper in my folder, I found another mod to help reliability (adding 100pF to collector/base to q5-6, 47pf to q3, update r17,r18...) in 1973. Yikes! More bandaids.
I don't know how old my pair of tiger 'B' (model 275) are (1976?) but they share similar output stages are earlier models but with (and I quote from the manual) 'we have added our complementary differential input circuit to this rugged old standby ..."
I haven't experienced the same issues listed here but have had one PS cap fail and one zener fail. I don't think this is too bad given '7' pages of comments on this old design. Maybe the model 275 finally is reliable.
I hope everyone is enjoying their tigers but when mine fail and I need a sand amp, I'll look at building a leach in my old Marantz 250 enclosure.
rick
I have Dan's original 1970 Popular Electronics article and wouldn't touch that schematic.
PB2 suggests using 1watt resistors in various places. Yep, this was suggested in 1972 in an Audio Amateur.
Later that year, Dan publishes 'Optimizing the Tiger 175-A' in Audio Amateur issue #3 This is much different than the original tiger (and hopefully more stable).
Digging deeper in my folder, I found another mod to help reliability (adding 100pF to collector/base to q5-6, 47pf to q3, update r17,r18...) in 1973. Yikes! More bandaids.
I don't know how old my pair of tiger 'B' (model 275) are (1976?) but they share similar output stages are earlier models but with (and I quote from the manual) 'we have added our complementary differential input circuit to this rugged old standby ..."
I haven't experienced the same issues listed here but have had one PS cap fail and one zener fail. I don't think this is too bad given '7' pages of comments on this old design. Maybe the model 275 finally is reliable.
I hope everyone is enjoying their tigers but when mine fail and I need a sand amp, I'll look at building a leach in my old Marantz 250 enclosure.
rick
djk wrote:
"To those of you that think the CF pair is unreliable, look at a Bryston 3B, 4B, etc. The original 3B looks just like a Tiger .01 with another pair of outputs in parallel. Bryston retroactively extended the waranty on these early models to 20 years, a safe bet as the ones I saw for service usually just need filter caps. They don't sound half bad either."
Thanks djk for the reference to this amp.
Has anyone taken a look at this schematic? What do you think?
"To those of you that think the CF pair is unreliable, look at a Bryston 3B, 4B, etc. The original 3B looks just like a Tiger .01 with another pair of outputs in parallel. Bryston retroactively extended the waranty on these early models to 20 years, a safe bet as the ones I saw for service usually just need filter caps. They don't sound half bad either."
Thanks djk for the reference to this amp.
Has anyone taken a look at this schematic? What do you think?
Electronics Experimenters Handbook 1971
has a different schematic, no complimentary output stage, and has output protection.
It's interesting that the ETI480, which isn't too different from the schem given in the first post, has a rep for stability. Perhaps it's worth considering upgrading unstable Tigers to ETI480s?
Pete McK
has a different schematic, no complimentary output stage, and has output protection.
It's interesting that the ETI480, which isn't too different from the schem given in the first post, has a rep for stability. Perhaps it's worth considering upgrading unstable Tigers to ETI480s?
Pete McK
The newer Bryston schematics on line are a bit different than the originals.
They still run the output stage with gain, but the driver drives the pair of outputs one EF and the other CF, at the same time.
The ETI 480 has been known to have some problems from time to time, most related to running too high a voltage on the 2N3055/2955 pair.
They still run the output stage with gain, but the driver drives the pair of outputs one EF and the other CF, at the same time.
The ETI 480 has been known to have some problems from time to time, most related to running too high a voltage on the 2N3055/2955 pair.
Similar designs - Tiger VAS degeneration
I've taken a look, and before I comment I'm wondering if anyone has looked at them with a "what's wrong with this picture eye".
The ETI is also very similar, but it does use a small "miller" (I actually prefer to call this cap Cdom) cap and nested HF feedback. Yet there are still reports of oscillations, can't be sure if they're reliable, but IIRC the designer did revise the HF compensation:
http://www.diyaudio.com/forums/showthread/t-15297.html
Regarding the Tiger schematic from the original post: I didn't notice that there was a lot of emmiter degeneration, or local feedback, at the VAS Q3 of 390 ohms which lowers the open loop gain and does help stability. Has anyone calculated or measured the open loop gain?
I've taken a look, and before I comment I'm wondering if anyone has looked at them with a "what's wrong with this picture eye".
The ETI is also very similar, but it does use a small "miller" (I actually prefer to call this cap Cdom) cap and nested HF feedback. Yet there are still reports of oscillations, can't be sure if they're reliable, but IIRC the designer did revise the HF compensation:
http://www.diyaudio.com/forums/showthread/t-15297.html
Regarding the Tiger schematic from the original post: I didn't notice that there was a lot of emmiter degeneration, or local feedback, at the VAS Q3 of 390 ohms which lowers the open loop gain and does help stability. Has anyone calculated or measured the open loop gain?
Undersized drivers
I like the fact that this original version of the Tiger has simply fuses in the +/- supply lines for output protection. However, I've always been suspicious that the driver transistors are also undersized. The design should be able to blow the fuses when driving a short circuit.
If we short the output, then the drivers see the parallel combination of R12 and R15 to ground for a combined value of 50 ohms. The collector of Q5 is just approximately one diode drop below the rail. Let's say for a 42 V supply there's 41 volts at the driver collector. If the amp is idling the rail will be at 42 and lets say the output was shorted. The volume is then turned up causing the drivers to be turn on, likely into clipping. Q5 saturates and the peak current is then 41/50 or .82 A, the drivers have an absolute maximum rating of .7 amps. The supply will of course sag but this cannot be good.
I like the fact that this original version of the Tiger has simply fuses in the +/- supply lines for output protection. However, I've always been suspicious that the driver transistors are also undersized. The design should be able to blow the fuses when driving a short circuit.
If we short the output, then the drivers see the parallel combination of R12 and R15 to ground for a combined value of 50 ohms. The collector of Q5 is just approximately one diode drop below the rail. Let's say for a 42 V supply there's 41 volts at the driver collector. If the amp is idling the rail will be at 42 and lets say the output was shorted. The volume is then turned up causing the drivers to be turn on, likely into clipping. Q5 saturates and the peak current is then 41/50 or .82 A, the drivers have an absolute maximum rating of .7 amps. The supply will of course sag but this cannot be good.
I used the MJE243/253 for drivers in a scratch built version of the large 200W SWTPC Tigersaurus amp back when I first suspected this problem. These are 4A/100V/15W but will only handle .1 A at Vce = 40V, or 4W due to SOA. They also have good beta at low Ic, and they're a good driver upgrade if ever rebuilding these amps. This large amp never had an oscillation problem. It had one output device fail, open as I recall, no smoke, and I believe it was just a defective part. But I never used it on the bench as a test amp, and I came up with my own grounding scheme.
The MJE15034/35 might also be a good choice, 4A/350V/50W which may seem like overkill, but their high Vce gives them better SOA characteristics. 1.25A at 40 V or 50W still on the constant power portion of the SOA curve. They also have much flatter beta characteristics being newer parts. Probably a good choice if one wants to experiment with high bias in the drivers of around a few watts.
Both of these parts require small heat sinks.
------------------------------------------
By the way, I've built the L'il Tiger SWTPC 22w kit from Popular Electronics (PE) December 1967:
http://www.swtpc.com/mholley/LittleTiger/PE_Index.htm
Built the SWTPC Universal Tiger stereo version kit from Popular Electronics October 1970, and repaired it probably more than 10 times.
Built a completely from scratch version of the HK Citation 12 amp around 1970.
Started a from scratch version of the large SWTPC Tigersaurus (200W) around 1973 but put it on hold and completed it around 1979, worked fine never any smoke.
Threw together a from scratch rats nest version of the Universal Tiger in 1976 that also worked fine, never any smoke.
-------------------------------------------
As far as the SWTPC amplifier history goes:
SWTPC L'il Tiger 22W, Popular Electronics, December, 1967
SWTPC Universal Tiger, 80W Popular Electronics, October 1970
SWTPC Plastic Tiger, 30W Popular Electronics, October 1971
This was the Universal Tiger with cheaper plastic output transistors.
"Optimizing the Tiger 175-A" Audio Amateur 3 - 1972
This article covered minor circuit changes to add output protection, and I believe minor bias changes, better diff amp transistors, and increased R12 and R14 to 2W, this was the Universal Tiger MKII version.
SWTPC Tiger .01 mono block, evolution of the Universal Tiger had a complementary diff amp and a triple output stage, rated at 60W into 8 ohms, less than .01% distortion.
SWTPC "Tigersaurus 250" 200W, Radio Electronics, December 1973
Series connected output devices, 8 per channel, fully complementary diff amp, gain of about 5 in driver/output stage, simple zener regulated front end.
The MJE15034/35 might also be a good choice, 4A/350V/50W which may seem like overkill, but their high Vce gives them better SOA characteristics. 1.25A at 40 V or 50W still on the constant power portion of the SOA curve. They also have much flatter beta characteristics being newer parts. Probably a good choice if one wants to experiment with high bias in the drivers of around a few watts.
Both of these parts require small heat sinks.
------------------------------------------
By the way, I've built the L'il Tiger SWTPC 22w kit from Popular Electronics (PE) December 1967:
http://www.swtpc.com/mholley/LittleTiger/PE_Index.htm
Built the SWTPC Universal Tiger stereo version kit from Popular Electronics October 1970, and repaired it probably more than 10 times.
Built a completely from scratch version of the HK Citation 12 amp around 1970.
Started a from scratch version of the large SWTPC Tigersaurus (200W) around 1973 but put it on hold and completed it around 1979, worked fine never any smoke.
Threw together a from scratch rats nest version of the Universal Tiger in 1976 that also worked fine, never any smoke.
-------------------------------------------
As far as the SWTPC amplifier history goes:
SWTPC L'il Tiger 22W, Popular Electronics, December, 1967
SWTPC Universal Tiger, 80W Popular Electronics, October 1970
SWTPC Plastic Tiger, 30W Popular Electronics, October 1971
This was the Universal Tiger with cheaper plastic output transistors.
"Optimizing the Tiger 175-A" Audio Amateur 3 - 1972
This article covered minor circuit changes to add output protection, and I believe minor bias changes, better diff amp transistors, and increased R12 and R14 to 2W, this was the Universal Tiger MKII version.
SWTPC Tiger .01 mono block, evolution of the Universal Tiger had a complementary diff amp and a triple output stage, rated at 60W into 8 ohms, less than .01% distortion.
SWTPC "Tigersaurus 250" 200W, Radio Electronics, December 1973
Series connected output devices, 8 per channel, fully complementary diff amp, gain of about 5 in driver/output stage, simple zener regulated front end.
Thank you Dimitri I missed those, the 4 channel amp was the Tiger .01 as you probably know, 4 of them physically located side by side. I don't remember the "Tigers that Roar" was that an early description of the Universal Tiger? It might have been the stereo version of the Universal Tiger with a shared power supply?
Spurious oscillations in transistors
I wanted to mention how it's possible for a single stage amplifier even an emitter follower to oscillate. Parasitic junction capacitances and lead inductance can form a Collpits oscillator. Here's a Colpitts:
http://www.ensc.sfu.ca/~ljilja/cnl/presentations/ljilja/iscas2002/sld016.htm
Any properly biased transistor will have lead or trace inductance in the collector load which provides L1 in the above circuit. C1 is the stray C to E capacitance, and Ce is the Miller capacitance as long as there's some capacitance from the base to ground. This probably explains how output stages oscillate but the base capacitance to ground is not clear or obvious, it could be the previous stages junction capacitance. We have to remember that the supply is also an AC ground to see how the Colpitts oscillator is formed.
It is somewhat risky to put a cap directly from the base to ground since this helps form the above circuit. The Tiger amp diff input has a 220 pF directly to ground, some series resistance here would help avoid oscillation in the diff amp. I don't know if this is a problem for sure in the Tigers but it's interesting that the Brystons do have this resistor.
I wanted to mention how it's possible for a single stage amplifier even an emitter follower to oscillate. Parasitic junction capacitances and lead inductance can form a Collpits oscillator. Here's a Colpitts:
http://www.ensc.sfu.ca/~ljilja/cnl/presentations/ljilja/iscas2002/sld016.htm
Any properly biased transistor will have lead or trace inductance in the collector load which provides L1 in the above circuit. C1 is the stray C to E capacitance, and Ce is the Miller capacitance as long as there's some capacitance from the base to ground. This probably explains how output stages oscillate but the base capacitance to ground is not clear or obvious, it could be the previous stages junction capacitance. We have to remember that the supply is also an AC ground to see how the Colpitts oscillator is formed.
It is somewhat risky to put a cap directly from the base to ground since this helps form the above circuit. The Tiger amp diff input has a 220 pF directly to ground, some series resistance here would help avoid oscillation in the diff amp. I don't know if this is a problem for sure in the Tigers but it's interesting that the Brystons do have this resistor.
Thermal Analysis - Self Papers
Does anyone know if these papers can be found online:
Self, D "Thermal dynamics of Power Amplifiers: I" Electronics World, May 1996
Self,D "Thermal dynamics of Power Amplifiers: II" Electronics World, June 1996
Self,D "Thermal dynamics of Power Amplifiers: II" Electronics World, Oct 1996
Does anyone know if these papers can be found online:
Self, D "Thermal dynamics of Power Amplifiers: I" Electronics World, May 1996
Self,D "Thermal dynamics of Power Amplifiers: II" Electronics World, June 1996
Self,D "Thermal dynamics of Power Amplifiers: II" Electronics World, Oct 1996
> thermal analysis of this driver and output stage? Self claims that the diodes should be in thermal contact with the drivers, which I believe stabilizes the driver idle current but not the output.
In the original, thermal compensation is mostly wishful, and the safest way to bias is with the outputs in Class B (fully off at idle).
In some Tiger designs, the output stage Vbe is inside the driver feedback loop. Output Vbe need not be compensated, but you do have to nail the driver thermal compensation. Meyer was at the cutting edge of big transistor amps, and bias compensation was still hit-n-miss.
> Does anyone know if these papers can be found online: Self, D ....
I bet all that material is in Doug's book, a useful reference.
In the original, thermal compensation is mostly wishful, and the safest way to bias is with the outputs in Class B (fully off at idle).
In some Tiger designs, the output stage Vbe is inside the driver feedback loop. Output Vbe need not be compensated, but you do have to nail the driver thermal compensation. Meyer was at the cutting edge of big transistor amps, and bias compensation was still hit-n-miss.
> Does anyone know if these papers can be found online: Self, D ....
I bet all that material is in Doug's book, a useful reference.