Hello,
I had repaired an old Trace Elliot GP7 SM300 (300W bass amp from a friend). Some transistors defective and a break on a pcb trace, which will have been the cause of it's first death. During the test on a 15" (4 Ohm) i cranked up the volume and at a loud level all the transistors went into meltdown again. After some research, the design basically does not seem to be so super destruction-proof. I assume the repair attempt was actually successful, it was just too much somewhere during testing.
The question: can the circuit be made a little more damage-proof with a manageable amount of effort?
There was already a modification (by someone else). The red resistor was soldered on the bottom side of the PCB.
The service manual is available here, for example: https://elektrotanya.com/trace-elliot_gp7_150-300watt_sm.pdf/download.html if necessary.
The toroidal transformer is primarily fused with T3.15A and there is approx. +-70V after the rectifier.
In my test at the moment when I was turning louder, the following transistors broke: TR 11 and on the +V side: 13,14,16 and on the -V side:
5,6,7,15 and the transformer fuse.
I had previously checked the DC offset, and a 1kHz sine wave looked clean on scope (does this bring more stability to reduce the quiescent current,
and e.g. rather have 0.5% THD instead of 0.01%?)
I had repaired an old Trace Elliot GP7 SM300 (300W bass amp from a friend). Some transistors defective and a break on a pcb trace, which will have been the cause of it's first death. During the test on a 15" (4 Ohm) i cranked up the volume and at a loud level all the transistors went into meltdown again. After some research, the design basically does not seem to be so super destruction-proof. I assume the repair attempt was actually successful, it was just too much somewhere during testing.
The question: can the circuit be made a little more damage-proof with a manageable amount of effort?
There was already a modification (by someone else). The red resistor was soldered on the bottom side of the PCB.
The service manual is available here, for example: https://elektrotanya.com/trace-elliot_gp7_150-300watt_sm.pdf/download.html if necessary.
The toroidal transformer is primarily fused with T3.15A and there is approx. +-70V after the rectifier.
In my test at the moment when I was turning louder, the following transistors broke: TR 11 and on the +V side: 13,14,16 and on the -V side:
5,6,7,15 and the transformer fuse.
I had previously checked the DC offset, and a 1kHz sine wave looked clean on scope (does this bring more stability to reduce the quiescent current,
and e.g. rather have 0.5% THD instead of 0.01%?)
and take a wild guess TR 12 and 5 also dead.
which are current limiters.
and likely missed and were dead from first fail
and possible damaged diodes
dont get the point of the resistor mod.
as with any magical mod amps not working.
usually just remove the mod.
or if repairing all damage will be where magic mod
is connected.
whole output from Q11 big mystery.
RV2 for bias, if dirty or goes open.
amp will launch itself to hard high high
meltdown. they put on wrong side of VBE
pretty much dingle ball design.
This amp design was fail from start.
worst MI amp ever made.
Has current limiters but no fly back protect diodes
on output, no dc protect either.
So when amp die it also make speaker die
if DC on output. Cab fryer
easy fix, dont fix
which are current limiters.
and likely missed and were dead from first fail
and possible damaged diodes
dont get the point of the resistor mod.
as with any magical mod amps not working.
usually just remove the mod.
or if repairing all damage will be where magic mod
is connected.
whole output from Q11 big mystery.
RV2 for bias, if dirty or goes open.
amp will launch itself to hard high high
meltdown. they put on wrong side of VBE
pretty much dingle ball design.
This amp design was fail from start.
worst MI amp ever made.
Has current limiters but no fly back protect diodes
on output, no dc protect either.
So when amp die it also make speaker die
if DC on output. Cab fryer
easy fix, dont fix
Your resistor mod makes no sense.
You'll likely find all transistors to the right of the Vbe multiplier dead. This includes the two over-current protection transistors Q5 and Q12 and possibly their series diodes D8 and D13.
Once you get the amp working you could entertain adding the flyback diodes mentioned above. So add a beefy diode from the amp output terminal to V+ and another from the output to V-. They'll need to be reverse biased during normal operation.
I'm also not a fan of the thermal switch in series with the output. I'd look for a way to get the amp to power off when it gets too hot.
But get the amp working before you start modifying it - if you modify it.
Tom
You'll likely find all transistors to the right of the Vbe multiplier dead. This includes the two over-current protection transistors Q5 and Q12 and possibly their series diodes D8 and D13.
Once you get the amp working you could entertain adding the flyback diodes mentioned above. So add a beefy diode from the amp output terminal to V+ and another from the output to V-. They'll need to be reverse biased during normal operation.
I'm also not a fan of the thermal switch in series with the output. I'd look for a way to get the amp to power off when it gets too hot.
But get the amp working before you start modifying it - if you modify it.
Oh? What's mysterious about it? It looks like a pretty standard emitter follower output. Adding the driver current to the emitter resistors for the output devices is a bit curious. Then there's a bit of current feedback from the speaker connection. That's not too uncommon.
Valid point. If the bias pot fails the output stage will be biased to the maximum bias current and likely melt down.
Tom
The mystery is not the topology.
Should be more complete sentence.
Big mystery why it blew up.
was making a point about " mods"
and repairing amps with mods.
you just look where the mods go.
and there will be a problem
or where it starts
there is no mystery, was being comical
+/-70 volts and 4 ohm speaker through only two pairs of outputs will result in meltdown within short order, if you are playing anything other than quiet background music. The way the current limiters are set, it will allow 16 amps peak at full Vout, and 6 amps reactive. Around 400W peak dissipation per device which it can only do in the kHZ region. Bass will blow it. The speaker will likely try to demand more, causing them to activate all the time. That’s not a reliable way to run an amplifier even WITH flyback diodes. Without, it’s even more certain to die than without the limiters. You also may be blowing drivers if you are indeed using TIPs, as the real breakdown on those is right at 140 volts, give or take a few so you might not be so lucky. They are not a good device to use anyway, even at rated voltage or less, and should be replaced with MJE15030/1.
My guess is this thing was intended to run off +/-55V, max. Which would be reliable.
My guess is this thing was intended to run off +/-55V, max. Which would be reliable.
Hi there,
This is the infamous "bipolar bear" output stage, yes?
I believe Trace Elliot designed a new MOSFET output stage to replace this as it was incredibly unreliable. But I don't think that is available any more;; Trace Elliot has been through many ownership changes, and I also think the devices used became unobtainable.
There are, or were, third parties out there offering different replacement boards if you wanted to go that route?
Cheers, and regards,
Ant.
This is the infamous "bipolar bear" output stage, yes?
I believe Trace Elliot designed a new MOSFET output stage to replace this as it was incredibly unreliable. But I don't think that is available any more;; Trace Elliot has been through many ownership changes, and I also think the devices used became unobtainable.
There are, or were, third parties out there offering different replacement boards if you wanted to go that route?
Cheers, and regards,
Ant.
I check the data sheets and as I suspected, TIP31/32C are rated 100V, not 140V, so these must have been built with selected parts. And two pair of 2SC4468/2SA1695 are good for about 100W, not 300W. So this amp was hopelessly optimistic. I would suggest about 4 pair of 2sc5200/2sa1943 (+2 more sets of 0.22 RE) and replace the drivers with something like MJE15030/31.
As others have said, you need to sure the protection parts (etc) are OK, since they probably got blown when the amp failed.
As others have said, you need to sure the protection parts (etc) are OK, since they probably got blown when the amp failed.
Rated to 100, but will usually take 140. Not always so you can’t count on it. Others have tried, some failed.
Two more pairs of outputs, increase Re to 0.39. Then the protection circuit would actually be correct for typical audio transistors on 70V. The extra pairs helps across the board an 4 ohm wont just kill it, but if you leave the resistors at 0.22 things will still die if you short the output. If you do 4 ohms with them set to 0.39 (where it’s safe) the protection will operate all the time. If you want to do low Z loads, add another 2 more pairs (6) with 0.39 Re’s. THEN it gets reliable for sure.
Two more pairs of outputs, increase Re to 0.39. Then the protection circuit would actually be correct for typical audio transistors on 70V. The extra pairs helps across the board an 4 ohm wont just kill it, but if you leave the resistors at 0.22 things will still die if you short the output. If you do 4 ohms with them set to 0.39 (where it’s safe) the protection will operate all the time. If you want to do low Z loads, add another 2 more pairs (6) with 0.39 Re’s. THEN it gets reliable for sure.
Well, I guess that confirms what I thought…. it’s intended to run on about +/-55 and shouldn’t be run on more. Clamping the VAS hard without adding a current limit for the VAS itself is also another Bad Idea.
There are better VAS transistors available than 2N3440/5415 these days. Those or MJE340/50 were the go-to parts back in the heyday of the Leach amps. TIP31/2 even graded to 140V is a Bad Idea. A really Bad Idea - probably one of the worst.
There are better VAS transistors available than 2N3440/5415 these days. Those or MJE340/50 were the go-to parts back in the heyday of the Leach amps. TIP31/2 even graded to 140V is a Bad Idea. A really Bad Idea - probably one of the worst.
I design, build, eat and breathe Guitar/Bass amplifiers for over 50 years now.
Over 14000 amps by now, of which over 1000 were 300W into 4 ohm, with +/-65V rails.
My power supplies are stiff, thanks to very low DCR transformers, to boot kept cold (first thing in front of side mounted fan) so they don't drop after 1-2 hours full power.
Used 4 output devices with no problems at all, in gruelling World Cup Stadiums conditions, 8 hour long open air festivals, etc.
Which devices?: metallic TO3 2N3773 for decades, which are "slow" Hometaxial , way stronger than those fast "HiFi" Japanese transistors.
Switched to 4 IRFP 250, sometimes IRFP240 if unavailable; never a problem.
Not sure about HiFi transistors or 70V rails but can vouch for 4 "industrial" transistor 300W amps if properly used.
Back to this T.E. amp of which I repaired a few, design is not as bad as mentioned above but quite typical of MI amps and it does have regular VI/foldback short protection, it should stand reaching clipping, at least into a resistive load.
That it did not, stinks of fake transistors, period.
Good real ones should work, at least as intended.
Upgrading them to better specs/ stronger ones, of course shouldn't hurt, at all.
Using more too, but they won't fit on current heatsinks.
Plan B: if I had them here on my bench, I would mod them for IRFP250, but can't offer detailed step by step instructions by "remote control" , sorry.
Over 14000 amps by now, of which over 1000 were 300W into 4 ohm, with +/-65V rails.
My power supplies are stiff, thanks to very low DCR transformers, to boot kept cold (first thing in front of side mounted fan) so they don't drop after 1-2 hours full power.
Used 4 output devices with no problems at all, in gruelling World Cup Stadiums conditions, 8 hour long open air festivals, etc.
Which devices?: metallic TO3 2N3773 for decades, which are "slow" Hometaxial , way stronger than those fast "HiFi" Japanese transistors.
Switched to 4 IRFP 250, sometimes IRFP240 if unavailable; never a problem.
Not sure about HiFi transistors or 70V rails but can vouch for 4 "industrial" transistor 300W amps if properly used.
Back to this T.E. amp of which I repaired a few, design is not as bad as mentioned above but quite typical of MI amps and it does have regular VI/foldback short protection, it should stand reaching clipping, at least into a resistive load.
That it did not, stinks of fake transistors, period.
Good real ones should work, at least as intended.
Upgrading them to better specs/ stronger ones, of course shouldn't hurt, at all.
Using more too, but they won't fit on current heatsinks.
Plan B: if I had them here on my bench, I would mod them for IRFP250, but can't offer detailed step by step instructions by "remote control" , sorry.
as recommended I will replace the TIPs with MJE15030/1 and remove the red 4k7.
R26 and R15 are currently 27k (according to the circuit diagram 27k are in the 300W version and 10k in the 150W version), then perhaps something like 22k would be an idea to add some safety?
is adding 5x20mm fuses like 4AT at +V and -V behind the electrolytic capacitors or one at the speaker out any good idea?
so that driving it to hard just melts the fuse and not all the transistors.
R26 and R15 are currently 27k (according to the circuit diagram 27k are in the 300W version and 10k in the 150W version), then perhaps something like 22k would be an idea to add some safety?
is adding 5x20mm fuses like 4AT at +V and -V behind the electrolytic capacitors or one at the speaker out any good idea?
so that driving it to hard just melts the fuse and not all the transistors.
This is a fascinating discussion.
From the datasheet, the 2SC4468 looks weak (100W and 10A maximum).
By comparison, the MJ15003/4 is 250W and 20A maximum. I use four to deliver half as much power. The amplifier lives an easy life in my living room.
The design margins are dramatically different.
Ed
From the datasheet, the 2SC4468 looks weak (100W and 10A maximum).
By comparison, the MJ15003/4 is 250W and 20A maximum. I use four to deliver half as much power. The amplifier lives an easy life in my living room.
The design margins are dramatically different.
Ed
Wrong way. You would increase the resistor value to reduce the amount of current before limiting sets in. The 300 and 150 watt versions would run on different supply voltages and that’s why 10k was used in the smaller version. If you used 10k with that much rail voltage you may as well not be using current limiters.
Maybe change those woosy output devices with MJL21193/4 or MJL4281/4302. That’s as big as youre going to go these days. You will want the better drivers if you use 2119x. Adding more is the only real fix and it’s probably not practical.
MJE340/50 are no better than the 3440/5415, eventually slower despite the official specs and with lower gain.
The VAS current is inherently limited since the LTP sink at most 6mA per branch, the voltage at the base of the VAS wont exceed 3.36V and its 100R emitter resistance will hence have at most 2.7V at its end, so that makes 27mA as peak VAS current.
That s a convenient way to protect the VAS in these kind of schematic, of course at the expense of the amp linearity, to render negligible the influence of this resistance require to load the VAS with a triple EF.
The +-70 v rails is a huge problem.
It is possible there is a 100 vac primary transformer tap for Japanese market, and this unit is set to that. Then changing to the 120 vac tap would pull rails down to something reasonable like +-50 v.
If not you can preregulate the rails with four more transistors and another heatsink. I have used two parallel TIP147 with a 72 v zener stack driving the base to regulate my 75 v rail down to 68. Use a resistor series the zener stack to keep the current under control, limited by zener wattage. I used 1.2 watt zeners. Requires a 1000 uf cap before the regulator and I used 3300 uf afterwards for two 75w channels. You want to regulate to more +-55 v unless a major upgrade of heatsink & output transistors is done.
A different transformer would be ideal because it would fit in the case, but do not know if your friend would spend that much.
It is possible there is a 100 vac primary transformer tap for Japanese market, and this unit is set to that. Then changing to the 120 vac tap would pull rails down to something reasonable like +-50 v.
If not you can preregulate the rails with four more transistors and another heatsink. I have used two parallel TIP147 with a 72 v zener stack driving the base to regulate my 75 v rail down to 68. Use a resistor series the zener stack to keep the current under control, limited by zener wattage. I used 1.2 watt zeners. Requires a 1000 uf cap before the regulator and I used 3300 uf afterwards for two 75w channels. You want to regulate to more +-55 v unless a major upgrade of heatsink & output transistors is done.
A different transformer would be ideal because it would fit in the case, but do not know if your friend would spend that much.
Last edited: