2n5551/2n5401 reliability experience?

[vintagelove]

Hello and thanks for looking. I wanted to get the opinions of some folks who have had a lot of experience repairing amps.

Long story short, I have a Tubeworks amp, that uses this combination to drive the output transistors. The thing is I came across a post from a well respected gentleman who suggested these parts were not up to the task and unreliable in this position.

What is your experience with amps using this transistor pair? Do you think it's a reliable part in a 40-60w power amp? Schematic is attached below. Thanks for your time and knowledge. Have a great day.

https://ibb.co/nbd79vf

[g1]

The thing is I came across a post from a well respected gentleman who suggested these parts were not up to the task and unreliable in this position.

That will have been JM Fahey over at MEF.  He will probably be around shortly and maybe will expand on the issue. 

[Loudthud]

The circuit looks strange, I suspect it may contain mistakes. I can see where one transformer winding powers the 12.6V tube heater and goes through a Voltage Doubler and could produce roughly +/- 15V. What is strange is the other winding has no center tap shown but drives two 4700uF 50V caps in series with the center grounded.

Another strange thing. See the output transistors. Looks pretty normal, NPNs on top, PNPs on bottom. Now look at the transistors just to the left. See how the NPNs are driven by an NPN transistor, but it's emitter is connected to the C+ rail. Same with the driver for the PNPs, a PNP transistor with it's emitter connected to the C- rail. Is current supposed to flow backwards through those transistors ?

[J M Fahey]

Yes, I agree with those being mistakes.

Not dissing anybody, but in my book B K Butler is a "pedal guy" and very successful at that.
Power amp guy? ... not that much.

At least some wild choices were made, plus the obvious errors.

1) yes, no CT on main power winding is weird.

There are a couple amps doing so, Revox A40 comes to mind, but they require* that current consumption on +V and -V rails match each other, at least on average, or rail voltages float around wildly.
Not a good choice.

Works with sinewaves which are symmetrical by definition, not so much with Guitar signals.

2) the multi function 12V winding is normal, in pedal supplies, specially on Tubedriver pedals.
Wall wart is not typical +9V DC but 12VAC, so it gives you 12VAC for filaments, and +/-16V (raw) which can feed Op Amps.
But filtering shown is poor, I bet on ripple on those rails.
I´d upgrade A caps from 470uF to 2200uF or at least 1000uV (25V) and B ones to 1000uF each, 25V would be safer than present 16V types

3) tube cathodes are not grounded 
Of course they must be, but schematic drawing is a mess.

4) power amp drawing is a mess, its NFB does not make sense at all.

5) the first (left) 2N5401/5551 pair is fine for its duty, voltage amplification and little current, 1 or 2 mA expected.

Then right pair is a mess, they are NOT suitable to drive a power transistor pair.
Puny TO92 case transistors can not dissipate whats needed.

I would use TO126 case transistors there, such as BD139/140 or equivalent US types (MJE340/350 or better?).

Pay attention to pinout.

What is the actual +/-C rail voltage?

Can you post  picture or two showing amp guts and a power transistor area closeup?

Including the driver transistors.

Lots of poor/fake schematics around.

Sometimes never rechecked, sometimes with errors on purpose (Mesa Boogie), sometimes not actually Factory originals but "reverse engineered" by Musicians.

Oh well.

[vintagelove]

Yes, I agree with those being mistakes.

Not dissing anybody, but in my book B K Butler is a "pedal guy" and very successful at that.
Power amp guy? ... not that much.

At least some wild choices were made, plus the obvious errors.

1) yes, no CT on main power winding is weird.

There are a couple amps doing so, Revox A40 comes to mind, but they require* that current consumption on +V and -V rails match each other, at least on average, or rail voltages float around wildly.
Not a good choice.

Works with sinewaves which are symmetrical by definition, not so much with Guitar signals.

2) the multi function 12V winding is normal, in pedal supplies, specially on Tubedriver pedals.
Wall wart is not typical +9V DC but 12VAC, so it gives you 12VAC for filaments, and +/-16V (raw) which can feed Op Amps.
But filtering shown is poor, I bet on ripple on those rails.
I´d upgrade A caps from 470uF to 2200uF or at least 1000uV (25V) and B ones to 1000uF each, 25V would be safer than present 16V types

3) tube cathodes are not grounded 
Of course they must be, but schematic drawing is a mess.

4) power amp drawing is a mess, its NFB does not make sense at all.

5) the first (left) 2N5401/5551 pair is fine for its duty, voltage amplification and little current, 1 or 2 mA expected.

Then right pair is a mess, they are NOT suitable to drive a power transistor pair.
Puny TO92 case transistors can not dissipate whats needed.

I would use TO126 case transistors there, such as BD139/140 or equivalent US types (MJE340/350 or better?).

Pay attention to pinout.

What is the actual +/-C rail voltage?

Can you post  picture or two showing amp guts and a power transistor area closeup?

Including the driver transistors.

Lots of poor/fake schematics around.

Sometimes never rechecked, sometimes with errors on purpose (Mesa Boogie), sometimes not actually Factory originals but "reverse engineered" by Musicians.

Oh well.

Hello, here are some pics of the area in question. I didn't have it out on the bench, so no chance to grab the voltages.

Are any of those parts you mentioned a drop in replacement?

So your main concern is if they are able to dissipate the heat needed?

Would a sufficient test be if I played a loop pedal for an hour through it at the desired volume (you would likely be thrown out of the establishment if you turned this thing above 5, it's LOUD), and monitored the temperature of that pair?

Let me state flatly, I really appreciate you sharing your thoughts on this, and you are far more knowledgeable about the subject than I am. The only reason I question if a test would show it as acceptable to my needs are you see a ton of these still floating around after all these years, and you never hear about problems with them. The only posts I've ever seen were the final output transistors going (and one of those accounts said it was a long story, which I take to mean he did something silly).


Anyway, let me know what you think. Thanks again


Having trouble attaching the photos from my phone, give me a few minutes and I'll have them up

[Loudthud]

The transistors on the layout have Q numbers, but not on the schematic.

The driver transistors can be spun around swapping Collector and Emitter because of the single inline footprint on the PCB. That makes a lot more sense.

It would be nice to know the C+ and C- Voltages.

[vintagelove]

The transistors on the layout have Q numbers, but not on the schematic.

The driver transistors can be spun around swapping Collector and Emitter because of the single inline footprint on the PCB. That makes a lot more sense.

It would be nice to know the C+ and C- Voltages.

Hello, not my measurement unfortunately, but from the original thread the reliability comment emerged from,

" For the voltage rails I disconnected the transformer secondary wires (thought the transformer might have been blown) but am getting 35V RMS on each leg (with center tap), which should get rails of 48ish V"

[g1]

Other tubeworks amps that used Fet's for outputs had the 2N5401/5551 drivers.
Probably a lot less drive current required for Fet gates.
I wonder if they might have just stuck with them for this model because "we're already doing it that way"?

examples attached.

[g1]

@vintagelove , as you mentioned, it seems they are getting away with using the TO92 devices for drivers in this case.  But this is the only amp I think I have ever seen doing so.  In my experience anything using bi-polar output devices has something heftier than TO-92 for drivers. 
Also, a lot of times the drivers are replaced when the outputs are blown, even if they check ok, as they will have been stressed.  It's possible this has been done on some of the 'blown output' repairs without being reported.

[vintagelove]

@vintagelove , as you mentioned, it seems they are getting away with using the TO92 devices for drivers in this case.  But this is the only amp I think I have ever seen doing so.  In my experience anything using bi-polar output devices has something heftier than TO-92 for drivers. 
Also, a lot of times the drivers are replaced when the outputs are blown, even if they check ok, as they will have been stressed.  It's possible this has been done on some of the 'blown output' repairs without being reported.

Is there anything you can think of that would drop in and be a bit more robust?

Do you think the stress test above would be sufficient to determine if they're reliable to do what I need?

Would you agree the main problem would be its ability to dissipate heat? And if so, if after that hour of constant playing at the desired volume the parts remain relatively cool, I would be able to rely on them?


Thanks so much.

[J M Fahey]

In no particular order and agreeing with my own "Law"  :

1) "do not mess with PCBs unless FORCED to", derived from the Truth: "you can buy *any* component for an Amp ... except the PCB", I suggest, "leave as is" for now.

2) marginal, but "if they work, they work", so ....

3) FULLY agree with G1: TO92 transistors are FINE for MosFet outputs, because MosFets draw NO drive current.

To be more precise, power MosFets have significant Gate capacitance, this is not 100% accurate but for analysis imagine they have, say, 2200 to 4700 pF in parallel to gates, which have to be charged/discharged on AC signals, but current involved is around 1 mA, easy peasy.

4) while those *bipolar* TIP31/32 will take Base current, big time.

Let´s do some "back of an envelope" Math:

47V rail/peak voltage into possible 5.5 speaker minimum impedance (DCR) amounts to worst case but possible 8.5A peaks.
Say 8A discounting transistor voltage drop but not much less than that.

So 4 A peak per device.

BUT TIP31/32 are THREE Ampere transistors 
Two of them would handle 6A peaks ... still short by a wide margin.

They also dissipate 40W each (under ideal conditions) so both "could" dissipate 80W

Now a classic similar power stage would use a pair of 2N3055 transistors, or modern equivalents.
Which are 15 Ampere transistors rated 115W dissipation and would be "just enough" there.
Now you know why this particular amp design sounds crazy to any Tech.

5) also: from datasheet, TIP31/32 current gain can drop as low as 10x at 3A (4A is not even mentioned, it is outside the curves) so drive current could be as high as 8A/10=0.8A which is stressing (bordering on impossible) for 2N5401/5551 which was our original concern.

But now we find more and more wild choices.

Oh well, if it works, works, leave as is.

IF you have any trouble, come back and we will help you solve it.

6) just as a side comment: both schematics posted by G1 look WAY more reasonable, not World´s clearest layout or drawing but no glaring errors either.
Tube cathodes are grounded, or better said, go to a connector pin which *might* be Ground or even -15V or so (have seen it in other amps) to increase tube working voltages.
Might even go to -69V ; not too easy to navigate around that schematic.

7) that schematic and the kind of errors it shows *looks* like those made by non-Techs trying to reverse Engineer (i.e. "copy") a Schematic by looking at a populated PCB.

Often 90/95% accurate, but always some errors creep in.

A Tech looks at them and says "no, this can´t be possible" (unconnected cathodes, wrong polarity and orientation drivers, impossible NFB connection, no PT center tap, etc.) and rechecks; a non Tech just draws what he thinks he saw.

Main transformer winding IS center tapped (in G1´s schematics).