Trace Elliot Quad Chorus issue (schematic included!)

[VintageGear]

A couple of years back I bought a couple of chorusses. Amongst them was a Trace Elliot Quad Chorus which did not work. As there was no schematic available, I just put it in a box somewhere, to be fixed later. I recently found it again, and still no schematic online. So I just tried TE customer support, and guess what, it was provided without hesitation, I was even allowed to share it. Excellent!!!
Unfortunately, I still need help. The power supply was burned out and the circuit was shorting. No half measures, I did the following:
1) Replaced the 2N3096
2) Replaced all zener diodes
3) Replaced all elec capacitors
4) Replaced and socketed all IC's (except the 2x M50198P; already socketed)

The circuit came back to life and yes, a great lush chorus is audible!
However, I notice the transistor runs very hot (can't touch it) and starts to smell. That cannot be right!
I measured, V+ is 8.5 volt and V- -9.6 volt, and +5V = 5 volt. Even when I take out all the IC's from their sockets, the voltages are off. Or well, at least, I expected +9 and -9V.

Any suggestions how to find out what is causing this imbalance, or are the voltages supposed to be like this, and is something else causing the transistor to burn up?

Help is appreciated.

[anotherjim]

Do you have the original 18v power supply?

What happens if you pull R71 out? Disconnects the 5v obviously, but are the + and - supplies then better with a cool transistor?

[VintageGear]

Thanks for the reply. I do not have the original PSU, but am using a Dunlop 18v, should be good.
Indeed, the transistor is much cooler with the 5v disconnected, but the 8.45 volt remains on the +9v line.
On the +5V there's not much going on: IC3, 5 and 6, which I pulled, and still runs hot. Replaced C6 to be certain it was not a bad badge from recap. The 2N3904 I replaced earlier.

Reconnected R71. Still runs hot, again.

[anotherjim]

I think that the transistor is responsible for the difference in voltage balance due to the base-emitter junction voltage drop. It doesn't need exact balance. The signal is being processed in 5v devices so the audio can't use the full +/-9v swing anyway.
I think there's a chance that at some time in the past it's had the wrong supply polarity or even AC applied, so any semiconductor or polarized cap is suspect. There's also a chance that with any gear, there's a loose bit of solder or wire in there ever since it was made that hasn't caused trouble - until now!

The problem seems to be on the +5v. Something is taking too much current.

[ElectricDruid]

Why do you say the problem is on the 5V, Jim? I'm not seeing it.

If the transistor that's getting hot is TR3 the 2N3906 in the power supply, I'd suspect a short from the 0V rail to one of the other power rails. Maybe the +V rail since that's low. Possibly a solder blob across the C52/C53 electors you replaced, but it could be anywhere.

[Rob Strand]

Any suggestions how to find out what is causing this imbalance, or are the voltages supposed to be like this, and is something else causing the transistor to burn up?

Unless I've missed something I don't see how it *can't* burn up.

They are pulling a lot of current off the positive supply specifically the +5V rail and it's not being balanced-up with current draw on the negative rail.

The +5V rail has 9V on one side of the resistor R71 and 5V on the other.   That means there's (9-5.1) / 56 = 70mA going down R71.  Most of that current is ending up going through the transistor TR3.   TR3 has 9V across it.  So it's dissipating 70mA * 9V = 0.63W.   The transistor has thermal resistance of 200 degC/W  so it will rise 0.63 * 200 = 126 deg C!!!  If ambient is 25C that means the junction is cooking at 151 degC.

There's also another zener Z5 and R74 on the top right which means more current through TR3, about 40mA!  So another 0.36W.  That bumps the load on TR3 to 110mA and dissipation to 1W.

With a 2N3906, you probably don't want TR3 dissipating anymore than 0.35W.   Which means a maximum current of 0.35 / 9 = 40mA.

One way around it is to make TR3 a large package which will handle the power dissipation.   That's the safest option.

Another way is to put a 1W across the collector and emitter of to off-load power dissipation.   We want to off-load 110mA - 40mA = 70mA.   So R_divert =  9 / 70mA = 128 ohm.      There might be a small risk of glitches on power-up with this.  It's probably safe since the 5V stuff has its own 5V regulator.

To me it looks like a design issue.  Maybe see if you can find any service notes on-line.

Edit: another idea to off-load power dissipation is to put a resistor in series with the collector of TR3 that reduces the voltage drop across CE and hence reduces the power dissipation.

[anotherjim]

Wasn't the heating stopped when the feed to the 5v Zener was disconnected?

[ElectricDruid]

Wasn't the heating stopped when the feed to the 5v Zener was disconnected?

I understood "much cooler" to mean "better, but not fixed", but maybe it *was* fixed.

If Rob's right and it's a design fault (sounded pretty convincing to me!), I'd pull it all out and stick a 7805 in a TO220 package in it instead. At least that is *intended* to drop several volts across it and has a nice metal tab to help keep it cool. I never had any trouble running those from 9V down to 5V. Running a 5V supply off a 15V rail does tend to make the regulator get warm, but that's not much of a surprise!

[Rob Strand]

I'd pull it all out and stick a 7805 in a TO220 package in it instead

I pondered about that too.  The main problem is the current from the 5V circuits still needs to flow down through to the negative rail via the 2N3906.   The idea was a 5V series regulator like a 7805 wouldn't need to "waste" current like the zener shunt regulator.  The wasted current needs to flow down the 2N3906.   However when I looked at the BBD chip, the current draw was pretty high.  High enough that it's not going to make a difference.

A devious scheme to bypass current around the 2N3906 without adding extra parts to the PCB is this:   Reduce R11 and R22 from 10k  to  say 180 ohm.  That will bypass (9-2.7V)/180 = 35mA each giving us or target bypass current of 70mA .    The downside is the higher zener current will cause the zener voltage to rise and that may affect the circuit.    You would need to measure what the zener voltage is now then put in a lower voltage zener so the voltage ended up the same with the higher zener current.    The 10k's are such high values for a 2.7V zener you might end-up reducing the zeners down to 1.8V!

Shoving a bigger transistor in there and moving on with your life has a lot going for it.

[jonny.reckless]

The Quad Chorus was one of my designs. If I remember correctly, TR3 was upgraded via a design change order to a BD136 for this very reason. Dissipation in TR3 is not quite as bad a Rob suggests, due to some of the other current that would otherwise be consumed in the return path of TR3 actually being used to power the op amps and R72 etc., and it was on a small clip on heatsink in production, but it definitely ran too hot. I was young and Reckless back then

[Rob Strand]

and it was on a small clip on heatsink in production,

That's likely what's missing.  Probably enough to save it.

FWIW, I'm NOT bagging you out I'm just trying to work out what's going on.  I like your designs.  There's not many people left with the analog nack like yourself.

[stallik]

That didn't take long fellas. Neat thread

[Rob Strand]

FWIW, I'm NOT bagging you out I'm just trying to work out what's going on.  I like your designs.  There's not many people left with the analog nack like yourself.

I fixed my typo. It said the opposite  of what I meant to say 

[VintageGear]

This has become one neat thread indeed! Thanks a bunch to Jonny Reckless to clarify on this issue, how cool's that to hear the suspected minor design flaw confirmed!

Here are some comments on several hypothesis raised:

Wasn't the heating stopped when the feed to the 5v Zener was disconnected?

I understood "much cooler" to mean "better, but not fixed", but maybe it *was* fixed.

Indeed, the heating was better, but still ran to approx 40-50 degrees in a 10 seconds or so, not very pleasant to the touch. Does not sound reliable to me.

Possibly a solder blob across the C52/C53 electors you replaced, but it could be anywhere.

I did check for this, no shorts or messy solderwork.

and it was on a small clip on heatsink in production

That one is missing then. I thought about cutting up a piece metal into a T-shape, the smaller tap heatshrinked to the flat side of the transistor, and the chunkier piece touch the metal face plate with some thermal paste. But that felt a bit.. messy..

Shoving a bigger transistor in there and moving on with your life has a lot going for it.

I agree with Rob -- I ordered a couple of BD136 as suggested by Reckless; let's see if a 25-year-old memory solves the issue 
If not, I'll go on the resistor-change-route.

Will let you all know in a couple of weeks when my order arrives and I swapped the 2N3906 for the beefier BD136

[Rob Strand]

I agree with Rob -- I ordered a couple of BD136 as suggested by Reckless; let's see if a 25-year-old memory solves the issue 
If not, I'll go on the resistor-change-route.

The BD136 comes in at 100 degC/W so it should survive all but the hottest environments.   You could be kind and put a tiny heatsink on it, I'll leave that up to you.

[VintageGear]

Update: Yes, it is working with a BD136! Added a heatsink to it, and there was nice empty space on the PCB for it next to the DC plug and the R input jack.
Still it runs hot, but doesn't explode anymore!  Thanks for all the help.