2N3055 failure (power supply)

[Ben Jackson]

I tracked down the fault in a failed power supply to a failed 2N3055
power transistor. It's conducting from C->B in a gang with 4 other
transistors driven by a TIP31. The net result was that the failure
caused the supply to jump to 22V+, far higher than its typical max
voltage, and with no current limiting. Are there any obvious mods
I should consider while I'm fixing it?

___
|
TIP31 | ___
LM723 ___ |/ |
[Vout>-|___|-o-----| |
22R | |> o---o---o---o---o
| | | | | | |
| ___ | | | | | |
'-|___|-o |/ |/ |/ |/ |/ 2N3055
2k2 '-----|---|---|---|---|
|> |> |> |> |>
| | | | |
.-. .-. .-. .-. .-.
| | | | | | | | | |
0.1R | | | | | | | | | |
5W '-' '-' '-' '-' '-'
| | | | |
o---o---o---o---o---->>>

(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

 

[Phil Allison]

"Ben Jackson"

I tracked down the fault in a failed power supply to a failed 2N3055
power transistor. It's conducting from C->B in a gang with 4 other
transistors driven by a TIP31. The net result was that the failure
caused the supply to jump to 22V+, far higher than its typical max
voltage, and with no current limiting. Are there any obvious mods
I should consider while I'm fixing it?

** Yes - change all those 0.1 ohm resistors to 0.33 ohms.

Also, use all *new* 2N3055s from the same batch to effect the repair -
or, if funds allow, use all new MJ15003s.

Otherwise device current sharing will likely be haywire.

BTW

Love the old Natsemi LM723 reg IC - versatile as all hell.




........ Phil

 

[John Popelish]

I tracked down the fault in a failed power supply to a failed 2N3055
power transistor. It's conducting from C->B in a gang with 4 other
transistors driven by a TIP31. The net result was that the failure
caused the supply to jump to 22V+, far higher than its typical max
voltage, and with no current limiting. Are there any obvious mods
I should consider while I'm fixing it?

___
|
TIP31 | ___
LM723 ___ |/ |
[Vout>-|___|-o-----| |
22R | |> o---o---o---o---o
| | | | | | |
| ___ | | | | | |
'-|___|-o |/ |/ |/ |/ |/ 2N3055
2k2 '-----|---|---|---|---|
|> |> |> |> |>
| | | | |
.-. .-. .-. .-. .-.
| | | | | | | | | |
0.1R | | | | | | | | | |
5W '-' '-' '-' '-' '-'
| | | | |
o---o---o---o---o---->>>

(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

Not having any resistance between the common base node and
the output bothers me. This gives the gang of 3055s a slow
turn off and a voltage overshoot when a .load is step reduced.

Something between 10 and 100 ohms might improve this.

 

[Brian]

I tracked down the fault in a failed power supply to a failed 2N3055
power transistor. It's conducting from C->B in a gang with 4 other
transistors driven by a TIP31. The net result was that the failure
caused the supply to jump to 22V+, far higher than its typical max
voltage, and with no current limiting. Are there any obvious mods
I should consider while I'm fixing it?

___
|
TIP31 | ___
LM723 ___ |/ |
[Vout>-|___|-o-----| |
22R | |> o---o---o---o---o
| | | | | | |
| ___ | | | | | |
'-|___|-o |/ |/ |/ |/ |/ 2N3055
2k2 '-----|---|---|---|---|
|> |> |> |> |>
| | | | |
.-. .-. .-. .-. .-.
| | | | | | | | | |
0.1R | | | | | | | | | |
5W '-' '-' '-' '-' '-'
| | | | |
o---o---o---o---o---->>>

(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

The 0.1R resistors are there to promote current sharing between the five
2N3055 transistors. If the circuits that the power supply is connected to,
are not real sensitive to over-voltage then you might replace the 0.1R
resistors with fusable resistors. If you do that, then you might want to put
a resistor and LED circuit across each fusable resistor (to let you know
when one is blown). If you need more protection than that, then you might
want to put a voltage sensing circuit on the output and a fuse in the
collector circuit. Have the voltage sense circuit trigger a SCR, that blows
that fuse in the collector circuit.

Brian Ellis

 

[DaveM]

I tracked down the fault in a failed power supply to a failed 2N3055
power transistor. It's conducting from C->B in a gang with 4 other
transistors driven by a TIP31. The net result was that the failure
caused the supply to jump to 22V+, far higher than its typical max
voltage, and with no current limiting. Are there any obvious mods
I should consider while I'm fixing it?

___
|
TIP31 | ___
LM723 ___ |/ |
[Vout>-|___|-o-----| |
22R | |> o---o---o---o---o
| | | | | | |
| ___ | | | | | |
'-|___|-o |/ |/ |/ |/ |/ 2N3055
2k2 '-----|---|---|---|---|
|> |> |> |> |>
| | | | |
.-. .-. .-. .-. .-.
| | | | | | | | | |
0.1R | | | | | | | | | |
5W '-' '-' '-' '-' '-'
| | | | |
o---o---o---o---o---->>>

(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

In addition to the other suggestions, I recommend that you consult the 723
datasheet and investigate the foldback current limiting suggestions given there.
It's well worth the trouble if there's a possibility of a short on the output.
Could save your power supply and an expensive repair job on connected equipment.

Cheers!!!!
--
Dave M
MasonDG44 at comcast dot net (Just substitute the appropriate characters in the
address)

Some days you're the dog, some days the hydrant.

 

[Jamie]

I tracked down the fault in a failed power supply to a failed 2N3055
power transistor. It's conducting from C->B in a gang with 4 other
transistors driven by a TIP31. The net result was that the failure
caused the supply to jump to 22V+, far higher than its typical max
voltage, and with no current limiting. Are there any obvious mods
I should consider while I'm fixing it?

___
|
TIP31 | ___
LM723 ___ |/ |
[Vout>-|___|-o-----| |
22R | |> o---o---o---o---o
| | | | | | |
| ___ | | | | | |
'-|___|-o |/ |/ |/ |/ |/ 2N3055
2k2 '-----|---|---|---|---|
|> |> |> |> |>
| | | | |
.-. .-. .-. .-. .-.
| | | | | | | | | |
0.1R | | | | | | | | | |
5W '-' '-' '-' '-' '-'
| | | | |
o---o---o---o---o---->>>

(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

put in a SCR crowbar circuit to clamp and pop the fuse when it over
voltages?

 

[Phil Allison]

"Brian"

Have the voltage sense circuit trigger a SCR, that blows that fuse in the
collector circuit.

** For the benefit of the OP - the simplest crowbar circuit would consist
of a 1 watt zener diode, a 56 ohm resistor and a 10 to 15 amp rated SCR (
TO220 pack).

The zener conduction voltage is chosen and tested to be the same as the
supply's nominal output voltage (ie 13.8 volts ?).

Connect the zener from the supply output to the G (gate) terminal of the
SCR, the 56 ohm goes from G to K (cathode ) while the A (anode ) terminal
connects also to the output positive rail. The K terminal also goes to the
output common or minus terminal.

The ( 20 amp ? ) fuse may be in the unregulated DC supply or low voltage AC
feed to the rectifier.




........ Phil

 

[[email protected]]

I tracked down the fault in a failed power supply to a failed 2N3055
power transistor. It's conducting from C->B in a gang with 4 other
transistors driven by a TIP31. The net result was that the failure
caused the supply to jump to 22V+, far higher than its typical max
voltage, and with no current limiting. Are there any obvious mods
I should consider while I'm fixing it?

___
|
TIP31 | ___
LM723 ___ |/ |
[Vout>-|___|-o-----| |
22R | |> o---o---o---o---o
| | | | | | |
| ___ | | | | | |
'-|___|-o |/ |/ |/ |/ |/ 2N3055
2k2 '-----|---|---|---|---|
|> |> |> |> |>
| | | | |
.-. .-. .-. .-. .-.
| | | | | | | | | |
0.1R | | | | | | | | | |
5W '-' '-' '-' '-' '-'
| | | | |
o---o---o---o---o---->>>

(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

I cant read this circuit very well, it appears to be an emmiter
follower. If thats the case then you have 2 basic problems. Firstly
your output is not inside the regulation loop so your regulation will
be poor. Secondly you have no short circuit protection, or indeed any
current control. Both these problems are easily solved by looking at
the application notes for the 723.

 

[Phil Allison]

<[email protected]

I cant read this circuit very well, it appears to be an emmiter
follower. If thats the case then you have 2 basic problems. Firstly
your output is not inside the regulation loop so your regulation will
be poor. Secondly you have no short circuit protection, or indeed any
current control.

** Nonsense.

The OP supplied a partial schem and STATED that it was a current limited
supply.

Both these problems are easily solved by looking at
the application notes for the 723.

** The OP very likely has a schem like figure 4 here:

http://www.national.com/ds/LM/LM723.pdf

Rsc is probably one of the 0.1 ohm ballast resistors shown in his schem,
resulting in a 6.5 amp limit per 3055 device.




........ Phil

 

[Ben Jackson]

Thanks all for the responses. I'll reply to all of them here:

Not having any resistance between the common base node and
the output bothers me. This gives the gang of 3055s a slow
turn off and a voltage overshoot when a .load is step reduced.

Something between 10 and 100 ohms might improve this.

The partial schematic I drew only covered the off-pcb parts of the circuit
that I had traced. There is a 330R from the common base to the output

** Yes - change all those 0.1 ohm resistors to 0.33 ohms.

Good idea, but I'm not sure this supply (BK Precision 1688) has room
for the 10W resistors I'd need. It was definitely not built for
maintenance. The final assembly consisted of pushing the PCB onto
the lugs of all the front panel stuff (outputs, meters) and soldering
it in place!

Also, use all *new* 2N3055s from the same batch to effect the repair -
or, if funds allow, use all new MJ15003s.

Yes, I'm hoping that a new batch of 2N3055s will fix any current sharing
problems. Digikey was out of MJ15003 and a few others I tried.

In addition to the other suggestions, I recommend that you consult
the 723 datasheet and investigate the foldback current limiting
suggestions given there.

The supply does include current limiting, but the nature of the failure
really took the 723 out of the equation. That's one of the reasons I
was looking for mod suggestions, which brings me to:

Jaime wrote and Phil Allison expanded:

** For the benefit of the OP - the simplest crowbar circuit would consist
of a 1 watt zener diode, a 56 ohm resistor and a 10 to 15 amp rated SCR (
TO220 pack).

The zener conduction voltage is chosen and tested to be the same as the
supply's nominal output voltage (ie 13.8 volts ?).

That seems like the best plan. The max in-spec voltage is 14.5V (I think
this was meant for automotive applications).

Thanks all.

 

[Brian]

<[email protected]

That's the way I read it too.

** Nonsense.

The OP supplied a partial schem and STATED that it was a current limited
supply.

The Op stated: "The net result was that the failure caused the supply to
jump to 22V+, far higher than its typical max voltage, and with NO CURRENT
LIMITING.

** The OP very likely has a schem like figure 4 here:

http://www.national.com/ds/LM/LM723.pdf

Rsc is probably one of the 0.1 ohm ballast resistors shown in his schem,
resulting in a 6.5 amp limit per 3055 device.




....... Phil

Unless you are seeing something that I'm not seeing, I don't see how you can
think that. In the partial schematic that the OP showed, the 0.1 ohm
resistors are being used to make sure the power transistors in parallel, are
sharing the current (close to equally).

Brian Ellis

 

[Phil Allison]

"Ben Jackson"

Good idea, but I'm not sure this supply (BK Precision 1688) has room
for the 10W resistors I'd need.

** How many amps will this supply deliver ?

The reason for the suggested change, BTW, is that 0.1 ohm ballast resistors
only work well with carefully Vbe matched devices.

If your new ones are all suitably Vbe matched, then use 0.1ohms.


....... Phil

 

[Brian]

That's the way I read it too.


The Op stated: "The net result was that the failure caused the supply to
jump to 22V+, far higher than its typical max voltage, and with NO CURRENT
LIMITING.


Unless you are seeing something that I'm not seeing, I don't see how you
can think that. In the partial schematic that the OP showed, the 0.1 ohm
resistors are being used to make sure the power transistors in parallel,
are sharing the current (close to equally).

Brian Ellis

When the output power transistor shorts out, there is no current (or
voltage) control.

Brian Ellis

 

[Phil Allison]

"Brian"

"Phil Allison"

The Op stated: "The net result was that the failure caused the supply to
jump to 22V+, far higher than its typical max voltage, and with NO CURRENT
LIMITING.

** Indicating that the current limiting inherent in the supply was rendered
ineffective by the faulty 3055.

The OP has now confirmed this IS in fact correct.

Unless you are seeing something that I'm not seeing, I don't see how you
can think that. In the partial schematic that the OP showed, the 0.1 ohm
resistors are being used to make sure the power transistors in parallel,
are sharing the current (close to equally).

** Look at the schem again and see what is blindingly obvious.

Any one of the 0.1 ohm ballast resistors can be used as Rsc.

Then the max output current = 5 times the current limit set by that
resistor.




......... Phil

 

[Phil Allison]

"Brian"

When the output power transistor shorts out, there is no current (or
voltage) control.

** Never in dispute.

Your earlier false assertions were.




........ Phil

 

[[email protected]]

<[email protected]



** Nonsense.

The OP supplied a partial schem and STATED that it was a current limited
supply.

Well i guess Phil, one of us cant read.

** The OP very likely has a schem like figure 4 here:


http://www.national.com/ds/LM/LM723.pdf

Maybe but since he didnt state it its just speculation on your part.

Rsc is probably one of the 0.1 ohm ballast resistors shown in his schem,
resulting in a 6.5 amp limit per 3055 device.

That would be too bad if it was since the 2n3055 wont survive a short
at that current for long.

 

[Brian]

"Brian"


** Indicating that the current limiting inherent in the supply was
rendered ineffective by the faulty 3055.

The OP has now confirmed this IS in fact correct.

You act like he finally came to that conclusion. He stated that from the
very beginning.

Even if that "should" be the case, so what, it has nothing to do with his
problem

** Look at the schem again and see what is blindingly obvious.

Of course it is.

Any one of the 0.1 ohm ballast resistors can be used as Rsc.

The operative word here is "can". But is it? It doesn't matter, it has
nothing to do with his problem.

 

[Ben Jackson]

"Ben Jackson"

** How many amps will this supply deliver ?

It's a 20A supply, so 4A per 2N3055, the .1R should dissipate about 1.6W
max and the .33 would be a little over 5W so I'd need the bigger package.
I would also have to trace the rest of the circuit and see what else
might be affected by the change (there's an LM339 looking at at least
temperature and maybe other things that will shut down the 723).

The reason for the suggested change, BTW, is that 0.1 ohm ballast resistors
only work well with carefully Vbe matched devices.

If your new ones are all suitably Vbe matched, then use 0.1ohms.

I ordered 10 so maybe I'll measure them and choose the best 5.

 

[Phil Allison]

"Brian"

You act like he finally came to that conclusion.

** You have a serious reading comprehension disability.

What the OP has confirmed is the supply IS indeed current limited.

Even if that "should" be the case, so what, it has nothing to do with his
problem.

** You have a serious reading comprehension disability.

Plus an utterly ASD fucked brain.

Go top yourself anytime - ASSHOLE.





......... Phil

 

[Phil Allison]

Well i guess Phil, one of us cant read.

** If you were not such a complete ASSHOLE might feel sorry for you.

Maybe but since he didnt state it its just speculation on your part.

** An educated speculation that has been now confirmed as correct.

Enjoying your sour grapes ????

That would be too bad if it was since the 2n3055 wont survive a short
at that current for long.

** Its within the DC safe area curve for a 2N3055.

A temp cut out switch on the heatsink would prevent failure.

PSU design is obviously way over your pointy head - ASSHOLE.




......... Phil