Looking to replace Germanium power transistors with Silicon

[John Robertson]

I figure this could work for upgrading old jukebox amplifiers, type AB
or B class, from the mid 70s that ran on Germanium transistors - almost
unobtanium these days. And never that reliable to begin with.

Is it possible to use MosFets to replace them? There is a bit of info on
line about replacing G with S, but I don't think I've seen anyone talk
about Mos...

The original amp runs on -46VDC to ground and is roughly 25W RMS into a
16R load - has an output transformer for 70V operation as well as taps
for various speaker volumes. I suspect I'd have to replace the entire
power amp side as mixing G with Mos is probably just not going to work.

Suggestions welcome!

Thanks!

John :-#)#
--
(Please post followups or tech inquiries to the USENET newsgroup)
John's Jukes Ltd.
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3
(604)872-5757 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."

 

[whit3rd]

On Monday, April 13, 2020 at 10:36:55 PM UTC-7, John Robertson wrote:

I figure this could work for upgrading old jukebox amplifiers, type AB
or B class, from the mid 70s that ran on Germanium transistors - almost
unobtanium these days. And never that reliable to begin with.

The usual AB bias for audio amplifiers requires diodes that match the transistors'
thermal character, SO either you have to buy oddball (sometimes
called 'stabistor' ) biasing diodes that connect to the transistor heatsinks,
OR you buy a module, prebuilt and mounted for heatsinking.

Something like this module
<https://www.st.com/content/ccc/resource/technical/document/datasheet/4f/18/a6/c8/21/33/41/8d/CD00001887.pdf/files/CD00001887.pdf/jcr:content/translations/en.CD00001887.pdf>

will save you time and money over trying to do a replacement at the component level.

 

[Jeff Urban]

It is not practical to use a FET. The bipolar transistor is driven by current, the FET by voltage. The drive circuit is different because of that.

To swing the output all the way to the rails with an enhancement mode FET requires voltage higher than the rails. At least I think so, because being germanium it is 99% that they are both PNP.

As such they are probably TO-3 from what you described. I that case I think you would do best with like MJ15023s. Decent gain, fairly linear and with not alot of voltage requirement for bias.

As I service vintage audio I am getting good at bias. The last unit for example does not have an adjustment, I changed a resistor value from 680 to 729 to make it come out right.

So go in with the mindset you are going to have to increase the bias VOLTAGE to get to the proper CURRENT. In some cases you only have a stack of diodes to produce the drop, in some cases you got a little transistor biased on, or almost all the way on and they use its thermal characteristics to match the outputs. Either way you probably know enough to make it drop a little more voltage. And I would venture to say that the power transformer in the unit doesn't have the balls behind it to blow MJ15023s.

To use a FET you have to give it voltage drive and that might mean adding a stage. The original drive stage is meant to use current, like the transistors.

Now there is a possibility that the outputs are transformer driven. Yup, they had split winding driver transformers for the SESAPP circuit. I you got that it is practically impossible to modify. If it was built when NPN transistors were scarce that was the way to do it. Without both polarities of bipolars it is a real bitch to design an SESAPP output circuit without a transformer. A few manufacturers did it but it was a fucking mess. Soy did it I know, and it was a fucking mess but so was half of everything they did. Yep, and I was the one who had to figure this shit out.

If you think they had it easy, try designing a power output circuit using transistors - but of only one polarity. A bitch. You have to depend on the specs of all the transistors or else it will be flooded with even order harmonic distortion.

On the other hand I think I got a set of four 2N5436s. Those are almost unblowable at lower voltages. Real germanium and TO-3 but with 60 mil pis. Good to 60 amps. If you blow those you is a bad mofo.

Just use the damn 15023s. they will do it and all you have to do is mess with the bias, the rest will all work.

Don't make problems for yourself. We have plenty of people to do that.

 

[John Robertson]

On 2020/04/14 1:22 p.m., Jeff Urban wrote:

It is not practical to use a FET. The bipolar transistor is driven by current, the FET by voltage. The drive circuit is different because of that.

To swing the output all the way to the rails with an enhancement mode FET requires voltage higher than the rails. At least I think so, because being germanium it is 99% that they are both PNP.

As such they are probably TO-3 from what you described. I that case I think you would do best with like MJ15023s. Decent gain, fairly linear and with not alot of voltage requirement for bias.

As I service vintage audio I am getting good at bias. The last unit for example does not have an adjustment, I changed a resistor value from 680 to 729 to make it come out right.

So go in with the mindset you are going to have to increase the bias VOLTAGE to get to the proper CURRENT. In some cases you only have a stack of diodes to produce the drop, in some cases you got a little transistor biased on, or almost all the way on and they use its thermal characteristics to match the outputs. Either way you probably know enough to make it drop a little more voltage. And I would venture to say that the power transformer in the unit doesn't have the balls behind it to blow MJ15023s.

To use a FET you have to give it voltage drive and that might mean adding a stage. The original drive stage is meant to use current, like the transistors.

Now there is a possibility that the outputs are transformer driven. Yup, they had split winding driver transformers for the SESAPP circuit. I you got that it is practically impossible to modify. If it was built when NPN transistors were scarce that was the way to do it. Without both polarities of bipolars it is a real bitch to design an SESAPP output circuit without a transformer. A few manufacturers did it but it was a fucking mess. Soy did it I know, and it was a fucking mess but so was half of everything they did. Yep, and I was the one who had to figure this shit out.

If you think they had it easy, try designing a power output circuit using transistors - but of only one polarity. A bitch. You have to depend on the specs of all the transistors or else it will be flooded with even order harmonic distortion.

On the other hand I think I got a set of four 2N5436s. Those are almost unblowable at lower voltages. Real germanium and TO-3 but with 60 mil pis. Good to 60 amps. If you blow those you is a bad mofo.

Just use the damn 15023s. they will do it and all you have to do is mess with the bias, the rest will all work.

Don't make problems for yourself. We have plenty of people to do that.

Hi Jeff,

Yeah, I figured FET was not practical.

I did see one page on line where the discussion about replacing G with S
was you double the size of the original G's emitter resistor. So, if it
was 75R you would change that to 150R. Have you any knowledge of that?

As for the Motorola MJ15023, they sound great, and are PNP which are
what I need. Vceo 200, IcContinuous 16A, 15/60hfe, 5MhZ, and 200W
dissipation. That will be hard to blow up!

I may go with a second diode to aid the temp stabilization, but I am no
expert on designing transistor amps, just a dedicated plodder.

Schematic here:

ftp://www.flippers.com/pdfs/Seeburg_TSA1_Amp_Schematic_Update.pdf

Thanks,

John :-#)#

--
(Please post followups or tech inquiries to the USENET newsgroup)
John's Jukes Ltd.
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3
(604)872-5757 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."

 

[Phil Allison]

John Robertson wrote:

------------------------

Yeah, I figured FET was not practical.

As for the Motorola MJ15023, they sound great, and are PNP which are
what I need. Vceo 200, IcContinuous 16A, 15/60hfe, 5MhZ, and 200W
dissipation. That will be hard to blow up!

** No hard at all.

I once had boxes of dead ones lying around - mostly extracted from amplifiers like the Phase linear 400 and 700 mk2s.

Like any bi-polar, they had limited "safe area" of about 2A at 80V for 0.5 seconds and much less at higher voltages.

Plus excessive chip temp is always fatal.

The lateral Mosfets made by Hitachi that come along in the 1980s were enormously more rugged.


...... Phil

 

On Tuesday, 14 April 2020 06:36:55 UTC+1, John Robertson wrote:

I figure this could work for upgrading old jukebox amplifiers, type AB
or B class, from the mid 70s that ran on Germanium transistors - almost
unobtanium these days. And never that reliable to begin with.

Is it possible to use MosFets to replace them? There is a bit of info on
line about replacing G with S, but I don't think I've seen anyone talk
about Mos...

The original amp runs on -46VDC to ground and is roughly 25W RMS into a
16R load - has an output transformer for 70V operation as well as taps
for various speaker volumes. I suspect I'd have to replace the entire
power amp side as mixing G with Mos is probably just not going to work.

Suggestions welcome!

Thanks!

John :-#)#

The differences between Ge & Si go further than just bias voltage. You could try it but don't get too excited, some amps simply won't fly with Si.

Russia still puts out lots of geraniums, don't know what power devices they have but I expect you'd find something usable. Prices are good. In extremis you could probably parallel a bank of small trannies with sharing resistors.

FET require much larger input voltage swings.


NT

 

[John Robertson]

On 2020/04/14 4:38 p.m., Phil Allison wrote:

John Robertson wrote:

------------------------

Yeah, I figured FET was not practical.

As for the Motorola MJ15023, they sound great, and are PNP which are
what I need. Vceo 200, IcContinuous 16A, 15/60hfe, 5MhZ, and 200W
dissipation. That will be hard to blow up!


** No hard at all.

I once had boxes of dead ones lying around - mostly extracted from amplifiers like the Phase linear 400 and 700 mk2s.

Like any bi-polar, they had limited "safe area" of about 2A at 80V for 0.5 seconds and much less at higher voltages.

Plus excessive chip temp is always fatal.

The lateral Mosfets made by Hitachi that come along in the 1980s were enormously more rugged.


...... Phil

My amp runs at -46V and about 1.5A before the line fuse fails. So being
derated to just under 50V should handle the limits you mention. If you
have better suggestions I am most open to them.

My goal is to have a reliable and relatively simple upgrade that folks
who wish to keep their jukeboxes mostly authentic can work with. Most of
the amp is stable after new electrolytics are installed, it is just the
outputs and their respective drivers that tend to fail due to, well, life...

Nothing to stop the owners from putting a modern amplifier in the
machine and just ignore the original tech, we do that when asked.

Thanks!

John :-#)#

--
(Please post followups or tech inquiries to the USENET newsgroup)
John's Jukes Ltd.
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3
(604)872-5757 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."

 

[Phil Allison]

John Robertson wrote:
----------------------

** No hard at all.

I once had boxes of dead ones lying around - mostly extracted from amplifiers like the Phase linear 400 and 700 mk2s.

Like any bi-polar, they had limited "safe area" of about 2A at 80V for 0.5 seconds and much less at higher voltages.

Plus excessive chip temp is always fatal.

The lateral Mosfets made by Hitachi that come along in the 1980s were enormously more rugged.



My amp runs at -46V and about 1.5A before the line fuse fails. So being
derated to just under 50V should handle the limits you mention.

** No doubt, long as nothing goes wrong.

The gross failures I saw were caused, not inherent.

Those causes however were predictable.

1. Overheating due to lack of good fan cooling.

2. Overloading with too many speakers.

3. Gradual failure of one or more speakers leading to a low impedance load that causes chip temp to exceed limits.

4. The fact they are *free standing* amplifiers, not incorporated in a unit like a juke box, so exposed to external influences the amp maker's never dreamed of.

IF appears to you that MJ15023s will do the needed job, by all means use them.

I probably would.


..... Phil

 

[Jeff Urban]

>I did see one page on line where the discussion about replacing G >with S was you double the size of the original G's emitter resistor. >So, if it was 75R you would change that to 150R. Have you any >knowledge of that?

This is not making sense here. What is G and what is S ? From what I heard G was gate, the equivalent of the base in a transistor and S was source equivalent to the emitter of a transistor, the C or D of course it the other one. Collector or drain.

There is no just changing S for G in any reasonable circuit. If you somehow wanted to reconfigure it to have the current drive drivers run the sources instead of the gates good luck. It'll take a ton of time to do the math to do something that probably will not work. Drivers cannot handle much current, that is why they are not outputs.

The schematic link does not work, path not found or some shit.

Being a Seeberg I would bet it uses driver transformers. That means to use silicon in it you change two resistors per channel, or perhaps add them to make the new value. Depends on where they fit.

Get the print to somewhere accessible so I can have a look at it.

 

[Fox's Mercantile]

On 4/16/20 4:26 AM, Jeff Urban wrote:

This is not making sense here. What is G and what is S ? > From what I heard G was gate, the equivalent of the base
in a transistor and S was source equivalent to the emitter
of a transistor, the C or D of course it the other one.
Collector or drain.

Oh for fuck's sake. Try and keep up will ya?
G = Germanium and S = Silicon.

--
"I am a river to my people."
Jeff-1.0
WA6FWi
http:foxsmercantile.com

 

[Jeff Urban]

That will be hard to blow up!


** No hard at all.

Well harder than an MJE2955. Harder than a 2N66609. They make oe really tougher pair, MJ2XXXX something. Probaably use those in Krells, just because. I mean they got like 128 output transistors ... nevermind.

>I once had boxes of dead ones lying around - mostly extracted from >amplifiers like the Phase linear 400 and 700 mk2s.

So those were all the complementary ones,not the quasi like I had. It was a two but still all NPN outputs. And of course when I got it and took it apart I found like four different types or putouts in it. Shit, this is only a 400 so it only has three pairs per channel.

>Like any bi-polar, they had limited "safe area" of about 2A at 80V >for 0.5 seconds and much less at higher voltages

Well it's not quite THAT bad, but... People do not get it sometimes. "It will be hard to blow that". sure, but if you get it hot enough it will fry. All transistors derate power handling. By 200ÂşC they can dissipate zero watts. Hot enough those badass MJs will fry, perhaps even from bias current. Not even one watt.

The way I take on it it that the thermal resistance to case is what limits that. The die can get so get hot so fast that it doesn't matter how good the heatsink is, it is not moving the heat FAST enough. That 200ÂşC is the JUNCTION temperature. And even with the best of mounting techniques, I mean run a bead or blob and squish it out so there are no air pockets, do it on clean surfaces and make it tight as hell, the thermal resistance is still atrocious.

One of the things I do understand some is power. I make a point of when I fix an amp it stays fixed. So I WILL use MJ15024s ans 15023s in fifty watt per channel amps.

Yes the power rating derates like an other, but this one you have to put in a deep fryer. LOL (not really) But then 375ÂşF is only 190ÂşC so we got ten degrees to go !

>The lateral Mosfets made by Hitachi that come along in the 1980s >were enormously more rugged.

If you got spares to spare I'll take them. But I don't remember right now, were they the ones with the really low Vgs or the ones you can parallel with no source resistors ?

And you know source resistors don't work worth a fuck on MOSFETs. I got a circuit that can balance a bank of the much better but, why ? Many people could do the same thing.

Ahh, I probably don't have to worry about this right now. A guy is bringing a Tascam, bodybuilder model four track Âź" four track.

Nice to do something different once in a while.

ACK ! I just looked at the time and I missed my shootemup show of the day. Dammit. Fucking Aussies.

 

[jfeng@my-deja.com]

On Thursday, April 16, 2020 at 2:26:36 AM UTC-7, Jeff Urban wrote:

The schematic link does not work, path not found or some shit.
It weere a t7po errorr; Try using http instead of fttp

 

[John Robertson]

On 2020/04/16 2:26 a.m., Jeff Urban wrote:

I did see one page on line where the discussion about replacing G >with S was you double the size of the original G's emitter resistor. >So, if it was 75R you would change that to 150R. Have you any >knowledge of that?

This is not making sense here. What is G and what is S ? From what I heard G was gate, the equivalent of the base in a transistor and S was source equivalent to the emitter of a transistor, the C or D of course it the other one. Collector or drain.

My apologies - I was shortening Germanium Transistor to "G" and Silicon
Transistor to "S". Considering the context of the discussion I thought
it was obvious. Sorry about that!

John :-#)#

There is no just changing S for G in any reasonable circuit. If you somehow wanted to reconfigure it to have the current drive drivers run the sources instead of the gates good luck. It'll take a ton of time to do the math to do something that probably will not work. Drivers cannot handle much current, that is why they are not outputs.

The schematic link does not work, path not found or some shit.

Being a Seeberg I would bet it uses driver transformers. That means to use silicon in it you change two resistors per channel, or perhaps add them to make the new value. Depends on where they fit.

Get the print to somewhere accessible so I can have a look at it.

--
(Please post followups or tech inquiries to the USENET newsgroup)
John's Jukes Ltd.
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3
(604)872-5757 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."

 

[John Robertson]

On 2020/04/16 6:46 a.m., jfeng@my-deja.com wrote:

On Thursday, April 16, 2020 at 2:26:36 AM UTC-7, Jeff Urban wrote:
The schematic link does not work, path not found or some shit.
It weere a t7po errorr; Try using http instead of fttp

Indeed, my mistake (yet again):

https://www.flippers.com/pdfs/Seeburg_TSA1_Amp_Schematic_Update.pdf

John ;-#)#

--
(Please post followups or tech inquiries to the USENET newsgroup)
John's Jukes Ltd.
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3
(604)872-5757 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."

 

[Fox's Mercantile]

On 4/16/20 11:50 AM, John Robertson wrote:

My apologies - I was shortening Germanium Transistor to
"G" and Silicon Transistor to "S". Considering the context
of the discussion I thought it was obvious. Sorry about that!

John :-#)#

The only thing obvious to Jeff is his own inflated self worth.


--
"I am a river to my people."
Jeff-1.0
WA6FWi
http:foxsmercantile.com

 

[Jeff Urban]

>Indeed, my mistake

Well thank you very much.

Looks like your best bet is to just put some silicon (diode) in series with CR107. I would recommend a pot for R127 so you can trim the bias. Forget specs just make it the same as the other channel.

And that is 47 volts supply ? That means 23.5 split. I wouldn't even bother with the MJ15023s, it is not worth it. The MJ2955 will handle that application no sweat. Just give it bias. Newark has then for like a buck and a quarter.

Of course check the drove circuitry first. To do that put 100 ohm resistors across the B and E of the outputs. The circuit should stabilise ad actually amplify but DO NOT put a load on it. Once it does that THEN you put the outputs in. Always on a DBY first.

 

[John Robertson]

On 2020/04/18 1:00 a.m., Jeff Urban wrote:

Indeed, my mistake

Well thank you very much.

Looks like your best bet is to just put some silicon (diode) in series with CR107. I would recommend a pot for R127 so you can trim the bias. Forget specs just make it the same as the other channel.

There was a 75R pot added in later amps of the same series with R127
reduced slightly in value. I am considering that.

Any particular silicon diodes?

And that is 47 volts supply ? That means 23.5 split. I wouldn't even bother with the MJ15023s, it is not worth it. The MJ2955 will handle that application no sweat. Just give it bias. Newark has then for like a buck and a quarter.

I have some MJ2955s in stock.

Of course check the drove circuitry first. To do that put 100 ohm resistors across the B and E of the outputs. The circuit should stabilise ad actually amplify but DO NOT put a load on it. Once it does that THEN you put the outputs in. Always on a DBY first.

I'll put up the schematic of the model I am looking to upgrade to with
my notes...past this model they changed the specs on the other
pre-driver germanium transistors which would require a bit more changes
to get the bias correct I suspect.

What is a "DBY"? Normally I use a series 40 to 60W lamp with the line
when doing test to current limit.


John :-#)#

--
(Please post followups or tech inquiries to the USENET newsgroup)
John's Jukes Ltd.
MOVED to #7 - 3979 Marine Way, Burnaby, BC, Canada V5J 5E3
(604)872-5757 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."

 

[Fox's Mercantile]

On 4/21/20 1:27 PM, John Robertson wrote:

What is a "DBY"? Normally I use a series 40 to 60W lamp with the line
when doing test to current limit.

Jeff is hard pressed to spell anything.
DBY = DBT, Dim Bulb Tester.

--
"I am a river to my people."
Jeff-1.0
WA6FWi
http:foxsmercantile.com

 

[whit3rd]

On Tuesday, April 21, 2020 at 11:27:45 AM UTC-7, John Robertson wrote:

On 2020/04/18 1:00 a.m., Jeff Urban wrote:

Looks like your best bet is to just put some silicon (diode) in series with CR107. I would recommend a pot for R127 so you can trim the bias. Forget specs just make it the same as the other channel.

There was a 75R pot added in later amps of the same series with R127
reduced slightly in value. I am considering that.

Any particular silicon diodes?

To be careful, diode-test a good Ge transistor and its associated diode, and
try to get the same voltage-offset (transistor BE drop minus diode drop). Within 20 mV
would be good.

As I read the schematics, the pulldown output drive transistor has an emitter resistor
and base-negative pulldown resistor; you might want to change the base pulldown
resistor, too, to bias the emitter current the same as the original.