Odd Transistor Choice

Gentlemen,

In the PSU section of that old Tek oscilloscope I've been
troubleshooting, I discovered several transistors had died. One of
these is puzzling me somewhat. It's a 2N5859 which they're using to
drive one of those big ol' TO-3 power transistors for the -15V output
section. I can't understand why they chose a fast switching device
with a GBP of 250Mhz. 

This being a linear PSU, in 1970 when these devices would have been
pretty expensive, it seems a curious choice. Fine, it can handle 80V
and pass 2 amps which is respectable in TO-39 in those days, but a
250Mhz Ft? What's the point? Some dumb old medium power audio
transistor could do that job, surely?


The schematic:
https://tinyurl.com/4m8ffw7e

Datasheet:
https://www.mouser.com/datasheet/2/68/2n5859-1368164.pdf

On Sunday, May 14, 2023 at 5:48:42 AM UTC-7, Cursitor Doom wrote:
> Gentlemen, 
> 
> In the PSU section of that old Tek oscilloscope I've been 
> troubleshooting, I discovered several transistors had died. One of 
> these is puzzling me somewhat. It's a 2N5859 which they're using to 
> drive one of those big ol' TO-3 power transistors for the -15V output 
> section. I can't understand why they chose a fast switching device 
> with a GBP of 250Mhz. 
> 
> This being a linear PSU, in 1970 when these devices would have been 
> pretty expensive, it seems a curious choice. Fine, it can handle 80V 
> and pass 2 amps which is respectable in TO-39 in those days, but a 
> 250Mhz Ft? What's the point? Some dumb old medium power audio 
> transistor could do that job, surely? 
> 
> 
> The schematic: 
> https://tinyurl.com/4m8ffw7e 
> 
> Datasheet: 
> https://www.mouser.com/datasheet/2/68/2n5859-1368164.pdf

Unless someone remember how much it costs in 1970s, we will never know why. However, when compares to the popular (and easy to remember) 2N2222, the 2A (vs. 0.8A) collector current seems attractive.  Both are 250MHz, so freq response is probably not an issue.

On 14/05/2023 1:46 pm, Cursitor Doom wrote:
> Gentlemen,
> 
> In the PSU section of that old Tek oscilloscope I've been
> troubleshooting, I discovered several transistors had died. One of
> these is puzzling me somewhat. It's a 2N5859 which they're using to
> drive one of those big ol' TO-3 power transistors for the -15V output
> section. I can't understand why they chose a fast switching device
> with a GBP of 250Mhz.
> 
> This being a linear PSU, in 1970 when these devices would have been
> pretty expensive, it seems a curious choice. Fine, it can handle 80V
> and pass 2 amps which is respectable in TO-39 in those days, but a
> 250Mhz Ft? What's the point? Some dumb old medium power audio
> transistor could do that job, surely?
> 
> 
> The schematic:
> https://tinyurl.com/4m8ffw7e
> 
> Datasheet:
> https://www.mouser.com/datasheet/2/68/2n5859-1368164.pdf

Ft 250Mhz was not particularly special even then, the Ft is not relevant 
to the choice there; they chose because the V/Ic/Hfe etc fitted the 
need. Even the 60 year old BC107-109 family you probably know well had 
Ft of circa 200Mhz.

I guess 2N2218A/2219A or BFY50 would do as equivalents to 2N5859. There 
are zillions likely to work well.

piglet

On Sunday, 14 May 2023 at 16:23:24 UTC+1, piglet wrote:
> On 14/05/2023 1:46 pm, Cursitor Doom wrote: 
> > Gentlemen, 
> > 
> > In the PSU section of that old Tek oscilloscope I've been 
> > troubleshooting, I discovered several transistors had died. One of 
> > these is puzzling me somewhat. It's a 2N5859 which they're using to 
> > drive one of those big ol' TO-3 power transistors for the -15V output 
> > section. I can't understand why they chose a fast switching device 
> > with a GBP of 250Mhz. 
> > 
> > This being a linear PSU, in 1970 when these devices would have been 
> > pretty expensive, it seems a curious choice. Fine, it can handle 80V 
> > and pass 2 amps which is respectable in TO-39 in those days, but a 
> > 250Mhz Ft? What's the point? Some dumb old medium power audio 
> > transistor could do that job, surely? 
> > 
> > 
> > The schematic: 
> > https://tinyurl.com/4m8ffw7e 
> > 
> > Datasheet: 
> > https://www.mouser.com/datasheet/2/68/2n5859-1368164.pdf
> Ft 250Mhz was not particularly special even then, the Ft is not relevant 
> to the choice there; they chose because the V/Ic/Hfe etc fitted the 
> need. Even the 60 year old BC107-109 family you probably know well had 
> Ft of circa 200Mhz. 
> 
> I guess 2N2218A/2219A or BFY50 would do as equivalents to 2N5859. There 
> are zillions likely to work well. 
> 
> piglet

They may have had commercial reasons to make that choice.  Perhaps they were
already buying millions to use in other products.  Maybe an alternative that might
have been a "better" choice would have added another part to the design.  In many
large companies there is a lot of pressure not to do that.
John

On 5/14/2023 10:27 AM, John Walliker wrote:
> On Sunday, 14 May 2023 at 16:23:24 UTC+1, piglet wrote:
>> On 14/05/2023 1:46 pm, Cursitor Doom wrote:
>>> Gentlemen,
>>>
>>> In the PSU section of that old Tek oscilloscope I've been
>>> troubleshooting, I discovered several transistors had died. One of
>>> these is puzzling me somewhat. It's a 2N5859 which they're using to
>>> drive one of those big ol' TO-3 power transistors for the -15V output
>>> section. I can't understand why they chose a fast switching device
>>> with a GBP of 250Mhz.
>>>
>>> This being a linear PSU, in 1970 when these devices would have been
>>> pretty expensive, it seems a curious choice. Fine, it can handle 80V
>>> and pass 2 amps which is respectable in TO-39 in those days, but a
>>> 250Mhz Ft? What's the point? Some dumb old medium power audio
>>> transistor could do that job, surely?
>>>
>>>
>>> The schematic:
>>> https://tinyurl.com/4m8ffw7e
>>>
>>> Datasheet:
>>> https://www.mouser.com/datasheet/2/68/2n5859-1368164.pdf
>> Ft 250Mhz was not particularly special even then, the Ft is not relevant
>> to the choice there; they chose because the V/Ic/Hfe etc fitted the
>> need. Even the 60 year old BC107-109 family you probably know well had
>> Ft of circa 200Mhz.
>>
>> I guess 2N2218A/2219A or BFY50 would do as equivalents to 2N5859. There
>> are zillions likely to work well.
>>
>> piglet
> 
> They may have had commercial reasons to make that choice.  Perhaps they were
> already buying millions to use in other products.  Maybe an alternative that might
> have been a "better" choice would have added another part to the design.  In many
> large companies there is a lot of pressure not to do that.
> John


Good point. Maybe also the low capacitances? Dunno.

On Sunday, May 14, 2023 at 8:48:42 AM UTC-4, Cursitor Doom wrote:
> Gentlemen, 
> 
> In the PSU section of that old Tek oscilloscope I've been 
> troubleshooting, I discovered several transistors had died. One of 
> these is puzzling me somewhat. It's a 2N5859 which they're using to 
> drive one of those big ol' TO-3 power transistors for the -15V output 
> section. I can't understand why they chose a fast switching device 
> with a GBP of 250Mhz. 

Ft of 250 MHz is general purpose, not real fast. That range of Ft means smaller reverse biased  output capacitance, Cob=7pF, making it much simpler to stabilize and easier to drive without destabilizing the driver electronics. It has a spectacular minimum hFE of 10 at Ic=1.0A and Tj= -55oC!!!- and probably typically 40 at 1.0A for other temps. They need that TO-39 package for power handling, TJmax of 200oC, and RJC of 35oC/W. The TO-3 power transistor probably has reduced hFE at full load ( <10 ) and elevated operating conditions, requiring a lot of drive current.

> 
> This being a linear PSU, in 1970 when these devices would have been 
> pretty expensive, it seems a curious choice. Fine, it can handle 80V 
> and pass 2 amps which is respectable in TO-39 in those days, but a 
> 250Mhz Ft? What's the point? Some dumb old medium power audio 
> transistor could do that job, surely? 

Not at all. If the regulator is for instrumentation and they want to push the 60Hz and its harmonics ripple feedthrough down by 100dB, then you need a high GBW closed loop feedback regulator. Input interference reduced by 1 + feedback beta x gain.

> 
> 
> The schematic: 
> https://tinyurl.com/4m8ffw7e 
> 
> Datasheet: 
> https://www.mouser.com/datasheet/2/68/2n5859-1368164.pdf

On Sun, 14 May 2023 09:45:04 -0700 (PDT), Fred Bloggs
<bloggs.fredbloggs.fred@gmail.com> wrote:

>On Sunday, May 14, 2023 at 8:48:42?AM UTC-4, Cursitor Doom wrote:
>> Gentlemen, 
>> 
>> In the PSU section of that old Tek oscilloscope I've been 
>> troubleshooting, I discovered several transistors had died. One of 
>> these is puzzling me somewhat. It's a 2N5859 which they're using to 
>> drive one of those big ol' TO-3 power transistors for the -15V output 
>> section. I can't understand why they chose a fast switching device 
>> with a GBP of 250Mhz. 
>
>Ft of 250 MHz is general purpose, not real fast. That range of Ft means smaller reverse biased  output capacitance, Cob=7pF, making it much simpler to stabilize and easier to drive without destabilizing the driver electronics. It has a spectacular minimum hFE of 10 at Ic=1.0A and Tj= -55oC!!!- and probably typically 40 at 1.0A for other temps. They need that TO-39 package for power handling, TJmax of 200oC, and RJC of 35oC/W. The TO-3 power transistor probably has reduced hFE at full load ( <10 ) and elevated operating conditions, requiring a lot of drive current.
>
>> 
>> This being a linear PSU, in 1970 when these devices would have been 
>> pretty expensive, it seems a curious choice. Fine, it can handle 80V 
>> and pass 2 amps which is respectable in TO-39 in those days, but a 
>> 250Mhz Ft? What's the point? Some dumb old medium power audio 
>> transistor could do that job, surely? 
>
>Not at all. If the regulator is for instrumentation and they want to push the 60Hz and its harmonics ripple feedthrough down by 100dB, then you need a high GBW closed loop feedback regulator. Input interference reduced by 1 + feedback beta x gain.
>

I wish I knew WTF you were talking about.

On Sunday, May 14, 2023 at 1:02:34 PM UTC-4, Cursitor Doom wrote:
> On Sun, 14 May 2023 09:45:04 -0700 (PDT), Fred Bloggs 
> <bloggs.fred...@gmail.com> wrote:
> >On Sunday, May 14, 2023 at 8:48:42?AM UTC-4, Cursitor Doom wrote: 
> >> Gentlemen, 
> >> 
> >> In the PSU section of that old Tek oscilloscope I've been 
> >> troubleshooting, I discovered several transistors had died. One of 
> >> these is puzzling me somewhat. It's a 2N5859 which they're using to 
> >> drive one of those big ol' TO-3 power transistors for the -15V output 
> >> section. I can't understand why they chose a fast switching device 
> >> with a GBP of 250Mhz. 
> > 
> >Ft of 250 MHz is general purpose, not real fast. That range of Ft means smaller reverse biased output capacitance, Cob=7pF, making it much simpler to stabilize and easier to drive without destabilizing the driver electronics. It has a spectacular minimum hFE of 10 at Ic=1.0A and Tj= -55oC!!!- and probably typically 40 at 1.0A for other temps. They need that TO-39 package for power handling, TJmax of 200oC, and RJC of 35oC/W. The TO-3 power transistor probably has reduced hFE at full load ( <10 ) and elevated operating conditions, requiring a lot of drive current. 
> > 
> >> 
> >> This being a linear PSU, in 1970 when these devices would have been 
> >> pretty expensive, it seems a curious choice. Fine, it can handle 80V 
> >> and pass 2 amps which is respectable in TO-39 in those days, but a 
> >> 250Mhz Ft? What's the point? Some dumb old medium power audio 
> >> transistor could do that job, surely? 
> > 
> >Not at all. If the regulator is for instrumentation and they want to push the 60Hz and its harmonics ripple feedthrough down by 100dB, then you need a high GBW closed loop feedback regulator. Input interference reduced by 1 + feedback beta x gain. 
> >
> I wish I knew WTF you were talking about.


It's right off the datasheet.

Write-up on performance level considerations for a prospective linear regulator *before* it's built:

(I just scanned quickly and it seems okay)

https://www.nisshinbo-microdevices.co.jp/en/design-support/basic/03-linear-regulator.html