Topics

7603 15+ rail dead. Options?


Dave Voorhis
 

I’m bringing a 7603 back from having released its magic smoke. The +15 volt rail is dead, the 0.1 ohm current sense resistor R935 having gotten hot enough to have unsoldered itself and fallen out. Q926, Q931, and Q936 are defective. Q933 and relevant surrounding components appear to have survived.

A replacement Q931 (2N5189) is easily and cheaply found, but a new replacement for Q936 (2N2919) costs about twice what I paid for the oscilloscope and Q926 doesn’t appear to have a specified substitute — though I presume any small NPN transistor capable of handling 60 volts or so would be fine.

Is there any good reason why I can't simply obtain +15 v from a 2 amp 15 volt regulator chip — like an L78S15CS — and ignore the +15v sense line?

I’ve verified that the ‘scope essentially works with +15v provided by my bench power supply and draws a little over 1 amp.

For reference, I’ve linked an image of the relevant page from the power supply schematic here: http://www.armchair.mb.ca/~dave/tmp/Tek7603_plus15v.png


Craig Sawyers <c.sawyers@...>
 

The only reason that Tek used a dual is - well because they were more easily available back then. I
would just take two generic NPN TO92 parts with gains of about 100-200 (you could easily match from a
bunch of them to 10% or so) and a Vceo of 60V or so, stick them together with heatsink compound and
shove them in.

It is just a long tailed pair error amplifier after all.

The actual single device that is in the dual is a 2N2484 (according to the Tek parts catalogue), which
are easily available and cheap in a TO18 package. Buy a bunch and match 2 for hfe to 10% and strap
them together. Job done.

Mouser $2 each or $18.50 for ten. Possibly even cheaper if you shop around.

Craig

A replacement Q931 (2N5189) is easily and cheaply found, but a new replacement for Q936 (2N2919)
costs about twice what I paid for the oscilloscope and Q926 doesn t appear to have a specified
substitute though I presume any small NPN transistor capable of handling 60 volts or so would be
fine.


Craig Sawyers <c.sawyers@...>
 

Here you go - PN2484, the plastic packaged version of the 2N2484. Mouser 42c each or $3.51 for 10. Buy
ten and find two with close hfe, shove them in.

A replacement Q931 (2N5189) is easily and cheaply found, but a new replacement for Q936 (2N2919)
costs about twice what I paid for the oscilloscope and Q926 doesn t appear to have a specified
substitute though I presume any small NPN transistor capable of handling 60 volts or so would be
fine.


Dave Voorhis
 

Excellent; thanks!

I might even have suitable transistors already knocking around in my parts bins.

Without much thought, I’d assumed the use of a dual packaging meant there was some critical need to match temperature as closely as possible, but if it's non-critical enough to solve with discrete devices and heatsink compound, that’s good.

On 26 Feb 2018, at 15:31, Craig Sawyers <c.sawyers@tech-enterprise.com> wrote:

Here you go - PN2484, the plastic packaged version of the 2N2484. Mouser 42c each or $3.51 for 10. Buy
ten and find two with close hfe, shove them in.

A replacement Q931 (2N5189) is easily and cheaply found, but a new replacement for Q936 (2N2919)
costs about twice what I paid for the oscilloscope and Q926 doesn t appear to have a specified
substitute though I presume any small NPN transistor capable of handling 60 volts or so would be
fine.
On 26 Feb 2018, at 15:23, Craig Sawyers <c.sawyers@tech-enterprise.com> wrote:

The only reason that Tek used a dual is - well because they were more easily available back then. I
would just take two generic NPN TO92 parts with gains of about 100-200 (you could easily match from a
bunch of them to 10% or so) and a Vceo of 60V or so, stick them together with heatsink compound and
shove them in.

It is just a long tailed pair error amplifier after all.

The actual single device that is in the dual is a 2N2484 (according to the Tek parts catalogue), which
are easily available and cheap in a TO18 package. Buy a bunch and match 2 for hfe to 10% and strap
them together. Job done.

Mouser $2 each or $18.50 for ten. Possibly even cheaper if you shop around.


Dave Hills
 

Is there any good reason why I can't simply obtain +15 v from a 2 amp 15 volt regulator chip — like an L78S15CS — and ignore the +15v sense line?>>
Three that I can think of.

First, you will disrupt the power-up sequences of the various DC supplies. Many IC's and discrete circuits don't like seeing positive
voltages before the negative rail comes up. May or may not be a problem, but you have to ask why the various regulators are sequenced as they are?
Second, what will happen to the calibration of your plug-ins when +15v is outside the limits of the original design and no longer tracks the -50v supply? Lastly,
+15v current limit, will something go @#poof--!! if your supply is not limited to the same value and function as the original?

Also, a thin layer of epoxy bonding together the flat sides of a pair of TO-92's is would be my choice over thermal compound.

Dave


Dave Voorhis
 

Thanks, point(s) taken.

I’ve ordered appropriate discrete parts, should have the power supply working in a few days, and can then move on to other problems.

On 27 Feb 2018, at 03:31, dadhills@mindspring.com wrote:

Is there any good reason why I can't simply obtain +15 v from a 2 amp 15 volt regulator chip — like an L78S15CS — and ignore the +15v sense line?>>
Three that I can think of.

First, you will disrupt the power-up sequences of the various DC supplies. Many IC's and discrete circuits don't like seeing positive
voltages before the negative rail comes up. May or may not be a problem, but you have to ask why the various regulators are sequenced as they are?
Second, what will happen to the calibration of your plug-ins when +15v is outside the limits of the original design and no longer tracks the -50v supply? Lastly,
+15v current limit, will something go @#poof--!! if your supply is not limited to the same value and function as the original?

Also, a thin layer of epoxy bonding together the flat sides of a pair of TO-92's is would be my choice over thermal compound.

Dave


 

I don't think it is quite that simple. It makes sense to use
transistor duals for differential pairs because it avoids having to
match separate transistors and the close thermal coupling improves
precision. Cheap integrated operational amplifiers pretty much
replaced them.

In this case. Q936 is the error amplifier for limiting the current.
Unfortunately Tektronix gives no offset voltage specifications for
their dual bipolar transistors but 100 millivolts of offset will yield
a current error of 1 amp through that 0.1 ohm current shunt. Given
the high cost of 6 lead dual transistors, matching a pair of 2N3904s
or 2N4401s or even better, high gain 2N5088, 2N5089s, or BC546s and
gluing them together will work fine.

The matching procedure is pretty easy. Short the base and collector
and measure the voltage drop across the base-emitter junction at the
operating current which in this case is 1.2 milliamps. In practice,
the diode test function on a digital multimeter which measures the
voltage will work fine. Then pick the two transistors with the
closest match.

The NPN BC546 mentioned above and its PNP complement, the BC556, are
good parts to keep around for replacing a variety of high gain
transistors used for differential amplifiers. Just watch out because
they use the CBE pinout instead of the ECB pinout common with 2N
parts.

On Mon, 26 Feb 2018 15:23:05 -0000, you wrote:

The only reason that Tek used a dual is - well because they were more easily available back then. I
would just take two generic NPN TO92 parts with gains of about 100-200 (you could easily match from a
bunch of them to 10% or so) and a Vceo of 60V or so, stick them together with heatsink compound and
shove them in.

It is just a long tailed pair error amplifier after all.

The actual single device that is in the dual is a 2N2484 (according to the Tek parts catalogue), which
are easily available and cheap in a TO18 package. Buy a bunch and match 2 for hfe to 10% and strap
them together. Job done.

Mouser $2 each or $18.50 for ten. Possibly even cheaper if you shop around.

Craig


 

On Mon, 26 Feb 2018 14:24:45 +0000, you wrote:

... and Q926 doesn’t appear to have a specified substitute — though I presume any small NPN transistor capable of handling 60 volts or so would be fine.
Hmm, Q926 is 151-0347-00 6-6&12 which is a 2N5551, 100MHz 160V 600mA
TO-92. You might get away with a 2N4401 selected for high Vceo but
one of the many high voltage TO-92 parts is the way to go and the
2N5551 is still produced and the best option.

Is there any good reason why I can't simply obtain +15 v from a 2 amp 15 volt regulator chip — like an L78S15CS — and ignore the +15v sense line?
The regulators are designed to track each other so that would not be a
good idea. If there was a short on the -50 volt supply which is used
as a reference, the +15 volt supply is suppose to drop.