Curious problem

[Cursitor Doom]

Hi all,

I'm now working on a different Philips scope which 'died' suddenly for no
apparent reason. Here's a summary of the key points:

The smps section had power in, but nothing coming out.
I checked the smps outputs for shorts/low-res.
One output marked 45V was trying to feed a board with an input resistance
of only 6 ohms.
I located the board in question. There were several multi-pin connectors
around the edge of it running off to other parts of the scope. Now it
turns out the smps runs perfectly fine when *one* of these connectors is
pulled from the board.
With the suspect connector's plug disconnected, the aforementioned 6 ohms
shoots up to about 15k and life is great.
*BUT* (and here's the thing) the resistance looking into the suspect plug
(which has only two wires) is several megohms. And yet this very high
resistance, when plugged into the board causes such a voltage drop that
the smps shuts down! How is this even possible?
I keep thinking there must be something simple I've overlooked, but can't
think what it might be.
If anyone has experienced something similar in the past and remembers
what the underlying issue was, then that could be very helpful!
Thanks.

 

[Cursitor Doom]

On Thu, 04 Feb 2016 10:26:46 -0500, Pat wrote:

Regarding "How is this even possible?", I can think of two things. One
is mechanical - plugging that 2 wire connector in causes a short near
the socket. Can you unplug the other end of that cable and see if you
get the result?

Already tried and eliminated, thanks.

The more likely cause is a voltage on that two wire

cable turns on a transistor or some other device hard wihich then shorts
the incoming line from the power supply due to a bad part somewhere else
on the board. Good luck.

That's what I was wondering. If my DVM on resistance range doesn't use
enough voltage to switch such a transistor or diode into conduction, then
that could account for it. I'll try it again with an old-style AVO.

 

[Cursitor Doom]

On Thu, 04 Feb 2016 16:08:38 +0000, MJC wrote:


> Mike.

"Current blasting"? It would have worked in this instance, but I've found
the problem with my good old trusty AVO which confirmed my previous
suspicions.
People often go on about how modern DVMs are great because they don't
turn on pn junctions, which is all very well, but there are odd times
when that is precisely what you *do* want the meter to do! This was one
of those.
Thanks, guys.

 

[Cursitor Doom]

On Thu, 04 Feb 2016 18:38:58 +0100, Reinhard Zwirner wrote:

> Would you be so kind and tell us what was wrong? Thanks in advance.

The problem was in the EHT converter section which creates 1600V from a
45V input. A BSW68 transistor in this section had fused low-resistance
between C&E and was attempting to dissipate 340 watts all by itself - and
it didn't even have a heatsink! The fact that its case had turned blue
also hinted that it had been very hot. I couldn't see it on inspection
initially as it was hidden by HV shielding.
Now I just have to find an equivalent for it...

 

On Thursday, February 4, 2016 at 1:44:26 PM UTC-5, Cursitor Doom wrote:

On Thu, 04 Feb 2016 18:38:58 +0100, Reinhard Zwirner wrote:

Would you be so kind and tell us what was wrong? Thanks in advance.

The problem was in the EHT converter section which creates 1600V from a
45V input. A BSW68 transistor in this section had fused low-resistance
between C&E and was attempting to dissipate 340 watts all by itself - and
it didn't even have a heatsink! The fact that its case had turned blue
also hinted that it had been very hot. I couldn't see it on inspection
initially as it was hidden by HV shielding.
Now I just have to find an equivalent for it...

The BSW68 crosses to an ECG/NTE 190. It would seem that an MPSU04 would work as well.

A BSW-68 seems to cross to an ECG/NTE 198. Here a TIP-47 or TIP-48 would do.

Check further. It may be easier to use a sub.

Dan

 

[Pat]

On Thu, 4 Feb 2016 13:59:27 -0000 (UTC), Cursitor Doom
<curd@notformail.com> wrote:

Hi all,

I'm now working on a different Philips scope which 'died' suddenly for no
apparent reason. Here's a summary of the key points:

The smps section had power in, but nothing coming out.
I checked the smps outputs for shorts/low-res.
One output marked 45V was trying to feed a board with an input resistance
of only 6 ohms.
I located the board in question. There were several multi-pin connectors
around the edge of it running off to other parts of the scope. Now it
turns out the smps runs perfectly fine when *one* of these connectors is
pulled from the board.
With the suspect connector's plug disconnected, the aforementioned 6 ohms
shoots up to about 15k and life is great.
*BUT* (and here's the thing) the resistance looking into the suspect plug
(which has only two wires) is several megohms. And yet this very high
resistance, when plugged into the board causes such a voltage drop that
the smps shuts down! How is this even possible?
I keep thinking there must be something simple I've overlooked, but can't
think what it might be.
If anyone has experienced something similar in the past and remembers
what the underlying issue was, then that could be very helpful!
Thanks.

Regarding "How is this even possible?", I can think of two things. One
is mechanical - plugging that 2 wire connector in causes a short near
the socket. Can you unplug the other end of that cable and see if you
get the result? The more likely cause is a voltage on that two wire
cable turns on a transistor or some other device hard wihich then
shorts the incoming line from the power supply due to a bad part
somewhere else on the board. Good luck.

 

[MJC]

In article <n8vlfv$7l5$1@dont-email.me>, curd@notformail.com says...

... *BUT* (and here's the thing) the resistance looking into the
suspect plug
(which has only two wires) is several megohms. And yet this very high
resistance, when plugged into the board causes such a voltage drop that
the smps shuts down! How is this even possible?

It's rather easily possible if the current draw is non-linear with
voltage. After all, the further the circuit deviates from a simple
resistor the more likely this is!

Apart from brutal ideas like providing the suspect plug with a very
beefy PSU and watching to see what burns up, I don't see what
alternative there is apart from gradual splitting off of parts of the
circuit...

Mike.

 

[Cursitor Doom]

On Thu, 04 Feb 2016 12:01:02 -0800, dansabrservices wrote:

On Thursday, February 4, 2016 at 1:44:26 PM UTC-5, Cursitor Doom wrote:
On Thu, 04 Feb 2016 18:38:58 +0100, Reinhard Zwirner wrote:

Would you be so kind and tell us what was wrong? Thanks in advance.

The problem was in the EHT converter section which creates 1600V from a
45V input. A BSW68 transistor in this section had fused low-resistance
between C&E and was attempting to dissipate 340 watts all by itself -
and it didn't even have a heatsink! The fact that its case had
turned blue also hinted that it had been very hot. I couldn't see it on
inspection initially as it was hidden by HV shielding.
Now I just have to find an equivalent for it...

The BSW68 crosses to an ECG/NTE 190. It would seem that an MPSU04 would
work as well.

A BSW-68 seems to cross to an ECG/NTE 198. Here a TIP-47 or TIP-48
would do.

Check further. It may be easier to use a sub.

Dan

Thank you for those suggestions, Dan. I've got scores of TIP BJT variants
lying around in my junk box.
Many years ago I had a copy of Towers' Transistor Selector which listed a
good range of equivalents for a vast range of devices. I guess these days
there are web sites that perform the same function, but finding one
worthy of trust that produces reliable results is another matter from
what I've seen so far.

 

For starters, I usually use the NTE cross reference. Since I have a rather large collection of NTE parts, I use these for experiments. I don't usually use NTE or ECG parts for a repair though. I try to use "real" parts. It does at least give me a quick try.

Dan

 

[Reinhard Zwirner]

Cursitor Doom schrieb:

On Thu, 04 Feb 2016 16:08:38 +0000, MJC wrote:


Mike.

"Current blasting"? It would have worked in this instance, but I've found
the problem with my good old trusty AVO which confirmed my previous
suspicions. ...

Would you be so kind and tell us what was wrong? Thanks in advance.

Best regardsw

Reinhard

 

[Cursitor Doom]

On Thu, 04 Feb 2016 14:05:40 -0800, dansabrservices wrote:

For starters, I usually use the NTE cross reference. Since I have a
rather large collection of NTE parts, I use these for experiments. I
don't usually use NTE or ECG parts for a repair though. I try to use
"real" parts. It does at least give me a quick try.

Good idea.
Maybe I'll look into subbing that BJT for a power MOSFET. A lot of older
gear uses BJTs in situations where a MOSFET would be de rigeur nowadays.
This transistor functions as a chopper switching an inductive load in
this circuit and it wouldn't require much tweaking to sub it for a better
device, I reckon. In fact maybe that's why the original part failed or
maybe its heatsink - if it ever had one - just fell off at some point, I
can't say.

 

[legg]

On Thu, 4 Feb 2016 18:41:31 -0000 (UTC), Cursitor Doom
<curd@notformail.com> wrote:

On Thu, 04 Feb 2016 18:38:58 +0100, Reinhard Zwirner wrote:

Would you be so kind and tell us what was wrong? Thanks in advance.

The problem was in the EHT converter section which creates 1600V from a
45V input. A BSW68 transistor in this section had fused low-resistance
between C&E and was attempting to dissipate 340 watts all by itself - and
it didn't even have a heatsink! The fact that its case had turned blue
also hinted that it had been very hot. I couldn't see it on inspection
initially as it was hidden by HV shielding.
Now I just have to find an equivalent for it...

Your Towers would have suggested 2n3439/2N3440 or a BUY60.
TO5 body sizes are no longer common, but those types are still listed
by digikey.

BSW68 is 150V, 1A, 40MHz, 800mW, 30min hfe.

A ZTX857, in a silicone E-Line body, could probably do it.

RL

 

[Cursitor Doom]

On Sat, 06 Feb 2016 02:09:03 -0500, legg wrote:

On Thu, 4 Feb 2016 18:41:31 -0000 (UTC), Cursitor Doom
curd@notformail.com> wrote:

On Thu, 04 Feb 2016 18:38:58 +0100, Reinhard Zwirner wrote:

Would you be so kind and tell us what was wrong? Thanks in advance.

The problem was in the EHT converter section which creates 1600V from a
45V input. A BSW68 transistor in this section had fused low-resistance
between C&E and was attempting to dissipate 340 watts all by itself -
and it didn't even have a heatsink! The fact that its case had turned
blue also hinted that it had been very hot. I couldn't see it on
inspection initially as it was hidden by HV shielding.
Now I just have to find an equivalent for it...

Your Towers would have suggested 2n3439/2N3440 or a BUY60.
TO5 body sizes are no longer common, but those types are still listed by
digikey.

BSW68 is 150V, 1A, 40MHz, 800mW, 30min hfe.

A ZTX857, in a silicone E-Line body, could probably do it.

RL

Thanks, legg. I'll have a rummage round my copious junk box. I doubt the
ft figure has much bearing in this application as whatever the rate this
thing switches at it won't be anywhere close to a fraction of that.
Likewise the hfe is definitely on the low side, so I shouldn't have any
trouble matching or bettering that in a subbed part.
Just to reiterate here, this is NOT the main chopper in the smps section;
it's a self-standing HT generator card with its chopper being driven by a
single transistor oscillator stage. In fact it really wouldn't take much
modding to swap the BJT for a MOSFET but I'd prefer to stick with the
original circuit if poss.

 

[legg]

On Sat, 6 Feb 2016 12:34:17 -0000 (UTC), Cursitor Doom
<curd@notformail.com> wrote:

On Sat, 06 Feb 2016 02:09:03 -0500, legg wrote:

On Thu, 4 Feb 2016 18:41:31 -0000 (UTC), Cursitor Doom
curd@notformail.com> wrote:

On Thu, 04 Feb 2016 18:38:58 +0100, Reinhard Zwirner wrote:

Would you be so kind and tell us what was wrong? Thanks in advance.

The problem was in the EHT converter section which creates 1600V from a
45V input. A BSW68 transistor in this section had fused low-resistance
between C&E and was attempting to dissipate 340 watts all by itself -
and it didn't even have a heatsink! The fact that its case had turned
blue also hinted that it had been very hot. I couldn't see it on
inspection initially as it was hidden by HV shielding.
Now I just have to find an equivalent for it...

Your Towers would have suggested 2n3439/2N3440 or a BUY60.
TO5 body sizes are no longer common, but those types are still listed by
digikey.

BSW68 is 150V, 1A, 40MHz, 800mW, 30min hfe.

A ZTX857, in a silicone E-Line body, could probably do it.

RL

Thanks, legg. I'll have a rummage round my copious junk box. I doubt the
ft figure has much bearing in this application as whatever the rate this
thing switches at it won't be anywhere close to a fraction of that.
Likewise the hfe is definitely on the low side, so I shouldn't have any
trouble matching or bettering that in a subbed part.
Just to reiterate here, this is NOT the main chopper in the smps section;
it's a self-standing HT generator card with its chopper being driven by a
single transistor oscillator stage. In fact it really wouldn't take much
modding to swap the BJT for a MOSFET but I'd prefer to stick with the
original circuit if poss.

ft must be equal or greater, similarly beta, which puts a modest
amount of doubt for the 3439/40 parts. They are slower general purpose
parts, with higher voltage ratings that are accompanied by lower beta
(possibly 25% lower at 1A).

The ZTX part is offered as an example of more recent physical
equivalents of higher speed switches in the same voltage range. It's
silicone package has the same temperature limts as hermetic and it can
benefit just as much by heatsinking methods common to TO5, if present
in the current application. It should gunction with whatever base
drive and SOA load line tailoring that was applied to the original,
with reduced switching and conduction losses.

I wouldn't advise a switch technology swap until you become more
familiar with the circuit functioning as originally intended. There is
seldom any noticeable benefit in low/medium power applications.

RL