Topics

Strange Tek2440 issue


Vince Vielhaber
 

I picked up a 2440 on ebay that had the dead nvram battery problem. Got that fixed up with an external battery. Ran thru the self cal and ext cal and all seems fine, except when the sweep rate is turned above 100 or 200 ns.

In trying to research this a bit, the only thing I could find was on one of the Tek forums back in 2017 but there were zero responses. Here is how the other person described the problem:

---------------------------------------------------------------------
The 5 volt regulated rail is getting crowbared when the scope Sec/Div is set to 100ns or faster. If save is pressed at a slower setting the Sec/Div can then be set to any value without the 5 Volt rail getting pulled down. It appears the issue only occurs during an active acquisition cycle at 100ns Sec/Div or faster.
----------------------------------------------------------------------

While I didn't measure the 5v rail, the screen shrinks a couple of times as the waveform changes, the inputs get switched to GND and the display goes blank. Sometimes, if I'm quick enough, I can turn the sweep back down (1us or slower) and it'll recover, but I still have to go to the input coupling and switch the inputs off of GND.

Anyone seen this or have any ideas what's going on?

Thanks!
Vince.
--
K8ZW http://www.metalworkingfun.com http://www.hamradio.fun


Bob Albert
 

Not your current problem but I replaced the NVRAM chips in my 2440 and the scope doesn't work.  I switched one of them back and still no joy.  I will try the other one.
I probably got some counterfeit modules from China.  I am hoping it will be fixed when I put the old one back.   How did you fix yours?
Bob

On Sunday, August 30, 2020, 02:37:52 PM PDT, Vince Vielhaber <vev@...> wrote:


I picked up a 2440 on ebay that had the dead nvram battery problem.  Got
that fixed up with an external battery.  Ran thru the self cal and ext
cal and all seems fine, except when the sweep rate is turned above 100
or 200 ns.

In trying to research this a bit, the only thing I could find was on one
of the Tek forums back in 2017 but there were zero responses.  Here is
how the other person described the problem:

---------------------------------------------------------------------
The 5 volt regulated rail is getting crowbared when the scope Sec/Div is
set to 100ns or faster. If save is pressed at a slower setting the
Sec/Div can then be set to any value without the 5 Volt rail getting
pulled down. It appears the issue only occurs during an active
acquisition cycle at 100ns Sec/Div or faster.
----------------------------------------------------------------------

While I didn't measure the 5v rail, the screen shrinks a couple of times
as the waveform changes, the inputs get switched to GND and the display
goes blank.  Sometimes, if I'm quick enough, I can turn the sweep back
down (1us or slower) and it'll recover, but I still have to go to the
input coupling and switch the inputs off of GND.

Anyone seen this or have any ideas what's going on?

Thanks!
Vince.
--
  K8ZW  http://www.metalworkingfun.com   http://www.hamradio.fun


Vince Vielhaber
 

Szabolcs, from here (TekScopes) posted a method of accessing the internal battery of the Dallas NVRAM modules using a heat gun and a scraper. The difference in these modules was that there were two batteries that were visible. Anyway I scraped down to pin 7 of the DS1210 and soldered a wire to it on each module. I ran those to a small circuit board where I had a battery holder and two diodes on it. The diodes fed the voltage to each of the modules.

When I put it back together it didn't work. Then I noticed I forgot to plug in one of the ribbon cables!

Anyway, the subject for the post I mentioned above is:

Easy method for NVRAM battery replacement

Vince.

On 08/30/2020 05:53 PM, Bob Albert via groups.io wrote:
Not your current problem but I replaced the NVRAM chips in my 2440 and the scope doesn't work. I switched one of them back and still no joy. I will try the other one.
I probably got some counterfeit modules from China. I am hoping it will be fixed when I put the old one back. How did you fix yours?
Bob
On Sunday, August 30, 2020, 02:37:52 PM PDT, Vince Vielhaber <vev@...> wrote:


I picked up a 2440 on ebay that had the dead nvram battery problem. Got
that fixed up with an external battery. Ran thru the self cal and ext
cal and all seems fine, except when the sweep rate is turned above 100
or 200 ns.

In trying to research this a bit, the only thing I could find was on one
of the Tek forums back in 2017 but there were zero responses. Here is
how the other person described the problem:

---------------------------------------------------------------------
The 5 volt regulated rail is getting crowbared when the scope Sec/Div is
set to 100ns or faster. If save is pressed at a slower setting the
Sec/Div can then be set to any value without the 5 Volt rail getting
pulled down. It appears the issue only occurs during an active
acquisition cycle at 100ns Sec/Div or faster.
----------------------------------------------------------------------

While I didn't measure the 5v rail, the screen shrinks a couple of times
as the waveform changes, the inputs get switched to GND and the display
goes blank. Sometimes, if I'm quick enough, I can turn the sweep back
down (1us or slower) and it'll recover, but I still have to go to the
input coupling and switch the inputs off of GND.

Anyone seen this or have any ideas what's going on?

Thanks!
Vince.


Roger Evans
 

100nsec per div and faster sounds like the switch from real time sampling to equivalent time interpolation (300Msample/sec real time). You can turn off the equivalent time mode by going to the Acquire menu and setting 'Repet' to Off. If that enables you to get to faster sweep speeds then the problem is clearly in the circuitry that does the equivalent time interpolation but I would need a thorough read of the manual to go into any detail!

Regards,

Roger


Vince Vielhaber
 

Repet on, Repet off, no difference.

Vince.

On 08/31/2020 01:42 AM, Roger Evans via groups.io wrote:
100nsec per div and faster sounds like the switch from real time sampling to equivalent time interpolation (300Msample/sec real time). You can turn off the equivalent time mode by going to the Acquire menu and setting 'Repet' to Off. If that enables you to get to faster sweep speeds then the problem is clearly in the circuitry that does the equivalent time interpolation but I would need a thorough read of the manual to go into any detail!

Regards,

Roger


Siggi
 

Hey Vince,

I have no idea what's going on there, though I'd suggest looking at the
power supplies as a first step.
You can also - perhaps - diagnose this a little ways by playing with
trigger and acquisition settings. My guess would be that a power supply is
collapsing when the sample clock exceeds some particular rate, though
that's of course just a guess.
Does this still occur when the acquisition is stopped? If not, you can
maybe narrow this down by e.g. going to NORM sweep/trigger mode, and
setting up a trigger that never hits. Does it still happen when acquisition
is started but no trigger fires? How about under a single-shot acquisition
in AUTO mode, etc.

Siggi

On Sun, Aug 30, 2020 at 5:37 PM Vince Vielhaber <vev@...> wrote:


I picked up a 2440 on ebay that had the dead nvram battery problem. Got
that fixed up with an external battery. Ran thru the self cal and ext
cal and all seems fine, except when the sweep rate is turned above 100
or 200 ns.

In trying to research this a bit, the only thing I could find was on one
of the Tek forums back in 2017 but there were zero responses. Here is
how the other person described the problem:

---------------------------------------------------------------------
The 5 volt regulated rail is getting crowbared when the scope Sec/Div is
set to 100ns or faster. If save is pressed at a slower setting the
Sec/Div can then be set to any value without the 5 Volt rail getting
pulled down. It appears the issue only occurs during an active
acquisition cycle at 100ns Sec/Div or faster.
----------------------------------------------------------------------

While I didn't measure the 5v rail, the screen shrinks a couple of times
as the waveform changes, the inputs get switched to GND and the display
goes blank. Sometimes, if I'm quick enough, I can turn the sweep back
down (1us or slower) and it'll recover, but I still have to go to the
input coupling and switch the inputs off of GND.

Anyone seen this or have any ideas what's going on?

Thanks!
Vince.
--
K8ZW http://www.metalworkingfun.com http://www.hamradio.fun




Vince Vielhaber
 

Acquisition stopped, doesn't happen. I can go all the way to 2ns/div.

Normal trigger but trigger set to +or- 90v (not triggering), it happens.

Single, doesn't happen, can go all the way to 2ns/div, can even trigger it (in single) and it doesn't happen. The probe was hooked to the calibrator for the triggers. Normally doesn't matter if there's a signal there or not.

Thanks!
Vince.

On 08/31/2020 01:43 PM, Siggi wrote:
Hey Vince,

I have no idea what's going on there, though I'd suggest looking at the
power supplies as a first step.
You can also - perhaps - diagnose this a little ways by playing with
trigger and acquisition settings. My guess would be that a power supply is
collapsing when the sample clock exceeds some particular rate, though
that's of course just a guess.
Does this still occur when the acquisition is stopped? If not, you can
maybe narrow this down by e.g. going to NORM sweep/trigger mode, and
setting up a trigger that never hits. Does it still happen when acquisition
is started but no trigger fires? How about under a single-shot acquisition
in AUTO mode, etc.

Siggi

On Sun, Aug 30, 2020 at 5:37 PM Vince Vielhaber <vev@...> wrote:


I picked up a 2440 on ebay that had the dead nvram battery problem. Got
that fixed up with an external battery. Ran thru the self cal and ext
cal and all seems fine, except when the sweep rate is turned above 100
or 200 ns.

In trying to research this a bit, the only thing I could find was on one
of the Tek forums back in 2017 but there were zero responses. Here is
how the other person described the problem:

---------------------------------------------------------------------
The 5 volt regulated rail is getting crowbared when the scope Sec/Div is
set to 100ns or faster. If save is pressed at a slower setting the
Sec/Div can then be set to any value without the 5 Volt rail getting
pulled down. It appears the issue only occurs during an active
acquisition cycle at 100ns Sec/Div or faster.
----------------------------------------------------------------------

While I didn't measure the 5v rail, the screen shrinks a couple of times
as the waveform changes, the inputs get switched to GND and the display
goes blank. Sometimes, if I'm quick enough, I can turn the sweep back
down (1us or slower) and it'll recover, but I still have to go to the
input coupling and switch the inputs off of GND.

Anyone seen this or have any ideas what's going on?

Thanks!
Vince.
--
K8ZW http://www.metalworkingfun.com http://www.hamradio.fun




Siggi
 

On Mon, Aug 31, 2020 at 2:04 PM Vince Vielhaber <vev@...> wrote:

Acquisition stopped, doesn't happen. I can go all the way to 2ns/div.

Normal trigger but trigger set to +or- 90v (not triggering), it happens.

Single, doesn't happen, can go all the way to 2ns/div, can even trigger
it (in single) and it doesn't happen. The probe was hooked to the
calibrator for the triggers. Normally doesn't matter if there's a
signal there or not.
Interesting. I'd look at the power supplies, should be easy enough to see
what's happening as this is reproducible. Do you have a second digital or
storage scope to look at what's happening on the power rails?
The only difference between single shot and run without trigger is that the
single shot capture has a very limited duration. Maybe the problem is that
a power rail is marginal, and so gets dragged down when the CCDs & capture
clock keep running for a while.
Once you start capturing the CCDs start storing (and discarding) the
signal, waiting for a trigger.


Vince Vielhaber
 

Yep, I have a 2430A and a couple of 11402s. I used the 2430A.

I checked the + and -5 and the + and -15 volt rails. Sure enough, the +5 volt rail drops exactly 1 volt while it's happening.

I'm thinking, would piggybacking another 5 volt source on that rail isolate if it's the power supply or something else?

Thanks!
Vince.

On 08/31/2020 05:09 PM, Siggi wrote:
On Mon, Aug 31, 2020 at 2:04 PM Vince Vielhaber <vev@...> wrote:

Acquisition stopped, doesn't happen. I can go all the way to 2ns/div.

Normal trigger but trigger set to +or- 90v (not triggering), it happens.

Single, doesn't happen, can go all the way to 2ns/div, can even trigger
it (in single) and it doesn't happen. The probe was hooked to the
calibrator for the triggers. Normally doesn't matter if there's a
signal there or not.
Interesting. I'd look at the power supplies, should be easy enough to see
what's happening as this is reproducible. Do you have a second digital or
storage scope to look at what's happening on the power rails?
The only difference between single shot and run without trigger is that the
single shot capture has a very limited duration. Maybe the problem is that
a power rail is marginal, and so gets dragged down when the CCDs & capture
clock keep running for a while.
Once you start capturing the CCDs start storing (and discarding) the
signal, waiting for a trigger.


Siggi
 

Hey Vince,

My bet would be that if you go back to the 5VD supply, you'll find that
C455 and/or C262 are shot. CR354 is also a suspect, but the bulk caps on
the 5VD supply are the primaries. If you can find a way to look at the
supply before L256, I betcha you're going to find it all jaggy and
disgusting.
If you have a way to measure the two capacitors, that'd work too. Assuming
F269 is not soldered in, you can pretty much isolate the two capacitors by
popping it out.

Good luck,
Siggi

On Mon, Aug 31, 2020 at 6:23 PM Vince Vielhaber <vev@...> wrote:

Yep, I have a 2430A and a couple of 11402s. I used the 2430A.

I checked the + and -5 and the + and -15 volt rails. Sure enough, the
+5 volt rail drops exactly 1 volt while it's happening.

I'm thinking, would piggybacking another 5 volt source on that rail
isolate if it's the power supply or something else?

Thanks!
Vince.


On 08/31/2020 05:09 PM, Siggi wrote:
On Mon, Aug 31, 2020 at 2:04 PM Vince Vielhaber <vev@...> wrote:

Acquisition stopped, doesn't happen. I can go all the way to 2ns/div.

Normal trigger but trigger set to +or- 90v (not triggering), it happens.

Single, doesn't happen, can go all the way to 2ns/div, can even trigger
it (in single) and it doesn't happen. The probe was hooked to the
calibrator for the triggers. Normally doesn't matter if there's a
signal there or not.
Interesting. I'd look at the power supplies, should be easy enough to see
what's happening as this is reproducible. Do you have a second digital or
storage scope to look at what's happening on the power rails?
The only difference between single shot and run without trigger is that
the
single shot capture has a very limited duration. Maybe the problem is
that
a power rail is marginal, and so gets dragged down when the CCDs &
capture
clock keep running for a while.
Once you start capturing the CCDs start storing (and discarding) the
signal, waiting for a trigger.


--
K8ZW http://www.metalworkingfun.com http://www.hamradio.fun




Vince Vielhaber
 

F269 is soldered but I was able to clip onto L256. A minor amount of ripple that only changed by a few millivolts when I added a 2200uf cap across it. No change when I switch below 200ns.

The 5V jumping around is the 5V regulated that C494 is on. There it drops a volt or so when I switch below 200ns. According to my Fluke, when it goes out completely it holds at 3.93v.

The 10v reference that divided and sent to U870B is solid.

Thanks!
Vince.

On 08/31/2020 06:43 PM, Siggi wrote:
Hey Vince,

My bet would be that if you go back to the 5VD supply, you'll find that
C455 and/or C262 are shot. CR354 is also a suspect, but the bulk caps on
the 5VD supply are the primaries. If you can find a way to look at the
supply before L256, I betcha you're going to find it all jaggy and
disgusting.
If you have a way to measure the two capacitors, that'd work too. Assuming
F269 is not soldered in, you can pretty much isolate the two capacitors by
popping it out.

Good luck,
Siggi

On Mon, Aug 31, 2020 at 6:23 PM Vince Vielhaber <vev@...> wrote:

Yep, I have a 2430A and a couple of 11402s. I used the 2430A.

I checked the + and -5 and the + and -15 volt rails. Sure enough, the
+5 volt rail drops exactly 1 volt while it's happening.

I'm thinking, would piggybacking another 5 volt source on that rail
isolate if it's the power supply or something else?

Thanks!
Vince.


On 08/31/2020 05:09 PM, Siggi wrote:
On Mon, Aug 31, 2020 at 2:04 PM Vince Vielhaber <vev@...> wrote:

Acquisition stopped, doesn't happen. I can go all the way to 2ns/div.

Normal trigger but trigger set to +or- 90v (not triggering), it happens.

Single, doesn't happen, can go all the way to 2ns/div, can even trigger
it (in single) and it doesn't happen. The probe was hooked to the
calibrator for the triggers. Normally doesn't matter if there's a
signal there or not.
Interesting. I'd look at the power supplies, should be easy enough to see
what's happening as this is reproducible. Do you have a second digital or
storage scope to look at what's happening on the power rails?
The only difference between single shot and run without trigger is that
the
single shot capture has a very limited duration. Maybe the problem is
that
a power rail is marginal, and so gets dragged down when the CCDs &
capture
clock keep running for a while.
Once you start capturing the CCDs start storing (and discarding) the
signal, waiting for a trigger.


--
K8ZW http://www.metalworkingfun.com http://www.hamradio.fun




Szabolcs Szigeti
 

Hi,

I would first check the power supply. I'm not a big fan of "the universal
internet method of fixing everything", that is indiscriminately recapping
stuff, but in the case of the 24xx family DSOs, I had 3 of those and all
had practically every cap failed in the PSU. So a complete replacement of
capacitors at the secondary side is one thing I would do first at the
slightest suspicion of power problems.

Szabolcs


Vince Vielhaber <vev@...> ezt írta (időpont: 2020. aug. 31., H,
20:04):

Acquisition stopped, doesn't happen. I can go all the way to 2ns/div.

Normal trigger but trigger set to +or- 90v (not triggering), it happens.

Single, doesn't happen, can go all the way to 2ns/div, can even trigger
it (in single) and it doesn't happen. The probe was hooked to the
calibrator for the triggers. Normally doesn't matter if there's a
signal there or not.

Thanks!
Vince.


On 08/31/2020 01:43 PM, Siggi wrote:
Hey Vince,

I have no idea what's going on there, though I'd suggest looking at the
power supplies as a first step.
You can also - perhaps - diagnose this a little ways by playing with
trigger and acquisition settings. My guess would be that a power supply
is
collapsing when the sample clock exceeds some particular rate, though
that's of course just a guess.
Does this still occur when the acquisition is stopped? If not, you can
maybe narrow this down by e.g. going to NORM sweep/trigger mode, and
setting up a trigger that never hits. Does it still happen when
acquisition
is started but no trigger fires? How about under a single-shot
acquisition
in AUTO mode, etc.

Siggi

On Sun, Aug 30, 2020 at 5:37 PM Vince Vielhaber <vev@...> wrote:


I picked up a 2440 on ebay that had the dead nvram battery problem. Got
that fixed up with an external battery. Ran thru the self cal and ext
cal and all seems fine, except when the sweep rate is turned above 100
or 200 ns.

In trying to research this a bit, the only thing I could find was on one
of the Tek forums back in 2017 but there were zero responses. Here is
how the other person described the problem:

---------------------------------------------------------------------
The 5 volt regulated rail is getting crowbared when the scope Sec/Div is
set to 100ns or faster. If save is pressed at a slower setting the
Sec/Div can then be set to any value without the 5 Volt rail getting
pulled down. It appears the issue only occurs during an active
acquisition cycle at 100ns Sec/Div or faster.
----------------------------------------------------------------------

While I didn't measure the 5v rail, the screen shrinks a couple of times
as the waveform changes, the inputs get switched to GND and the display
goes blank. Sometimes, if I'm quick enough, I can turn the sweep back
down (1us or slower) and it'll recover, but I still have to go to the
input coupling and switch the inputs off of GND.

Anyone seen this or have any ideas what's going on?

Thanks!
Vince.
--
K8ZW http://www.metalworkingfun.com http://www.hamradio.fun




--
K8ZW http://www.metalworkingfun.com http://www.hamradio.fun




Siggi
 

On Mon, Aug 31, 2020 at 8:36 PM Vince Vielhaber <vev@...> wrote:

The 5V jumping around is the 5V regulated that C494 is on. There it
drops a volt or so when I switch below 200ns. According to my Fluke,
when it goes out completely it holds at 3.93v.
Oh, interesting. I assumed you were talking about the 5VD, which is the
main switch-mode regulated supply that feeds all the digital logic.

You're talking about the linear regulated 5V supply. Same deal though, go
back to the first decoupling cap, see whether the supply is jaggy and
whether the raw supply holds up when the regulated supply drops. Most
likely - as others have noted - this is due to geriatric bulk caps in the
secondary, so check that first.
If the raw supply is holding up OK when the regulated supply drops, then
you move downstream and see what's up. This supply is current limited, so
you can infer the current it's using by measuring across the shunt.


Siggi
 

On Tue, Sep 1, 2020 at 10:04 AM Siggi via groups.io <siggi=
undo.com@groups.io> wrote:

On Mon, Aug 31, 2020 at 8:36 PM Vince Vielhaber <vev@...> wrote:

The 5V jumping around is the 5V regulated that C494 is on. There it
drops a volt or so when I switch below 200ns. According to my Fluke,
when it goes out completely it holds at 3.93v.
Oh, interesting. I assumed you were talking about the 5VD, which is the
main switch-mode regulated supply that feeds all the digital logic.

You're talking about the linear regulated 5V supply. Same deal though, go
back to the first decoupling cap, see whether the supply is jaggy and
whether the raw supply holds up when the regulated supply drops. Most
likely - as others have noted - this is due to geriatric bulk caps in the
secondary, so check that first.
If the raw supply is holding up OK when the regulated supply drops, then
you move downstream and see what's up. This supply is current limited, so
you can infer the current it's using by measuring across the shunt.
Incidentally I'm not finding a power distribution diagram in the service
manual. Looking at schematic <10>, though, you'll see that the peak
detectors and the CCD arrays run on the +-5V linear regulated supplies, so
that'd explain why one or both of those is getting dragged low when the
sampling speed is jacked up.


Vince Vielhaber
 

Ok, Q879 collector is solid. When things go sour, it stays. The base voltage is moving, as is the emitter. So it's being told to compensate for the change. However, under normal conditions, the collector voltage is lower by a few hundred mv than the schematic calls for while the base and emitter are within a few mv. Under normal conditions there's a 300mv drop across the shunt.

Going to take a closer look at Q879's feed.

Vince.

On 09/01/2020 10:03 AM, Siggi wrote:
On Mon, Aug 31, 2020 at 8:36 PM Vince Vielhaber <vev@...> wrote:

The 5V jumping around is the 5V regulated that C494 is on. There it
drops a volt or so when I switch below 200ns. According to my Fluke,
when it goes out completely it holds at 3.93v.
Oh, interesting. I assumed you were talking about the 5VD, which is the
main switch-mode regulated supply that feeds all the digital logic.

You're talking about the linear regulated 5V supply. Same deal though, go
back to the first decoupling cap, see whether the supply is jaggy and
whether the raw supply holds up when the regulated supply drops. Most
likely - as others have noted - this is due to geriatric bulk caps in the
secondary, so check that first.
If the raw supply is holding up OK when the regulated supply drops, then
you move downstream and see what's up. This supply is current limited, so
you can infer the current it's using by measuring across the shunt.


Harvey White
 

Assuming a VBE drop across the sensing resistor, then you're drawing roughly half an amp.

Harvey

On 9/1/2020 8:15 PM, Vince Vielhaber wrote:
Ok, Q879 collector is solid.  When things go sour, it stays.  The base voltage is moving, as is the emitter.   So it's being told to compensate for the change.  However, under normal conditions, the collector voltage is lower by a few hundred mv than the schematic calls for while the base and emitter are within a few mv.  Under normal conditions there's a 300mv drop across the shunt.

Going to take a closer look at Q879's feed.

Vince.


On 09/01/2020 10:03 AM, Siggi wrote:
On Mon, Aug 31, 2020 at 8:36 PM Vince Vielhaber <vev@...> wrote:

The 5V jumping around is the 5V regulated that C494 is on.   There it
drops a volt or so when I switch below 200ns.  According to my Fluke,
when it goes out completely it holds at 3.93v.
Oh, interesting. I assumed you were talking about the 5VD, which is the
main switch-mode regulated supply that feeds all the digital logic.

You're talking about the linear regulated 5V supply. Same deal though, go
back to the first decoupling cap, see whether the supply is jaggy and
whether the raw supply holds up when the regulated supply drops. Most
likely - as others have noted - this is due to geriatric bulk caps in the
secondary, so check that first.
If the raw supply is holding up OK when the regulated supply drops, then
you move downstream and see what's up. This supply is current limited, so
you can infer the current it's using by measuring across the shunt.



Siggi
 

On Tue, Sep 1, 2020 at 8:15 PM Vince Vielhaber <vev@...> wrote:

Ok, Q879 collector is solid. When things go sour, it stays. The base
voltage is moving, as is the emitter. So it's being told to compensate
for the change. However, under normal conditions, the collector voltage
is lower by a few hundred mv than the schematic calls for while the base
and emitter are within a few mv.

How are you measuring the collector voltage?
The switching frequency is ~25kHz, and it's going to be ugly as heck if the
capacitor is out, so the instrument you're using to measure this and its
bandwidth are quite important. The average or RMS voltage there just isn't
super important :).
What happens if you tack on extra capacitance from Q879 collector to
ground? You probably want a capacitor that's fairly low ESR, the ripple
current is going to be something wicked if the average current is 3A (see
below).


Under normal conditions there's a
300mv drop across the shunt.
That looks sane according to the voltages annotated on the schematic.
That's a 3A nominal load, though.
U870A/B are the current/voltage controls, respectively, or-ed together
through the diodes. You can tell which one is in control by the voltage out
of pin 1/7.


Going to take a closer look at Q879's feed.
Have fun!


Vince Vielhaber
 

Both the voltmeter and the 2430A. There's not much garbage on it but I did hang another cap on it and it made no difference.

Here is a shot of both sides of R873 on the 2430A. At power off, both traces are together. The top one is the at the junction of R873, R876, Q879 E. The lower one is on the C494 side. This was caught just when the sweep was turned to 100ns/div. I quickly hit Save when it happened.

http://www.hamradio.fun/pix/R873.jpg

Thanks!
Vince.

On 09/01/2020 08:55 PM, Siggi wrote:
On Tue, Sep 1, 2020 at 8:15 PM Vince Vielhaber <vev@...> wrote:

Ok, Q879 collector is solid. When things go sour, it stays. The base
voltage is moving, as is the emitter. So it's being told to compensate
for the change. However, under normal conditions, the collector voltage
is lower by a few hundred mv than the schematic calls for while the base
and emitter are within a few mv.

How are you measuring the collector voltage?
The switching frequency is ~25kHz, and it's going to be ugly as heck if the
capacitor is out, so the instrument you're using to measure this and its
bandwidth are quite important. The average or RMS voltage there just isn't
super important :).
What happens if you tack on extra capacitance from Q879 collector to
ground? You probably want a capacitor that's fairly low ESR, the ripple
current is going to be something wicked if the average current is 3A (see
below).


Under normal conditions there's a
300mv drop across the shunt.
That looks sane according to the voltages annotated on the schematic.
That's a 3A nominal load, though.
U870A/B are the current/voltage controls, respectively, or-ed together
through the diodes. You can tell which one is in control by the voltage out
of pin 1/7.


Going to take a closer look at Q879's feed.
Have fun!


Siggi
 

On Tue, Sep 1, 2020 at 10:06 PM Vince Vielhaber <vev@...> wrote:

Both the voltmeter and the 2430A. There's not much garbage on it but I
did hang another cap on it and it made no difference.
Here is a shot of both sides of R873 on the 2430A. At power off, both
traces are together. The top one is the at the junction of R873, R876,
Q879 E. The lower one is on the C494 side. This was caught just when
the sweep was turned to 100ns/div. I quickly hit Save when it happened.

http://www.hamradio.fun/pix/R873.jpg

I guess next up is to capture U870 pins 1 & 7, see what's up there when
your rail collapses.
On one hand, you could be hitting the current limit, which could be because
of actual overcurrent, or because the resistor network on U870A pin 2 has
drifted. Also the -5V rail could be compromised, as the current limiter is
referenced to that. Also check all the other rails, as the +5 needs the +8
to be up and sound as that's where Q870 draws from.

On the other hand it could be that Q870/Q879 have degraded and can no
longer do the lifting required.


Vince Vielhaber
 

I had actually done that last nite, but didn't have the parts layout handy to see which probe was on which pin (I had them on the diodes>.

Here they are with voltage measurements:

http://www.hamradio.fun/pix/2440.html

Thanks!
Vince.

On 09/02/2020 01:25 PM, Siggi wrote:
On Tue, Sep 1, 2020 at 10:06 PM Vince Vielhaber <vev@...> wrote:

Both the voltmeter and the 2430A. There's not much garbage on it but I
did hang another cap on it and it made no difference.
Here is a shot of both sides of R873 on the 2430A. At power off, both
traces are together. The top one is the at the junction of R873, R876,
Q879 E. The lower one is on the C494 side. This was caught just when
the sweep was turned to 100ns/div. I quickly hit Save when it happened.

http://www.hamradio.fun/pix/R873.jpg

I guess next up is to capture U870 pins 1 & 7, see what's up there when
your rail collapses.
On one hand, you could be hitting the current limit, which could be because
of actual overcurrent, or because the resistor network on U870A pin 2 has
drifted. Also the -5V rail could be compromised, as the current limiter is
referenced to that. Also check all the other rails, as the +5 needs the +8
to be up and sound as that's where Q870 draws from.

On the other hand it could be that Q870/Q879 have degraded and can no
longer do the lifting required.