Dear experts,
One of my 7A26's distorts a 1 ns falling edge (PG506, followed by 50 Ohm attenuator) as shown in this photo:

https://groups.io/g/TekScopes/album?id=255438

Both channels show the same. I've looked for what I considered obvious problems like the Display Switching area (7A26 SM dwg 4) but haven't found the cause yet. Other 7A26's show the edge fine.
Any ideas?

Thanks,
Raymond

On Wed, Oct 21, 2020 at 04:20 AM, Raymond Domp Frank wrote:

I've looked for what I considered obvious problems like the Display Switching
area

The fact that the circuit is current driven makes following the signal a bit inconvenient...

Raymond

Is the rising edge OK?
--John Gord

Raymond,
The 7A26 has a number of small ceramic trim caps of a design that has proved to be prone to going open circuit. You may have failures at the same point of both channels, unlikely as that seems. I would suspect C1432 and C2432 in particular, since they seem likely to affect the highest frequency signals. Try turning them: if they have gone open circuit turning them will have little to no effect. If those are not the problem, try others!
--John Gord

Raymond,

The picture suggests that the signal might be switching from the full bandwidth to the 20MHz version at around the 40% level. Have you looked at the biassing of Q860, Q880 and also for leakage in Q860, Q880, CR860 CR880? It seems to me that the problem must be related to the semiconductors since a purely linear circuit should affect positive and negative edges equally.

Regards,

Roger

Raymond,

A simple test - lift one end of each of CR860, CR880 and see if the signal behaves as it should.

Roger

On Wed, Oct 21, 2020 at 06:27 AM, John Gord wrote:

Is the rising edge OK?

Hi John,
Both edges show the same behavior.

Raymond

On Wed, Oct 21, 2020 at 06:55 AM, John Gord wrote:

The 7A26 has a number of small ceramic trim caps of a design that has proved
to be prone to going open circuit. You may have failures at the same point of
both channels, unlikely as that seems. I would suspect C1432 and C2432 in
particular, since they seem likely to affect the highest frequency signals.
Try turning them: if they have gone open circuit turning them will have little
to no effect. If those are not the problem, try others!

Hi John,
I know the problem with the trimmer caps. I found several that had no effect, initially blaming that on the lack of HF content in the signal as it appears and not expecting a multi-device failure.
I temporarily put some small caps (3 - 10 pF) in parallel with one trimmer-cap at a time and so far, I've found 5 (yes, five) that immediately made the pulse steeper, without overdoing it, in both channels!
I'm afraid this is going to exhaust my donor supply...

Raymond

On Wed, Oct 21, 2020 at 07:52 AM, Roger Evans wrote:

The picture suggests that the signal might be switching from the full
bandwidth to the 20MHz version at around the 40% level. Have you looked at the
biassing of Q860, Q880 and also for leakage in Q860, Q880, CR860 CR880? It
seems to me that the problem must be related to the semiconductors since a
purely linear circuit should affect positive and negative edges equally.

Hi Roger,
Since these components constitute most of the common parts, I had done some testing in that area. I checked Vbe across Q820 / Q840, Q860 / Q880 and they were correct. I pulled Q860 and Q880 and lifted one end of CR860. No difference.
Next, I bridged some of the trimmer caps with small fixed caps each at a time and found a significant influence. I'm going to check and replace trimmer caps!

Raymond

On Wed, Oct 21, 2020 at 04:20 AM, Raymond Domp Frank wrote:

One of my 7A26's distorts a 1 ns falling edge (PG506, followed by 50 Ohm
attenuator) as shown in this photo:

https://groups.io/g/TekScopes/album?id=255438

Both channels show the same. I've looked for what I considered obvious
problems like the Display Switching area (7A26 SM dwg 4) but haven't found the
cause yet. Other 7A26's show the edge fine.

This is my newest 7A26 (main PCB code 670-2549-21)!
S/N 709928, so Heerenveen. It's the only one in my collection with a green PCB and 155-0078's with date code "429", I guess that's 8429.

Raymond

On Wed, Oct 21, 2020 at 02:49 PM, Raymond Domp Frank wrote:

I temporarily put some small caps (3 - 10 pF) in parallel with one trimmer-cap
at a time and so far, I've found 5 (yes, five) that immediately made the pulse
steeper, without overdoing it, in both channels!
I'm afraid this is going to exhaust my donor supply...

The problem was indeed caused by defective trimcaps: I started by replacing 4 (four !) open trimcaps in Ch2, made preliminary adjustments and took another picture. See album.

I'd never have thought that practically (?) all trimcaps of the problem-type would be defective in the one unit but then again, it *is* a systemic problem.

Thanks to John and Roger for your help and suggestions!

Raymond

Raymond,

I interpreted your first post to mean that only the falling edge was distorted, that was the reason for thinking it couldn't be only the passive (or more accurately the linear) components at fault.

Anyway it is good to be warned that the caps are so unreliable.

Regards,

Roger

On Thu, Oct 22, 2020 at 09:57 PM, Roger Evans wrote:

I interpreted your first post to mean that only the falling edge was
distorted, that was the reason for thinking it couldn't be only the passive
(or more accurately the linear) components at fault.

Anyway it is good to be warned that the caps are so unreliable.

I later realized that I should have mentioned that both the falling and rising edges were affected equally or only mentioned "edge". In my first fast edge generator (a PG506), the falling edge was a bit cleaner than the rising edge, which resulted in my continued preference for using a falling edge in measurements using a PG506. I didn't realize that this might from time to time cause confusion.
Your post (msg #172512), made clear the significance of the detail / omission.

I ended up finding 12 out of 14 trimcaps in this 7A26 - my most recent - being defective! The remaining two trimcaps have a slightly different design.

Raymond

Raymond,

My PG506 (very early model) has a similar behaviour but it is the edge moving towards ground which is cleaner so the -ve output has a clean 'positive' edge.

Roger