开云体育

When the AC/DC/GND switch is in the AC position, the signal is the same, with good risetime (using the BNC soldered directly at the source follower). In the GND position the signal switched to approx 50% amplitude but still with good risetime (is because my 50 ohm input termination is paralleled with the R104, 56 ohm).

Oops... at the end of last paragraph commited a typo: please read "shunted" instead of "paralleled"
The " B" contacts of the AC/DC/GND selector are inaccessible.
So i'll proceed to dismantling all...

I'm totally baffled.
I've dismantled and cleaned the attenuator's pcb with IPA (several hours of work) then remounted all the assembly carefully: signal is the same as before.
Changed Q150 with another one NOS. Changed CR130 with another one NOS. The displayed waveform is still the same, mocking....

I ran out of all possible tests.
What do you advise me to do?
I'm very discouraged.

Max

On Fri, May 20, 2022 at 02:53 PM, unclebanjoman wrote:

I'm totally baffled.
I've dismantled and cleaned the attenuator's pcb with IPA (several hours of
work) then remounted all the assembly carefully: signal is the same as before.
Changed Q150 with another one NOS. Changed CR130 with another one NOS. The
displayed waveform is still the same, mocking....

This is a real puzzle. In the pictures there is a metal connection between attenuator and the amplifier, roughly between CR130 and CR507 (CR507 shows up in board layout but not in schematic). Probably a ground wire. Could that connection have a cold solder and give poor grounding?

Did you replace C140 and C158 (10uF caps)? Although 1nF caps are doing the filtering at fast edges perhaps there is some influence of C140 and C158.

Perhaps something is faulty at amplifier input and impedance seen by Q150 is giving you a distorted waveform. Do you see expected voltages at pins of U350? Does removing U350 clean up the waveform at right side of R10 (where you used to see the half step)?

It is not related to what you are seeing but I find schematic around pin 13 of U350 confusing. Drawing looks like C190 was replaced with a short circuit but a short circuit would not work for DC biases and voltages you measured also suggest open circuit. Pins 13 and 16 are connected together by ~ 4-ohms internally.

Ozan

Hi, Ozan,

Just finished another trip with the 7A16A.
Here are the responses to your answers and observations.
1) C190 was discontinued starting from serial numbers B079999-up. Mine has S/N B138015. In its place there's simply a jumper wire connecting the ground between the attenuator/input follower board and the main board and it's well soldered.
2) I Put two 22 uF in parallel to C140 and C158: no changes.
3) Removing U350 gives me a slight better waveform at pin 14, but still out of specs (and the half step is still perceptible): /g/TekScopes/photo/262075/3433488
Remember that I'm using a 500 ohm 10:1 probe together with my Philips 3340 scope. I don't have no other probes to test such fast signals.

In my version of the board pin 1 and pin 13 of U350 simply has not connection. There are two nude holes instead of the socket IC pins.
As Raymond pointed out, the DC voltages I wrote on the schematic (you can see it at /g/TekScopes/photo/262075/3430589) seems reasonably.

Let me know if you want other photos and /or waveforms or other tests.
I appreciate very much your effort to help me in this puzzling situation!

Max

I noticed another thing while measuring the signals using the plugin's BNC as input.
During the tests R10 broke down at the BNC. To not waste time, I directly connected the BNC pin to the actuator board.
To my surprise the signal changed totally, with an overshoot of 20% and a damped oscillation of a few cycles. The waveform amplitude also was greater, approx 20%
Soldering a new 56 ohm resistor the signal returned as usual (always lousy).
It seems to me that the whole assembly of the attenuator board has been carefully calibrated by the Tek designers. But looks strange to me that a simple 56 ohm 5% resistor can have a such grat influence on the signal!

Max

It seems that I have found the culprit.

Another 7A16A arrived today. It had marked overshoots and oscillations which were not recoverable with trimming. I have located C756 open. Changed it, the plugin started working decently.

Thinking about my original problem, without too much conviction I tried to change C756 with a new one even on the faulty plugin.
To my surprise it started to work somewhat better.

After a bit of tweaking I managed to get a waveform practically similar to the expected one, when I trimmed it a year ago.
C756 did not seem broken because turning it had a more than visible effect.
So I carefully measured it with an RLC meter (HP4275A): it showed a capacitance variation between 1 pF and 8 pF instead of the required 3-15. Not in specs, not open but partially functional.

Moral of the story: a small piece created big problems, leading me to look for the fault where it was not there.

The most important thing is that all the various trim points also seem to be made to recover any rise-time problems in the input follower stage. Basically it's all a team game between the many calibration points.

Comparing the signals between the plugin arrived today and the (ex) faulty one, the waveform in the input follower shows practically the same.
This would mean that the signal in some manner is cleaned up and compensated and enhanced in the transitions between successive stages.
This is my opinion. Yours are welcome.

I thank everyone who took their time to help me!

Tomorrow I'll finish reassembling all the pieces of the (tortured) plugin and I will try it inserted directly in 7854.
Since with the flex entender it reaches about 0.85 ns of rise-time, surely the waveform will be better with the plugin plugged directly.

I'll give you confirmation tomorrow.
Now I rest listening to some jazz because this whole story has stressed me a lot.

Max

On Sun, May 22, 2022 at 01:16 AM, unclebanjoman wrote:

C756 did not seem broken because turning it had a more than visible effect.
So I carefully measured it with an RLC meter (HP4275A): it showed a
capacitance variation between 1 pF and 8 pF instead of the required 3-15. Not
in specs, not open but partially functional.

Congratulations, Max, that's the reward of not giving in!
Lots of information for many of us during your quest, thanks for that.
Normally, I'd expect to notice if a trimmer works but its range of adjustment isn't enough. I guess that wasn't clear because of the many variables (pun intended).

Raymond