开云体育

Hey folks,

Wondering if there is someone here who is a readout system whisperer. I
have me a 7904 that wasn't working when I got it. It's a rather interesting
one, with all available options equipped. This includes the fast write crt
with P11 phosphor. A bunch of shorted tantalum capacitors replaced later,
and I have a mostly-working scope. I did rob some parts (including the crt
and vertical amp) from a similarly optioned parts unit 7903 I have.
Unfortunately someone burned the original tube very badly. Lucky to have a
drop in replacement!

At any rate, there's a trace and it's even reasonably calibrated fir
vertical gain, as long as the trace doesn't stray too far from center.
Essentially, there seems to be no happy medium between setting the vertical
centering and the readout center on the vertical amp. The readout
essentially is stuck low on the crt...what should be at the top of the
graticule is in the center and squishes up if you move it any higher.
Additionally, the vertical gain drops, the lower down the trace goes. This
all was observed on both tubes so I doubt it's the tube.

So far I've tried curve tracing all the transistors in the network on the
vert amp formed by Q694, Q698, Q710, Q716, Q723, and Q728. They are all
fine. I *also* tested the transistors in the output of the readout X and Y
signals on the readout board. They are also fine. Additionally, I swapped
some ICs from the 7903's readout board. This actually helped a little, but
the readout still squishes up if you try to move it (and the trimmer pot
tops out before it gets anywhere near the top of the graticule anyway). I'm
running out of ideas on what to try, so was wondering if anyone on the list
has encountered similar issues or knows what might cause them.

Thanks!

Sean

Hi Sean,

On the vertical amplifier board, take a look at resistors R657/R659,
R662/R663, R670/R672 and make sure their values match and are approximately
to spec. I think these are carbon composition and can drift high. A
mismatch will push the main trace up or down, so that when you centre the
main trace the auxiliary amplifier runs out of control authority and the
readout can't reach one edge.

Regards,
Nick

Thanks everyone. I think Nick here found the winner. R670 and 672 have
both drifted over 5 ohms on both vertical amps I have. I'll try replacing
those two first.

Sean

On Tue, Jan 26, 2021 at 07:04 AM, Nick Shipman wrote:

Hi Sean,

On the vertical amplifier board, take a look at resistors R657/R659,
R662/R663, R670/R672 and make sure their values match and are

approximately

to spec. I think these are carbon composition and can drift high. A
mismatch will push the main trace up or down, so that when you centre the
main trace the auxiliary amplifier runs out of control authority and the
readout can't reach one edge.

Regards,
Nick

Hi Sean,

A bit of drift shouldn't be a big problem if both resistors in a pair do it
by the same amount — it's when one differs from the other that you'll get
the trace shifting up or down. But with any drift the overall sensitivity
will change to some degree, so you might want to look at the vertical
calibration once you're happy with the trace positioning.

Make sure you use non-inductive resistors for the replacements, try to get
ones of about the same size, and avoid bending anything (I seem to remember
it's all a bit 3-D here). It'd be a shame to spoil that amazing performance.

Regards,
Nick

On Tue, Jan 26, 2021 at 06:34 PM, @0culus wrote:

Thanks everyone. I think Nick here found the winner. R670 and 672 have
both
drifted over 5 ohms on both vertical amps I have. I'll try replacing those
two
first.

Drift of these low-value resistors (R657, R659, R662, R663, R672) in the
vertical amplifier destroys impedance matching and so completely destroys
frequency behavior.
By "drifted over 5 Ohms", do you mean "from their nominal 4.7 Ohm to over 5
Ohm" or "by 5 Ohm", so both to more than 9 Ohm?
Unless you have low-induction replacements, I wouldn't touch them, especially
if the former.
From your description, I wouldn't conclude that drift of R660 and R662, even
to double their intended value, would cause anything like what you're
describing.
Could you create a small video to show what exactly is the behavior?

Raymond

Yeah, it will need a full cal regardless...I changed out the CRT from the 7903 parts unit I have. I do have good quality non-inductive resistors on hand.

Since I have two identical vertical amplifiers thanks to the parts unit, I did the resistor swap on one of them. It did help a little, but it's not the whole issue. Unfortunately, the vertical output IC on the original amplifier that was in the scope is very much bad. And the output IC on the board from the parts unit already had crap high frequency response because it looks like a gorilla bodged together the little output wires that go to the CRT vertical deflection inputs. The ones on the bad output IC are single wires; the working one has about three sloppily soldered together...which cannot be a good impedance match at all.

I hate to say it, but I may need another vertical amplifier board with known-good ICs altogether.

Sean

On Tue, Jan 26, 2021 at 10:41 AM, Nick Shipman wrote:

Hi Sean,

A bit of drift shouldn't be a big problem if both resistors in a pair do it
by the same amount — it's when one differs from the other that you'll get
the trace shifting up or down. But with any drift the overall sensitivity
will change to some degree, so you might want to look at the vertical
calibration once you're happy with the trace positioning.

Make sure you use non-inductive resistors for the replacements, try to get
ones of about the same size, and avoid bending anything (I seem to remember
it's all a bit 3-D here). It'd be a shame to spoil that amazing performance.

Regards,
Nick

If you can find a junker PG502, that's another potential source of that
vertical amplifier.

--Tom

Tom,

I wonder if that's where this one came from, the one with the bodged leads. Since I'm guessing in the pulse generator application, it's not driving a crt?

Sean

On Tue, Jan 26, 2021 at 05:00 PM, Tom Lee wrote:

If you can find a junker PG502, that's another potential source of that
vertical amplifier.

--Tom

Hi Sean,

With instruments of uncertain provenance, it's hard to say what a previous
owner might have been thinking (or failing to think), so anything's possible.
It's hard for me to understand why they would touch those leads at all, but
I've seen so many weird mods and bodges that I've sort of stopped asking
questions. :) One assumes that what they were doing made sense to them at the
time, but why they thought so is often elusive...

In both the PG502 and the 7904, the vert output amp's pins just go straight
into socket pins, and drive nominally resistive loads. The 7904's vertical
deflection plates are actually a transmission line, and are terminated
resistively (with a parallel RL damping network to tame resonances created by
lead inductance), so present a broadband resistance to the amp. The PG502
similarly terminates the IC in a purely resistive load, so the two instruments
aren't using the IC dramatically differently.

-- Cheers,
Tom

That's a good point. At any rate, I got these instruments for free, so I'm not complaining too much, especially if I can resurrect this 7904. :o)

I took the working output IC (with the bodges) and re-bent the output legs to match the burned out one (which isn't bodged...essentially the output leads are super long "legs" compared to the others on the IC). This _massively_ improved the high frequency response, so it appears geometry of these leads is more important. So much so that testing the frequency response using a known-good signal standardizer fixture + RF signal generator to make the CW shows nominal 3 dB point. But, one problem at a time!

On the amp that I did the resistor change on, I do have a little more range in the centering controls. Unfortunately, the RO center is basically cranked all the way and the top readout line is squished up.

It's also worth nothing at this time two further gremlins (that may or may not be interrelated): (1) the units shown for the time base setting is wrong in a couple places in the "A" time base (showing ms instead of us, when it clearly should be us) and (2) the "A" time base will not trigger internally, but it does work with an external trigger.

Sean

On Tue, Jan 26, 2021 at 06:01 PM, Tom Lee wrote:

Hi Sean,

With instruments of uncertain provenance, it's hard to say what a previous
owner might have been thinking (or failing to think), so anything's possible.
It's hard for me to understand why they would touch those leads at all, but
I've seen so many weird mods and bodges that I've sort of stopped asking
questions. :) One assumes that what they were doing made sense to them at the
time, but why they thought so is often elusive...

In both the PG502 and the 7904, the vert output amp's pins just go straight
into socket pins, and drive nominally resistive loads. The 7904's vertical
deflection plates are actually a transmission line, and are terminated
resistively (with a parallel RL damping network to tame resonances created by
lead inductance), so present a broadband resistance to the amp. The PG502
similarly terminates the IC in a purely resistive load, so the two instruments
aren't using the IC dramatically differently.

-- Cheers,
Tom

Hi Sean,

In the PG502, all leads of that IC are the same length, I believe (I will
check this eventually). That might explain what you saw -- the previous user
apparently tried to add back the lost length when doing a transplant to the
7904. Very interesting. Roughly what is that extra length? I'm just curious.
The inductance of that lead is apparently part of a peaking network.

Tom

Hi Tom,

The leads on the output IC that is bad are all normal length, except for the two outputs. Those I would estimate between 1.5-2 inches long if you straightened them out of the "L" shape they make and include the length of the socket that goes over the pin in the crt. The working IC which has patched leads is about the same length, so if nothing else they were at least careful about that. The output pins are short like the others; they soldered the extensions directly to them.

Sean

On Tue, Jan 26, 2021 at 08:09 PM, Tom Lee wrote:

Hi Sean,

In the PG502, all leads of that IC are the same length, I believe (I will
check this eventually). That might explain what you saw -- the previous user
apparently tried to add back the lost length when doing a transplant to the
7904. Very interesting. Roughly what is that extra length? I'm just curious.
The inductance of that lead is apparently part of a peaking network.

Tom

So at least they gave it a solid try!

That's a substantial lead inductance -- perhaps of the order of 30nH or so,
give or take. Very interesting.

Thanks for the info, Sean.

--Tom