开云体育

I wasn't thinking things through completely. After I figured out I could unscrew the nylon (that's what we'll call it for now) coupler I saw that the shaft had the nylon connection between the pot shaft and the knob shaft. It took me a bit to realize the design is isolating the high voltage focus pot from the low voltage intensity pot. But once that light-bulb went off I realized I might be able to extract the knob shaft and replace it. It wasn't a case of it shifting - it was sheared off at the end of the outer intensity shaft. Unfortunately when I tried removing the focus knob shaft it snapped of just inside the nylon connector.

Rather than fight with it I'm getting an earlier model 100k/5M pot - earlier serial number models used this. I'm going to scavenge the wiper and shaft (hopefully) from the 100k/5M Clarostat and use it in the 100k/6M pot. Or some other rearrangement of parts. I'll have to see once I have parts in hand.

Some carbon composition resisters have a humidity absorption issue where they can change in value. This depends greatly on the outer casement material.

Zen

Off topic - speaking of resistors out of tolerance. I was working on a TM500 module recently, and there was a resistor that measured high. About 20% or so as I remember.
I had about 5 resistors in my bin (new resistors btw) that were the exact value, 1% tolerance. And 4 of them also measured really high, like 10-15% high. The last one was right on the money.
Crazy coincidence that they were similar to the "bad" one. Of course I suspected a measurement error. Obviously my meter / probes were adding a lot of R to the measurement, but no - checked other values and they were what they should be.
What are the odds that I would have 4/5 bad (new) resistors? Okay, we all know that "new" means "been in my resistor drawer since I was 14 years old" right?

I have restored one of these. The transistor types are not too critical, you could use just about any old Ge small signal transistors of the right polarity if you can find them. Maybe even silicon. I had to find a substitute for the unobtainable thermistor - now THAT was a challenge!

Morris

They're simple 78xx or 79xx regulators.? Do please ask. Datasheets available (at least on the one I have).


Harvey

Right here seems

Looking for a place to ask a repair question for TEK 1103 Scope Probe Power Supply

So the collector of Q386 was running a little less negative than in the working unit. That's a diode drop below the gate of Q392 (confirmed), which seemed relevant. It's also a diode drop below the low end of 10V zener VR296 (confirmed). The other end of the zener is at ground (confirmed), so that voltage is determined by the zener and it's wrong so the zener is bad. I don't have a 10V zener but I have some TL431s. Solved!

Before getting too deep into crafting an artisanal 10V zener (having visions of an SOT23 and a couple of 0603 resistors on a tiny chip of perfboard and sugar plum fairies), I lifted one end of the bad zener to confirm the theory by wiring a AA cell in series (via test leads hanging out the empty middle bay) to compensate for the ~1.5V deviation. ... No joy. Same double trace; same voltages. ... Ok, maybe inlining the battery was a faulty concept. So I breadboarded a TL431 (TO92!) and trimmer pot and dialed in 10V to make the more valid test. No joy. So I pulled the bad zener and measured it separately. 10 Volts. So I put it back. It appeared that the task required more understanding. Back to the schematic...

Since the sources of Q382 & Q392 can't go more than a diode drop below ground, these -10ish V levels from Q386 should be pulling the gates and sources of Q382 & Q392 apart to disconnect ch2. That was working for Q382. And working for Q392 in dual trace mode. The diodes CR383 & CR393 could differ, but the forward voltages of the diodes and so gate voltages of the ch2 FETs were about the same -- as were their source voltages and so their Vgs. When showing ch1 as two traces, both traces were shifted from "correct" display, suggesting Q392 was conducting (not isolating ch2) to two different degrees i.e. that the two Q392 gate voltage levels in ch1 mode were both in the FET's linear region rather than on or off. The two levels were about 1V apart, and about 1.5V less negative than the working example, so the more negative level was only about 1/2V less negative than the less negative working example. And Q392 did turn off in dual trace mode when its gate couldn't be going very much more negative. I don't suppose Tek would have had any reason to select FET pairs for matched Vgs cutoff levels, so I supposed it was at least plausible that the Vgs cutoff levels were different enough between Q382 and Q392, within normal variation, for good parts to operate differently at these somewhat less negative than intended Vgs levels in effect when displaying ch1 only. That bit of figuring didn't achieve any progress, just a little more understanding of what was/was supposed to be.

So Q286 and channel 1 appeared to be ok. And the voltages at the base and emitter of Q386 appeared to be ok. But Q386 wasn't pulling it's collector quite low enough -- when conditions at its emitter and base matched to good example. In either the working or broken unit the different collector levels were a handful of volts different from the emitters, so Q386 operates in its linear region (vs on/off) in this display mode, so something else is in the loop determining the collector voltage.

R386 connects Q386 collector to +5V, which ties Q386 collector voltage to current through R386. Current into Q386 from ch2 includes current through R382 & R392. When the zener is regulating collector voltage, which is supposed to be the case, current into Q386 also includes current from the zener via CR386. Some ideally small current also comes into the base. All current out of Q386 goes through R298 to -30V, which ties Q398 emitter current to voltage across R298. Using Tek's indicated voltages for the test condition (ch1 only, inputs grounded), that's ... close enough to need a calculator. V(R298=2.2kΩ)=13.7V giving 6.2mA out of Q386. V(R386=15kΩ)=15.6V and V(R382=R392=3.65kΩ)=9.3V giving 1.04+2×2.55=6.1mA into Q386 from channel 2 switching, leaving just 0.1mA for the zener to make up. Neglecting base current. In other words: if everything else is right except Q386 has weak β this fails. Or if some bit of "everything else" drifts the wrong way this fails. Q386 is one thing vs. "everything else", so I pulled that and stuck it in the transistor tester.

β=300.

To be continued...

My question is whether the RU800 would work better in the 2445 than in the 2465. I understand that there is some bandwidth limitation with the RU800, and that might not be a problem when the scope has half the bandwidth.

—Jeff Dutky

I'd fix up the 2465A first. Put U800 on a socket if you have to replace it. It will make putting in an RU800 so much easier if necessary.
FWIW I installed the first generation RU800 in my 2465 and it works quite well.

Hi Dave, If it is the one I had sold him, then you're braver man than I thats for sure. But i do still have some notes on my repairs from when i was fixing it feel free to contact me if you'd like to know the work i carried out.

So the displayed voltage of channel 1 was shifting with trace alternation.

The rest of this will make more sense with reference to "output amplifiers <2>" schematic. See page 51 in this PDF on TekWiki:

I don't have a plugin extender. Of course the output stuff is on the right side of the plugin, not accessible through the left side of the scope frame. So diagnosis continues by pulling out the plugin, clipping leads onto points of interest and replacing the plugin in the left slot with leads hanging out from the empty right slot. Repeat.

I set the ch1 inputs to GND and ran a sweep slow enough to see the alternation.

Of course output voltage between the drains of Q282/Q292 changed with alternating sweeps. The drain of Q282 was steady and Q292 alternated. So that narrowed it down a bit. The emitters of Q242 & Q252 were both steady 19.4V, as indicated on the schematic. The difference between collectors of Q260 & Q270 alternated with Q260 steady and Q270 alternating. Q242 & Q252 drive the bases of Q260 & Q270, so at Q270 the base is right and the collector is wrong.

FETs Q282/Q292 handle channel switching by connecting/disconnecting the ch1 outputs to/from the mainframe interface board. They were clearly disconnecting ok because ch1 was off when it was supposed to be off. When "on", the source and drain of Q292 tracked each other, so it was connecting ok when ch1 should be connected. When displaying ch1, diodes CR283 & CR293 are back biased, which should isolate the ch1 signal from any influence by the channel switching logic. Puzzle.

Oh yeah... remembered the mainframe shorts the two channels together, connecting the drains of Q292 & Q392 and bringing ch2 into the picture.

Which should have no effect because when ch1 is on, Q_3_82 & Q_3_92 are off and, apart from bringing R394 into parallel with R294 that should do nothing to ch1-. And Q382 (ch2+) was switching off and the same signal switches Q382 & Q392. And Q392 responds correctly to that signal in dual trace mode. Puzzle.

It wasn't clear how the channel switching circuit could produce this result. Reviewing assumptions... was this really synchronous with ALT/CHOP logic? Seemed pretty clearly so. Could that timing come from anywhere but the timebase? The CHOP clock comes from the interface board, but the timebase clearly has control over CHOP vs ALT and the ALT timing can't come from anywhere but sweep timing. How could CHOP/ALT timing affect the plugin? The B21 CHAN SWITCH SIGNAL line, of course. Anything else? That bit of cirtrickery also connects to A21 CHAN SWITCH SUPPLY, which ... the interface board connects directly to -30V for both vertical slots. So something shifting the -30V supply might have some effect, but it seemed like that would affect a lot more than the -signal out from one plugin and whatever it might be doing isn't happening when ch2 is enabled. Could the CHAN SWITCH signal be doing anything different with/without ch2? Does CHAN SWITCH normally chop/alternate when not needed? The interface board schematic does look like CHAN SWITCH always either CHOPs or ALTs regardless of vertical plugin display mode so the ch1/2/dual state shouldn't influence the incoming switching signal. Other connections include signal, power, ground, trigger, readout. I have an OPT1 scope sans readout so didn't give that much thought. I'm already watching the signal lines. Can't see how anything coming back through the trigger could produce that result. Normalcy in all other respects seemed to point away from ground or any other supply connections.

Well, I have another 5A26. I picked up this second unit because I have a "someday" project in mind for four differential inputs. That unit worked normally, further pointing away from anything outside the plugin. And provided a baseline for comparison.

After some more puzzling over the schematic I clipped five leads to the collectors and bases of Q286 & Q386 and their connected emitters on the good 5A26. The schematic shows expected voltages for these points when displaying ch1 only with grounded inputs, which is what I was already doing. The collector and base of Q286 were steady and close to the indicated voltages. The three terminals of Q386 alternated up & down by about a volt between values bracketing the indicated values. I took that as close enough to the indicated values to work with, leaving the question of why Tek provided "DC" voltages without indicating that they weren't steady. In retrospect, I suspect their test case was either chopping or sweeping rapidly enough for a voltmeter to filter away the oscillation while I was instead running slow sweeps, which one would ordinarily CHOP, in ALT mode to artificially slow the switching to human observable rate. Anyhow, that gave me a working baseline for the channel switching logic, which is a little bit subtle in that it depends on a handful of resistor dividers calibrated to about 3%, which might be worth a paragraph.

At the same test points, the wonky unit showed the same conditions at all points except the collector of Q386, which oscillated between levels about 1.5 volts less negative than the good unit. That seemed like progress.

To be continued...

Rebonjour! Update:

Cleaned a bit, can use a standard 9V alkaline batt with snaps.

Output is IN CAL for freq and 1V at jack.

Needed an adapter, 1.125" to 0.500" reducer to fit my GR mics into the calibrated cavity, they were in the originlas kits but super rare.


I kluged a PVC plastic pipe adapter and used felt gaskets...Voila! ......

With the three 3 GenRad meters, ONE is in CAL at 114 db, the other read a bit low. As all are from 1970s, this is remarkable.

All it needs is some careful corrosion cleaning, new battery snap and replacing the corroded transistors, 2N1304, 2N1305.

For the $25 on ebay I am a happy guy!

Bon journee,

Jon

Dave,
I think that the concentric pot assembly has nylon spacer joining the two pots. Within that spacer is an insulated shaft. There is a nylon section between two metal shafts. It is possible that the metal portion that extends from the intermediate coupler is missing. You might be able to fit a replacement metal tip to the remaining nylon shaft coupling.
The nylon spacer that joins the two pots is threaded. It is tight but it can be unscrewed.
As for the mundane switch, it is not common in my junk pile because it is a NC rather than NO configuration. The narrow UNS #8 mount is also less common. It’s not so mundane after all.

Output clip and integral clip may be what you need to add to your code...

I found this tutorial very helpful...

Now I'm not sure.? I'll have to go look and find out.? Was using a TDS540A, which should go to 500 Mhz.? I can switch to an active probe and see what goes on.

Back after a bit.

Thanks.

Harvey

Steve many thanks very long time!

Many thanks for the fast advise! Built 1972..1978 at least.

Found three of my GenRad sound level meters to check it . Must fab a 1.125" to 0.500" aluminum cylinder to adapt the transducer.

1/ ERROR! Battery 9 V CARBON-ZINC not mercury! #1600 1" dia ~ 1.5" Wow, no Hg!

2/ Corrosion most at two Transistors, battery - snap, edge of outer frame, fortunately the Shure transducer seems OK.

3/ Transistors 2N1305,5 corroded but IT STILL WORKS! Seeking replacements TO-5 metal can NPN, PNP Geranium original were Texas Instruments

Will proceed with cleanup and report later.


Bon Journee,
Jon

Rogerio,

R110/1/2/37 are stressed at 1/2W. Mouser has MFR200 types in stock, if you do not have 1 or 2W in stock. The 2W Mouser has is cheaper than the 1W types and the additional power capacity will not hurt anything. The 127,000 ohm resistors mentioned by Dan are suspect. A 100,000 ohm and 27,000 ohm 1/2W in series will get you the resistance. They will need to be 1% as the original is 1%. See about the 47,000 ohm resistors in the emitter circuits. You may have to unsolder some resistor leads to check. Because a resistor looks good does not mean it is in spec. Some can check good in circuit and be too high in resistance when checked with one end out of circuit. Dan is right about the filters possibly being bad. Zen is right about the switches being dirty and being in need of being cleaned! Those will cause a lot of problems.

Mark

On Wed, Jan 12, 2022 at 06:07 PM, Harvey White wrote:

...
On the 7D14, I measured the input up to the tunnel diode, and that's
fine.? I get the waveform that I'm supposed to get (I think) after the
tunnel diode, which is the sine wave with glitches (I used a sine
input).?

Perhaps input frequency is too high for your scope to show the sharp edges but otherwise sine wave is not expected at U329 inputs, it should look like a square wave. Do you see the correct waveform at prior points, e.g. waveform 13?
Ozan

Rogerio,
Beamfinder switch is a sore spot in a LOT of tek gear. Clean and lubricate if you even suspect a problem with it. On the 576 the "cal zero and invert" switch pack is a realy soar spot and caluses lots of bizarre issues with the trace.

Zen