Cossor CRO 3100 Timebase unit repair (model #3122)

[golexica]

First time poster.

I recently acquired a cossor 3100 oscilloscope, and it needs some tlc. I came across this forum thread from 2008 https://www.vintage-radio.net/forum/...ad.php?t=25919

User Chris55000 posted most of the manual online, for which I am eternally grateful. Unfortunately, the problem seems to be related to the plug in dual time base unit , that Chris did not post/have the manual for.

i suspect the timebase unit because : the trace won't sweep, vertical deflection seems good on testing, I can still adjust horizontally with dials, but internal timebases are heavily biased to the side of the screen, if i connect to external sweep/x amp, i can get a lissajous).

It's marked model no 3122, and has 3 internal pcbs marked A, B, C.
B, C seem fine, but A has a very sad resistor on it - brown black black...black- shattered, and crud on the back side that suggests tracking. Based on the other values on the board, it's probably either a 10ohm or 100ohm, as all other resistors of that beginning brown are either a 10 or a 100, but which, i don't know.
Testing other diodes and such, and it seems they're all intact.

User Peter Cole apparently designed the thing back in the seventies and has the manual, but they haven't posted since 2013.

If possible, i'd like to get hold of the manual, - it's a rat's nest of wires, and there are some things that look like tunnel diodes, which would be basically impossible to replace if they broke.

If I can't I see only 2 real options - clearing the tracking and trying first a 100ohm, then a 10ohm if the 100 doesn't work.

Alternatively, board A and B have a lot in common, but with B being quite a bit more complicated, with trim poots and the like.

Suggesting perhaps that A and B are each responsible for a timebase each, probably A the main, and B the delayed based on the front panel, and the complexity of the boards I don't know whether perhaps it would be wise to just remove A completely and see if that resolves the fault, and whether that could damage the rest? I doubt it, but the tunnel diodes are making me jumpy.

Advice would be appreciated.

P.S. Sorry about my location, I forgot to include my County (Middlesex) along with the city London, and the forum won't let me change it for some reason?

[WME_bill]

Cossor timebase 3122 for the 3100 (CDU160) main frame.
Peter Cole many years ago kindly lent me his original manuals to scan for the 3100 main frame, 3122 dual TB and 3102 Yamp. 35Mhz.
In production Cossor uprated the bandwidth to 50Mhz. I don't have any information on this. Probably a change of transistors and some more feedback.
To a designer used to valve techniques, transistors could often exceed the expected performance, so you often see uprated Y amps of 1972 designs.

Your board A is the main timebase, with two AEY25 tunnel diodes (5ma, same as 1N3716 or 1N3717 or Russian).
Board B is the delayed TB, also with two AEY25. My board B has been modified sometime, and has only has one tunnel diode.
Board C is X amp, TriggerAmp, Blanking etc.
On Board A, almost all the power lines each have a 10ohm 1/8w resistor in series:
point 12=+25v,
point 8=+10v, also has a tantalum capacitor C12 22u fed from R62 10 ohm, and C25 22u fed from R71 1kohm.
point 18=-25v ,
point 42=120v.
So try 10ohm, and if it is R62 in the middle of the board, check C12 beside it and C25 near the top with R71. Both are tantalum which are well known to short readily.
I attach TB circuit. I have posted advice on tunnel diode testing previously. Do not test with an ohmmeter, it will destroy it.
wme_bill

[golexica]

Thank you so much , R72, a 100ohmer, capacitor on that line seems to be ceramic, but i'll be sure to test the others just in case- bumblebees right? will have to trace everything out properly tomorrow, I hear the diodes are quite ESD sensitive. hope they haven't blown already, one can hope.

speaking of which, one of them was replaced? that sounds interesting, with what?

[golexica]

welp on those tunnel diodes, by "russian" would https://www.ebay.com/itm/GI304B-Ge-A...-/141378748975 do? seem to be the closest i could find? - germanium, close in max current/voltage and a bit lower in capacitance, but nothing crazy

[MotorBikeLes]

I would suggest your starting point with tunnel diodes is to assume they are OK. I think MOST failures are people "testing" them. I made a proper tester 40 years ago, and WMEBill has put one out, but starting point is "leave alone" until everything points that way.
Les.

[WME_bill]

Cossor 3122 timebase.
I endose what MotorbikeLes says about tunnel diodes.

To test them, you do really need another oscilloscope. Otherwise make up a little dc rig, never an ohmmeter. Both noted on my extract as below.
Be sure the max current can never go over twice the rated peak current, forward or back. So for your 1N3716, keep at all times below 10ma.
These tunnel diodes are very sensitive to heat when soldering, and to any current overloads. No however affected by radiation, if that is any consolation to you.

For the Russian ones as on ebay, the only parameter you need to worry about is in the first column of maximum parameters -In. For germanium, not GeAs.
So 1l304 or 1G304 meet this. The other variants of In are the tolerance of the peak current. It is sufficient as you can normally adjust in circuit for satisfactory operation.
The other parameters: capacitance Sn, valley voltage Un, ratio of peak to min current In/Iv, (quoting from the usual Russian data tables) are really irrelevant for a replacement part.

These tables have been posted on the forum. So also my summary of the usual GE/RCA/Mullard/Tek/HP ones, and a possible tester, based upon Wireless World March 1965 p108.
If stuck, I can re-post both.
wme_bill

[golexica]

Ooof, Glad you mentioned those caps now-c25 is fine, but someone has put the wrong inrush resistor on it- 100ohm instead of 1k. Can't imagine that higher inrush would be good for its longevity so imma switch it .

getting the feeling this thing has been worked on by somebody before without access to the correct components/manuals.

there were some Transistors in the x amplifier that had been replaced by a different model but with lower breakdown Tr4 and Tr3.
not much lower 45->35, but on the diagram those look like they might be working in avalanche mode? in which case wouldn't that alter the oscillation/switching? ( not EE trained, so can't be sure).

they looked kinda funky too- the leads were all black. tested on a hfe, and they came out fine, but they just look weird, maybe just corrosion? they have silvery leads as opposed to the gold plated ones on the rest.

[golexica]

Progress has been made! still won't sweep, but probing the sweep out and i get a 10V Sawtooth that corresponds to the timebase...now i just need to figure out where it's getting lost en route to the screen

[golexica]

well, i'm preplexed it's not that the sweep isn't getting to the screen at all, it's just responding very strangely to it- attached some images below.

see the one shaped like a hook? vertical axis is the triangular sweep waveform, it seems to respond normally to the sweep, untill it reaches a certain voltage level, at which point it snaps backward and stays there, but while staying lit, so not obviously blanking?

i can even trigger off that small piece of "correct" sweep behaviour, as you can see with the square wave, but you can see where it suddenly decides to stop sweeping with the two bright dots, but that they're slightly behind the furthest point swept to?

[Chris55000]

H!!

I did try and get the 3100 Y amp and TB manuals but so far I've not found a source for them – I would certainly have included scans of these with my mainframe scan I have listed if I'd been able to get them!

(I have a search saved for a 3100 with bookwork, and as soon as I can get this the bookwork will be scanned & uploaded!)

New manuals for other models, including the Gould–designed Dartron D12, are currently in the writing and drawing phase!

The fault you've pictured looks either like the operating point of the X Amplifier has shifted enough to bring the scan too far to the left (too high voltage on X1 plate and too low voltage on X2 plate) or the threshold of the flyback trigger is incorrectly set so the flyback is initiated before the correct length of sweep has been achieved!

Can you get a full range of movement of the X shift control or does it need a "hard stuck" point before the r.h.s. of the screen?

If you can move the trace, such as it is, over the full width of the screen without the squares expanding or contracting, etc., then the X final amplifier is almost certainly OK and correctly set up!

Some 'scopes have a preset adjustment for the "trigger point" that might have a bearing on the trace display you're getting – I'll study the circuit WME_Bill posted and return presently!

Chris Williams

[WME_bill]

Cossor 3100 +3122.
It seems that your timebase is working. But the X Amplifier in the main frame is not. I suggest you have good look at that.
Stop the sweep operating, by XAmp on the TB. The XAmp should now give you a central spot (or a vertical line is you have a Y input as in your first photograph.)
Check all the voltages on each transistor. Start with those at the high voltage end as they tend to fail more often. Tr7,8 = MM3001.
It looks as though the beam blanking is working. It is a complex circuit to understand, and made worse by the circuit being drawn over about 3 sheets of the manual.
You have the main frame manual 3100 and the TB 3122 circuits. The other parts of the manuals do not offer any further help. They are rather typical Cossor, little explanation and some very oddly drawn diagrams. You will wish for Tek or HP or Marconi.
wme_bill m0wpn

[golexica]

getting the feeling that nothing's technically broken, but that there are some oscillator stability issues at play. The trace has places that it likes to "hitch" on the screen, or oscilate back and forth between two points untill you "knock it out of its groove".

you mentioned before about being able to get higher bandwidth by modifying it with higher binned transistors, but better transistors can introduce this kind of instability and make the whole thing go haywire unless you're very careful (not a mod for the faint hearted)...and the average quality of transistors has improved drastically since 1970...

this is going to get fiddly -_-

can see what you mean about cossor, I got so confused as to why they were sinking the output of the sweep into ground through a 10ohm resistor, untill i figured out through testing that the diagram was just plain wrong- the 10 ohm is connected to pin 15, not 14.

Is it possible i could also have the schematic for the "input" part of the oscilloscope?
channel 2 is out of calibration a bit, and there are so many different screws, that I want to make absolutely sure i'm turning the right ones, Although maybe i'm getting ahead of myself, I need to fix the rest first

[golexica]

little by little it's getting better , fiddling around with transistors in the X Unit, and it works fully...below 2ms/division, above that it starts acting all screwy.

back to looking at the timebase again i guess, or fiddling with the X unit some more, but first, Lunch

[MotorBikeLes]

Two possibilties. Could the output transistors be breaking down as the frequency increases? I have had trouble years ago (with a FET I think) where things behaved oddly. I think I changed and cured problem, but cold checks on component looked normal. I put it on the CT71 curve tracer and as the current and voltage increased, the curves started flattening off. lower down, perfect.
One (very remote) thought. Is the delaying timebase somehow "breaking through". Most unlikely, but they do operate after the start of a scan so the shape of the output sawtooth becomes very different according to settings and mode of operation --- dirty switches leaking mode accros. Just a crazy thought.
Les.

[golexica]

by 2ms or below, i mean that it locks up on slow sweeps, but works fine for any sweep 2ms-0.5micro

it sweeps once across the screen, and then gets locked into a tiny oscillation on the right hand side of the screen, untill it's forcefully reset by some other means- a trigger or something, but only at 5ms or above

not the delay, i did just figure out how that works ("delay" is a really confusing word for what that does XD) but the lockup still happens whatever the delay setting is, even off.

[Chris55000]

Hi!

AH!

If it's only 5ms and above, then it's most likely a leaky or faulty timing capacitor on the sweep timing switch assembly, pay attention to C110A (4u7 electrolytic) connected to wafer WB/F of the main TB speed switch S7, also make sure the 10M resistor R114A on wafer WD/F is OK, and pay attention to all the wiring between PCB 'A' and the timing switch assembly – do the board's plug in ? – the drawings Bill provided suggests there's an edge connector on each of them!

Chris Williams

[golexica]

Originally Posted by MotorBikeLes

Two possibilties. Could the output transistors be breaking down as the frequency increases? I have had trouble years ago (with a FET I think) where things behaved oddly. I think I changed and cured problem, but cold checks on component looked normal. I put it on the CT71 curve tracer and as the current and voltage increased, the curves started flattening off. lower down, perfect.
One (very remote) thought. Is the delaying timebase somehow "breaking through". Most unlikely, but they do operate after the start of a scan so the shape of the output sawtooth becomes very different according to settings and mode of operation --- dirty switches leaking mode accros. Just a crazy thought.

Originally Posted by Chris55000

If it's only 5ms and above, then it's most likely a leaky or faulty timing capacitor on the sweep timing switch assembly, pay attention to C110A (4u7 electrolytic) connected to wafer WB/F of the main TB speed switch S7, also make sure the 10M resistor R114A on wafer WD/F is OK, and pay attention to all the wiring between PCB 'A' and the timing switch assembly – do the board's plug in ? – the drawings Bill provided suggests there's an edge connector on each of them!

This bumblebee (4.7mic) is, if not dead, extremely unhappy. It looks like a baked bean (peeling) and i can't get a reliable meter reading off of it. It's being switched in at the 5ms range too, looks like i need a new one No idea of the voltage tolerance, i think 16? how do you tell the polarity of a bumblebee, doesn't seem particularly obvious, seeing as the caps themselves are polarized. is it possible to substitute a more reliable form of capacitor? an MLCC/ something non polarized?

Btw, does the forum automatically strip Metadata from photos? almost uploaded a photo with my exact coordinates XD

[golexica]

I miss-spoke, apparently this is not a bumblebee capacitor, a bumblebee capacitor, is a thing that looks far less like an actual bee than this one

[Chris55000]

Hi!

That's a 4u7 35V tantalum bead electrolytic capacitor!

The colour coding is such that the positive lead is the right hand side holding the component with the spot of the colour–code pointing towards you!

I would turf that in the bin and fit a good–quality AVX, Kemet or Vishay replacement, you should be able to get them up to 50V d.c. working. New ones are normally rated ±20% but I think Mouser/DigiKey can supply ±10% tolerance components.

Modern types are much more clearly marked for polarity with a plus sign – the positive side connects to the timing switch wafer assembly and the negative lead to chassis–earth!

Unfortunately I've not been able to get the full 3122 manual but I can advise a suitable replacement from a pic of the original, as in this case!

Chris Williams

PS!

You should be able to obtain a quality polyester film 4u7 50V radial capacitor that should fit in the space available behind the timing switch assembly and this will be more reliable and stable in operation!

[Chris55000]

PS!

Cossor oscilloscope circuit diagrams aren't really that difficult to follow – I've seen a lot worse!

Chris Williams