-
[FIXED] Tektronix 2465A repair: very bright fuzzy blob
Posted by
Bill Woodbridge
on 15 Jul, 2020 19:47
-
I’ve just finished the recap and reassembly of my 2465A. On power-up, all the front panel indicators light up together, but then all stay on and don’t move beyond that into the self-test routines etc. At the same time, a big (1cm diagonal, roughly rectangular) fuzzy and very bright blob appears on the CRT, just off centre. None of the CRT controls (intensity, readout intensity, focus, beam find) affect it at all – it just sits there looking like it wants to burn a hole through the tube.
(When I first bought the scope, it was working perfectly but I decided on a precautionary recap. I had to remove the power supply boards (A2 and A3) for photography to identify the specific version, but on reassembly the scope stopped working and instead showed the fault I have just described. Rather than stopping to investigate, I pressed ahead with the recap hoping that would cure it, but now I’ve come back to exactly the same problem.)
Given the brightness of the blob I was afraid of tube damage, so I tried to turn down the grid bias pot as a temporary measure while I investigated further. But turning the pot all the way down (it was nearly at its minimum already) made very little difference and the blob was still there. Turning it up made it even brighter.
It seems sensible to look at the power rails first as the fault seems so fundamental. There is a thread at
https://groups.io/g/TekScopes/topic/7650976 describing a similar problem (caused by power supply issues), and containing much detailed component-level advice, which I want to use as my faultfinding guide. But I’m scared to leave things powered up for more than a few seconds due to the blob. Am I safe to just ignore the blob (it’s so bright!) and crack on, or is there another easy and safe way to ‘kill’ the tube temporarily while I investigate?
Many thanks for any advice please.
-
#1 Reply
Posted by
mawyatt
on 16 Jul, 2020 13:49
-
I had some issues with repairing a couple 2465s. One was after fixing the test waveform circuit & recapping, the larger wire harness that goes from the power supply to the main PC board had a broken pin which prevented one of the supply voltages reaching the main PCB. A check on the main PCB top side voltage test socket showed this. I removed all the soldered in connector pins and just soldered the wires directly to the main PCB. Now this scope works fine and directly soldered wires looks like they are originally that way.
On the other scope after repairing both channel 1 and 2 input hybrids & recapping, the display was all messed up. It took awhile to find this with lots of voltage measurements and studying the schematics. Eventually leading to the CRT bias voltage circuits. It was the small single pin connector that plugs into the small PCB that sits on top of the CRT had pulled loose but still engaged!! This is the board that has a single pot. Once this connector was reseated the scope works fine.
Hope this helps,
Best,
Mike
-
-
Thanks very much indeed - your first suggestion sounded very convincing (I also checked that single pin connector on the small PCB, which was OK). It does feel like one (or maybe more) of the power rails just isn’t connected, and those two power connectors did feel rather sloppy when I reinserted them.
So I tested them through for continuity with power off, going from the relevant components on the A2 regulator board, mostly the regulator output capacitors (it’s impossible to get at the connector header pins themselves once it’s all back together in the chassis) to the voltage test socket on the main A1 PCB. It seemed like everything was getting through to there. But the -15V unregulated isn’t brought out to that test socket, and I’m not yet convinced I’ve got continuity on that one. ‘Upstream’ it passes straight through the A2 regulator board (as you might expect), so I want to get back to its source on the A3 board which is more difficult to probe.
Downstream it simply passes through the A1 board (apart from one resistor I think) all the way to the A9 high voltage board where it feeds the high voltage oscillator on which much of the CRT depends, via another connector which also felt sloppy. Which is an encouraging fit to the symptoms I’ve got, so I’m very much hoping the -15V unreg is simply open circuit somewhere along the line through all those boards and connectors A3-A2-A1-A9. It would have to be the most difficult of the lines to probe!
Thanks again, will revert with the final result in the hope it will be useful for others.
-
#3 Reply
Posted by
Kwakerman
on 17 Jul, 2020 17:16
-
The power supplies on these scopes comes up fairly quick (quicker than the tube heaters). Set a DVM up to read the 10V supply on the test socket then turn the scope on, you should get a reading in a second or so then turn off, go to the next test point and repeat the procedure to check all the supplies are close to nominal. I had to do this when I had a tripping issue and managed to spot the -15 didn't come up when all the others did as one of the caps on the main PCB was shorted
-
-
Many thanks for the tip. You encouraged me to bite the bullet as I was getting nowhere with my continuity tests (the -15V turned out to be a red herring as continuity was perfectly OK right through to the A9 board transformer). So I risked turning on the machine for long enough to whizz around all the pins on the test socket (about 15 seconds), shouting out the readings to my wife to scribble down (which was very kind of her, although she seems less fascinated by the scope and its problems than I am, strangely).
Which finally gave me something concrete to go on: all the rails were OK with the exception of the +15V and +5V, which had both collapsed to less than 0.25V. The +5V is derived from the +15V so obviously I looked at the +15V first, as the manual instructs.
The unregulated +15V into the regulator looked OK at 18.8V. I wanted to isolate the fault to either the regulator circuit itself, or to a low resistance or short somewhere downstream, so I unplugged the A2-A1 power connectors and looked into the +15V input to A1 with the multimeter (power off of course) and saw a short to chassis (0.4 ohms).
Unfortunately the +15V rail powers a huge number of devices, not only on the A1 board but then downstream on all the others too. I guess the first step is to disconnect the downstream boards and see if I can isolate the short to a particular board, but I’m not sure where to go after that other than brute force lifting component-by-component, and the thought of that is pretty off-putting especially on the giant A1 board. Start with the caps I guess as the most likely culprits, and perhaps pay attention to milliohm differences that might give a clue to the physical proximity of the short to the test point? It’s slightly worrying though that the fault was there before I started the comprehensive recap, and it’s still there exactly the same now the job is finished.
So investigations continue (and at least I don’t have to risk the blob while I'm only measuring DC resistance) but I just wanted to thank you in the meantime.
-
#5 Reply
Posted by
tautech
on 19 Jul, 2020 00:42
-
Which finally gave me something concrete to go on: all the rails were OK with the exception of the +15V and +5V, which had both collapsed to less than 0.25V. The +5V is derived from the +15V so obviously I looked at the +15V first, as the manual instructs.
IIRC these are infamous for an incorrect PCB silkscreen on the PSU PCB that has those recapping them installing the ecaps in the wrong polarity.
I'm sure this is mentioned more than once in the big 2465B thread and it's not an uncommon trap for those recapping these and for those that are aware of this they replace one cap at a time ensuring each is installed with the same polarity as the one removed.
Just one of the mentions:
https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/msg1129304/#msg1129304Tracing the polarity through should find where you went wrong.
-
#6 Reply
Posted by
tggzzz
on 19 Jul, 2020 07:48
-
Which finally gave me something concrete to go on: all the rails were OK with the exception of the +15V and +5V, which had both collapsed to less than 0.25V. The +5V is derived from the +15V so obviously I looked at the +15V first, as the manual instructs.
IIRC these are infamous for an incorrect PCB silkscreen on the PSU PCB that has those recapping them installing the ecaps in the wrong polarity.
I'm sure this is mentioned more than once in the big 2465B thread and it's not an uncommon trap for those recapping these and for those that are aware of this they replace one cap at a time ensuring each is installed with the same polarity as the one removed.
Just one of the mentions:
https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/msg1129304/#msg1129304
Tracing the polarity through should find where you went wrong.
If you are thinking of the bige electrolytics on the PSU board, I haven't heard the polarity being wrong. There are errors w.r.t. C1132 and C1115 being swapped, e.g.
https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/msg2906934/#msg2906934There will be more information on the groups.io TekScopes board.
-
-
Thanks both - the C1132 / C1115 error on the schematic was well signposted to me by the seller of my recap kit, and I'd already read about it online, so I think I avoided that pitfall. Likewise I was paranoid about wrong polarities, as I'd watched Dave's 2465 repair when one of the caps near there gets put in the wrong way, with amusing results! So I think I'm in the clear on that one.
I've just made a start trying to find the +15V supply short on the A1 board (it's somewhere there and not on any of the downstream boards as far as I can tell) but it's hard going at the moment. I think I will have to remove the board again as most of the electrolytic capacitor terminals aren't probe-able from the component side because I've soldered them too close to the board. I'm trying to use the technique of looking for the resistance decreasing as I probe physically nearer to the short, but it's only 0.4 ohms at the test socket and my meter only has 0.1 ohm resolution, so it's going to be tough going.
I read about a technique to inject a small current into the shorted +15V supply rail and instead read millivolt voltage drops across the various components, which decrease with proximity to the shorted one, but I don't like the idea of connecting any power at all to the A1 board, especially with the other power rails floating.
-
#8 Reply
Posted by
Gyro
on 19 Jul, 2020 19:23
-
I read about a technique to inject a small current into the shorted +15V supply rail and instead read millivolt voltage drops across the various components, which decrease with proximity to the shorted one, but I don't like the idea of connecting any power at all to the A1 board, especially with the other power rails floating.
This is a very useful technique. Obviously do it with the scope powered down so you don't have to worry about the other rails. Set your bench PSU to a very low output voltage (<1V if possible) and connect to the +15V rail. Set the current limit reasonably high - 1A will give you 1mR per mV resolution so it is a very sensitive technique. Just be careful that you keep the PSU voltage knob wound down low enough that it can't back-drive the active parts of the regulator if [Edit: your probing suddenly] clears the fault.
-
#9 Reply
Posted by
tautech
on 19 Jul, 2020 21:33
-
Thanks both - the C1132 / C1115 error on the schematic was well signposted to me by the seller of my recap kit, and I'd already read about it online, so I think I avoided that pitfall. Likewise I was paranoid about wrong polarities, as I'd watched Dave's 2465 repair when one of the caps near there gets put in the wrong way, with amusing results! So I think I'm in the clear on that one.
I've just made a start trying to find the +15V supply short on the A1 board (it's somewhere there and not on any of the downstream boards as far as I can tell) but it's hard going at the moment. I think I will have to remove the board again as most of the electrolytic capacitor terminals aren't probe-able from the component side because I've soldered them too close to the board. I'm trying to use the technique of looking for the resistance decreasing as I probe physically nearer to the short, but it's only 0.4 ohms at the test socket and my meter only has 0.1 ohm resolution, so it's going to be tough going.
I read about a technique to inject a small current into the shorted +15V supply rail and instead read millivolt voltage drops across the various components, which decrease with proximity to the shorted one, but I don't like the idea of connecting any power at all to the A1 board, especially with the other power rails floating.
Good so not the recap then.
Unless you've left a tiny solder bridge somewhere.
Tants, do 246* scopes have tants ? If so those would be on my list to check and particularly any that see voltages close to their max rating as we derate them more these days than what designers did when tants were new.
-
-
Progress at last! I was in the process of disconnecting everything from the A1 board prior to embarking on the current injection technique, the final one being the tiny 2-pin connector to the A8 display illumination board, and all of a sudden the short on the +15V line disappeared. Sure enough, it was the illumination board (aka three incandescent bulbs in series) that was shorted. Although that connector was mentioned as having a 15V supply in the power distribution diagram 11 that I had been using, my only excuse was that it was tucked away in the small print amongst the many other things fed by that rail, and I’d missed it first time around.
As usual, it should have been more obvious looking back as my only repair to the scope (prior to the recap) had been to replace the three bulbs which were all blown. My replacements worked perfectly – for a while until this short problem appeared, and after that I just assumed they weren’t coming on because of a 15V short elsewhere.
So now (not even stopping to investigate A8 but just leaving it unplugged for now) the front panel goes through its self-tests and settles down, but sadly I didn’t get the happy ending I was hoping for yet: it fails diagnostics on ‘Test 04 Fail 13 XX <D128>’. That’s a calibration checksum error, so something has damaged the calibration data. The RAM battery (mine is the Keeper battery version, not the Dallas) has always looked very healthy at 3.7V, and I purposefully didn’t recap at the RAM (although it is one of the two tants in the machine, thanks for the previous comment) to avoid risking the OV soldering iron shorting the battery, but even so I’m now in for another world of pain with calibration I guess. But it might be a good excuse to buy some more vintage test equipment.
Thanks again to everyone who has helped me to get this far, at least calibration is a more well-trodden path hopefully.
-
-
I’ve been reading up on the principles of 2465A calibration and trying to weigh up the strategy going forwards.
To dip my toe in the water, I tried the easiest calibration routine namely CAL05 (hours of operation and number of power-ups), resetting both to zero. The machine appeared to save this data, since it appeared on the next power-up (incrementing the number of power-ups by one as expected) so from that I’m assuming the RAM itself and its battery is OK, at least for the moment. The checksum error still occurs, so CAL05 didn’t rewrite it successfully – from that I’m assuming that the checksum isn’t a single global value that’s automatically regenerated after every successful partial CALxx (although some online comments suggested this might be the case).
There is currently no trace (just display and scale factors) even using beam find. I embarked on the start of CAL08 (CRT adjustments) and the Grid Bias test displayed the single spot as it should, which I could adjust with the Grid Bias pot. After that, however, I had to abandon CAL08 due to lack of test equipment as I haven’t even got a simple function generator to put a sine wave on the screen as needed for the later focus adjustments. There’s still no trace, and my assumption here is that the calibration constants have been so badly damaged that the machine can’t even display a simple horizontal sweep. There seem to be plenty of adjustments in the various CAL routines that, if not correctly done, could kill the trace entirely. So if things are this bad, the checksum problem is entirely secondary and will clear once the calibration is done correctly.
So my plan is to:
1) Replace the Keeper battery – although the voltage is OK, it’s the original one so clearly it should go, and anyway I’ve got nothing more to lose from disconnecting the battery at this point
2) Acquire enough test equipment to get through the calibration somehow. I’m not averse to spending money as I need a function generator and a frequency counter anyway, but it’s a shame that the 2465’s requirements are rather more specialised than anything I’m ever likely to need again.
What worries me more about the world of calibration isn’t the complexity as such, but all the online talk about ‘hidden’ procedures not in the manual, and the possibility of getting permanently stuck in a calibration loop because the machine thinks something is out of range and won’t exit/store cleanly. I’ve already noticed that the CAL08 tests don’t come up in the order advertised in the manual. My serial number is 131499, and I’m using the updated manual pages dated 6/5/87 for serial numbers B016276 and above, but I wonder if there was a later firmware change and manual update that I haven’t got?
If I run into calibration problems, which seems more than likely, I’d be grateful for the advice on forum etiquette please: should I carry on here, start a new thread or add to the main 2465 Teardown thread?
-
#12 Reply
Posted by
tggzzz
on 22 Jul, 2020 18:19
-
I’ve been reading up on the principles of 2465A calibration and trying to weigh up the strategy going forwards.
...
There is currently no trace (just display and scale factors) even using beam find.
...
If I run into calibration problems, which seems more than likely, I’d be grateful for the advice on forum etiquette please: should I carry on here, start a new thread or add to the main 2465 Teardown thread?
No point in calibrating it if there's no trace.
Use the TekScope archive and knowledge at
https://groups.io/g/TekScopes The denizens are friendly to those that think and do some research
If you add to the main 2465 thread, any useful info will be lost in the middle. Once you have found the beam and this thread's title is solved, I'd consider starting a new thread or piggybacking on another directly relevant calibration thread (if any).
-
-
Thanks again. The trace is back, which is a huge relief. I’m not sure if it was just finger trouble while I was confused in the depths of CAL08, but it all now looks very solid. Apart from calibration of course which is way out – the cursors say the test 5kHz square wave is 4.48kHz. And the focus and intensity controls interact a bit too much for my liking, but I’m sure that’s because I haven’t finished CAL08.
So I’ll amend the title of this thread to fixed, with thanks to everyone for the speed and expertise of the responses.
I’ve also had a bit of luck, and have just bought for a reasonable price most of the calibration equipment I need, namely the Tektronix SG503 signal generator, TG501 time mark generator and PG506 calibration generator all in a TM503 chassis. The plan is to sell it again on eBay once it’s done its job.
FOOTNOTE CONCERNING THE A8 ILLUMINATION BOARD PROBLEM 25/7/20
For completeness and of case it is of use to others, here is the final twist in the tail.
My home-rebuilt A8 illumination board that was shorting the +15V rail turned out not to have any blown bulbs - it was perfectly OK. (I had been puzzling how three bulbs in series could all simultaneously fail short-circuit). Rather, I had made a stupid mistake in reassembly by soldering the bulbs to the wrong side of the board, and then installing it in front of the perspex block which couples the light up to the graticule, not behind where it should be. It all fits together that way, but the problem is that when it's in front rather than behind, the pads on the board all press up tight against the inner metal bezel for the tube as it's tightened down. Hence the short!