Tek 2445A Repair, trace drift, CH1 out of spec, long startup

[fivefish]

That's strange.  Do the delta-V, delta-t, and SLOPE +/- switches work normally individually?

Yes, they work fine in normal operation.

Another way to get into diag mode is to hold down any non-latching button while the scope is booting.  It will think it has a stuck switch and enter diag mode automatically.

No go. It just displays TEST xx FAIL xx message (thinking a key is bad or stuck?)  It doesn't present me the menu option (according to service manual) I should get.


Another thing, I don't have a AM6012 DAC chip, and the layout of my control board almost looks like the diagram in the Service Manual, but different enough. The chip #s designations are different. I'm going bonkers here trying to correlate what the service manual says and what the actual PCB looks in front of me.

Heres some photo.  It says J651 and J652 for the left ribbon connectors, and that's what I have on my board too... so I know I'm looking at the right Control Board. But my IC is U2101 vs. U2134, U2201 vs U2234, etc.... every thing is different from the service manual. The service manual is 2445, while my unit is 2445A. I didn't know the difference is that huge, different part# designations, drastically different board layout.





I want to bang my head to the wall, because I'm studying the Service Manual, and then look at my board and "Where the heck is that chip? Put your meter on test point XXX, and I don't have test point XXX on my board." (maybe it's called differently on mine) . Very frustrating.

[MarkL]

Please read these two previous posts where I provided links to the manual.  Snippet below.  Looks like your board to me.

  https://www.eevblog.com/forum/repair/tek-2445a-repair-trace-drift-ch1-out-of-spec-long-startup/msg693467/#msg693467

  https://www.eevblog.com/forum/repair/tek-2445a-repair-trace-drift-ch1-out-of-spec-long-startup/msg693482/#msg693482

[siggi]

The service manual is 2445, while my unit is 2445A.

The 2445 and the 2445A are a generation apart. See <https://en.wikipedia.org/wiki/Tektronix_analog_oscilloscopes> - there's three years between them, and you really do need the right service manual.
IMHO it's well worth the money to buy a good quality scan of the service manual. All the Artek <http://artekmanuals.com/manuals/tektronix-manuals/> manuals I've bought have uniformly been of great quality, and they're OCR processed, so you can do (some) text searching in them. Artek has the 2445A - (the "A" is critical , and has bundle pricing for all the manuals (no relation except as a satisfied customer).

On a tangent, one thing that can cause wonky deflection ratios is oxidized contacts in the attenuators.
If you set the channels to DC coupling, you can measure the input impedance with your DMM. Then scrub the channel through the different attenuation ratios (there are 2 10X attenuators in each attenuator hybrid, so you want to hear two relay clicks), and see whether the input impedance stays at 1MOhm. If you see readings like 1.8MOhm or open, it's a good bet that you have a "mechanical" issue with the attenuators - most likely oxidation.

[MarkL]

"TEST xx FAIL xx" is what it should do, but then it should dump you into the diag monitor.

As far as not getting the "DIAGNSTIC. PUSH A/B TRIG TO EXIT" display, perhaps it's trying to display it but it's positioned off the screen.

None of your screen shots have the top status line on them either, which may be more evidence of a problem with positioning or the readout logic.  Try the beam finder to see if there's anything outside the edges of the screen.


Maybe there's a setting to disable diagnostic mode, but I'm not aware of it.

[fivefish]

Thanks all. I didn't know about the non-A and A difference.

I bought the Artek manual, now just waiting for the emailed links.

[PA4TIM]

I bought a Tek 2710 manual from Artec and that was well wordth the money. I had a hard to read pdf from an other subversion of the 2710.  Daves copy was a very high quality and included the updates/changes/faults info. That made repair a whole lot more easy . I will not try to repair a scope without the right documentation.( And I'm "used" to repairing instruments without documentation)

Only use the right manual. The A is not sort of subversion, it is a complete different scope.

[fivefish]

On an earlier post, I mentioned I can't get my scope to go into Diagnostics mode (pressing delta V, delta T, slope).

Upon closer look, it seems my SLOPE button isn't working?  I don't think I've ever used this button during normal mucking around.

I shot a video of it, can someone confirm?  The LED lights are not switching from (+) to (-).
https://www.youtube.com/watch?v=BQNo6bC5MXc&feature=youtu.be

[MarkL]

Yep, that's broken.

[MarkL]

It would be worthwhile to play with all the buttons and switches to see what else isn't working.  Could be something in common.

[fivefish]

Thanks for the confirmation Mark.

UPDATE: Confirmed... the Push-Button switch for SLOPE is broke, tested it with a DMM.  Gotta find a replacement. 

Unfortunately, also broke one knob in half while in the process of removing them... didn't see there was a hex set screw. Bought a replacement on eBay.

[siggi]

Thanks for the confirmation Mark.

UPDATE: Confirmed... the Push-Button switch for SLOPE is broke, tested it with a DMM.  Gotta find a replacement. 

If you have some (70-99%) isopropyl alchohol - or better some DeOxit - it's worth giving the button a squirt and a bit of a workout. Sometimes this is just contact oxidation and you can bring them back. It sure beats waiting for a replacement .

[fivefish]

Thanks for the tip siggi... worth giving it a shot.

[fivefish]

Alcohol cleaning didn't work. I think it's a mechanical problem.

I don't have an exact Tek push button switch, but I do have a low-profile, short travel push button switch from my parts collection. The legs do not align, the Tek switch has a wider pitch... so bent the legs of my switch slightly, to go into the original holes. Made sure the switch is level and soldered it.



Next problem, the push button caps don't work with my switch. It doesn't have the right "profile" for the switch cap to snap in. So I had to break the legs of the switch caps so the switch caps will sit directly on top of plunger, and be able to actuate the switch.

Screwed together the front panel switch and pot board, making sure I can still hear/feel the clicking of all the switches while I align the switch caps with the panel. At this point, feeling good that the replacement switch  is working.

Installed front panel assembly back to the scope, connected the 2 ribbon cables and powered it up.... pressed the SLOPE switch and now it works, and changes from (+) to (-) and vice versa.




Now, the real goal.... go into Diagnostics mode. Press Delta-V, Delta-T, and Slope.... YES! Scope goes into Diagnostics mode.  Did the All Test, and everything passes test. 



Now I can continue following the rest of the service manual Calibration procedures. 

Since all TESTS ran successfully, I'm thinking maybe the scope just badly needs calibration?
So that's my next to do item.

[fivefish]

The 2.82V marking you see on the scope is the actual Vpp of that sine waveform. I double-checked and verified that Vpp value with my Rigol scope.

As you can see, there is a problem around the BNC jack of Channel 1 causing it to display bigger/taller waveforms.... thus making it appear that it's more than 2.82Volts.   

Frequency value is correct though.

https://www.youtube.com/watch?v=-KFMzFOb9XU&feature=youtu.be

[siggi]

As you can see, there is a problem around the BNC jack of Channel 1 causing it to display bigger/taller waveforms.... thus making it appear that it's more than 2.82Volts.   

That's ... interesting ... Looks to be a mechanical problem of some sort. I replaced a worn-out BNC on my 2467, which caused intermittent connections. The amplitude would tend to drop from nominal, though, not increase.
It's easier to get at the attenuators than it looks like at first glance.
Remove the front panel bezel and a metal bar running across CH1/2, then the various screws fastening them. There are a couple of screws you get at from below the mainboard, inside the hollow left by the front panel assembly. You need to de-solder a couple of connections, and then they just pull out - they're socketed.

From your list of symptoms, it seems to me the channel 1/4 trace drift is still unexplained. This isn't something that'd be affected by calibration. Also I think you'll find that the calibration procedure is a PITA to get through unless you have the equipment specified, as the software routines are somewhat intimate with the specified calibration generators.

[MarkL]

From your list of symptoms, it seems to me the channel 1/4 trace drift is still unexplained. This isn't something that'd be affected by calibration. Also I think you'll find that the calibration procedure is a PITA to get through unless you have the equipment specified, as the software routines are somewhat intimate with the specified calibration generators.

Plus, I wouldn't attempt the calibration without fixing or at least explaining the drift problem.  If the problem is an unstable offset, you might not even be able to get through the calibration.

[fivefish]

I looked more deeply in the calibration steps and I think it will be impossible for me to do.
I don't have a time mark generator. I only have a signal generator.

I think the un-centered trace *offsets* I see upon powering up is from old/non-valid cal data anymore stored in the EPROM? Because it seems consistent, though different for ch 1 to 3 traces. Only ch4 is perfectly centered upon power up. After I manually  centered the traces, the Volts peak to peak and frequency readings are correct (except for that CH1 flakiness)... the trace just starts uncentered from power-up. -- I can live with that.

And even if I do have a time-mark generator, I do need to fix that Ch1 BNC problem first. For now, this is the most serious problem. And my next to-do list.  Thanks siggi for that disassembly instructions.

Re: Trace drifting up or down.... I think this is another serious problem than the power-up uncentered offsets.  I admit I haven't seen this problem again in the last few days. Maybe because the unit has been left powered more often and for longer periods of time? As I said earlier, this trace drift disappears when the unit gets warm and running for maybe half an hour. Still something I'd like to troubleshoot.

[fivefish]

I'm going through the TEST/EXER/CAL menu (thanks to my now working Slope button) and I cleared all the saved settings in the scope. I think these setups were there originally since I bought the scope almost 6 years ago. 

After clearing everything in memory, I adjusted all the 4 channels to be centered and in my preferred initial time/volt settings I want, and SAVED that new setup to #1. Then going to EXER 06, made the scope recall setup #1 during power up. 

Now everything is centered during scope power on! 

https://www.youtube.com/watch?v=giZzZTKgn88&feature=youtu.be

Bottonline: I think the scope was just recalling a bad stored setup during power up. Problem fixed.  Two down, two more to go.

[fivefish]

HA! Found the problem why my Ch1 is out of spec. It looks like the problem is mechanically related, specifically in the attenuator section.

So earlier I noticed if I wiggle my BNC hard, I can make the Ch1 reading come out correct. Also noticed that if I bumped the unit, it will cause disturbance in the Ch1 waveform reading. Hmmmmm.... a clue....

I repeated the experiment on Ch2 and this vibration problem doesn't happen with Ch2, no matter how hard I thump/bang on the unit. But on Ch1, I can see slight mechanical vibrations affect the signal I'm seeing. It will go from showing the correct Vpp reading to something out of whack.

The attenuator section have (4) relays inside them. And it looks like my problem is one of these mechanical relays not working properly.... stuck closed or open in it's position. Thumping and whacking the unit, sometimes dislodges that relay and then I get a correct reading.

This may also explain why my earlier 50-ohm overload test didn't work for Ch1.  The relay connected to that 50-ohm resistor didn't move, and at this point, possibly already burned up that internal 50-ohm resistor too.





Here's a video showing my "hammer test" method . 500mVpp is the correct reading for that waveform, but thumping the attenuator with the palm of my hand, or my finger, would cause it to give wrong readings (super high). Then I had to whack it again to make it show the correct reading.

https://www.youtube.com/watch?v=cJdmQHiXiFs&feature=youtu.be

I can buy these attenuators from eBay for $35 to $45 apiece. And this should fix my Ch1 out of specs problem.

So now, the only problem left is the trace drift...

[tautech]

i've seen other Tek scopes where the BNC solder joints have cracked from physical stresses.
Check for this first.

I'm not sure if 2445 BNC's solder directly to the PCB, others will know. 

[siggi]

HA! Found the problem why my Ch1 is out of spec. It looks like the problem is mechanically related, specifically in the attenuator section.

So earlier I noticed if I wiggle my BNC hard, I can make the Ch1 reading come out correct. Also noticed that if I bumped the unit, it will cause disturbance in the Ch1 waveform reading. Hmmmmm.... a clue....

The attenuators are quite repairable - I cracked them open for cleaning on my 2430 and replaced the BNCs on my 2467.
I posted some pictures of the attenuator guts from my 2430 in that repair thread. See e.g. https://www.eevblog.com/forum/repair/attempting-to-repair-a-tektronix-2430/msg431422/#msg431422. The 2400-series all uses the same attenuators AFAIK.

However, I can't think of a fault in the attenuators that'd lead to larger readings than nominal, except perhaps if the substrate under the 1M film resistor was cracked, or maybe if the attenuator grounding was compromised.

So now, the only problem left is the trace drift...

Can you reproduce the drift?

This might be nothing more than contact oxidation on one of the hybrids. My 2430 had all kinds of glitching initially, but with some cleaning and subsequent use it all leveled out.

If you can reproduce, you can measure your way to a diagnosis for sure, or you can just release and re-seat the hybrids - perhaps cleaning the contacts while you have them out with a bit of IPA. The latter only takes 10 minutes.

[fivefish]

siggi, thanks for the reply.

Right now, there's a couple of "related problems" with Channel 1. So my hunch is it's the relays in the attenuator that's bad, or stuck?

I'm thinking if it's just a bad trace in the BNC path, then the signal would either be there or not.  But in this case, the signal goes from "correct" reading to "going ridiculously large"... i.e. 500mVpp becomes a little bit above 2.5Vpp shown by the scope.  Tap and whack the attenuator, and the reading going back to what it should be, in my example, 500mVpp.

Also, the 50-ohm overload test as outlined in the Service Manual doesn't work as expected with my scope. I don't hear any relays clicking. So my hunch is there's a problem with the relays.... and maybe even possibly burned out the 50-ohm resistor.


Can you reproduce the drift?

No not yet... want to fix Ch1 first before I tackle this.

My observation on the drift are: if unit is left powered on for long periods of time, and I turn off/on the unit, it doesn't exhibit drift problem. (Maybe because unit is still warm? or all caps still charged? )  The drift happens *usuallly* the first few minutes the scope is powered on (after being unpowered for some time).

Also... sometimes the scope takes a long time to turn on (i.e. displaying a flat trace). More likely to happen if the scope hasn't been powered for a long time. So first time you turn it on, it either takes a long time before it shows a trace on screen, or it "hangs up" and never actually show a trace on screen.  Turning scope off then on again, fixes this "slow bootup" time. The second power on, everything starts fast and trace shows up on screen. Maybe there's some bad/failing caps in the PSU (or some other board) I'm thinking.

Also, I bought a temperature sensor gun (the one with the laser pointer) and pointing it at the various caps, on the PSU low-voltage board shows 2 caps that seems to be reading a couple of degrees higher than their neighboring capacitors of the same kind/size. One of these caps also had some slight residue on the top. It's not bulging, but you can tell some electrolyte did leak out ever slightly..

[tautech]

Of course Ch1 gets most of the work on any scope and most of the abuse too, this is what I guess is your problem, the input has been overlaoded at some point. Who knows what a previous owner did to it. 

So that should point to the dividers, fets etc, sure an attenuator module should sort it, but then where's the fun in fix gone? 

[fivefish]

SUCCESS!  Cost me $22+shipping (cost of Channel 1 attenuator module) to fix my scope!



Replaced Channel1 attenuator.  And scope is now working properly, giving good and correct readings for Channel 1.



After replacing the attenuator and powering up, I got a TEST 04 FAIL 10. -- oh oh, what did I screw up?
Reading the manual, it seems I need to re-calibrate the scope... or replace the EAROM.



Problem is I don't have a pulse generator, so I cannot calibrate this scope properly.
So what I did is I just performed CAL01, and just continued to press NEXT (up Coupling button) without
adjusting anything.  Exit, power off, power on and the FAIL message is gone.

I then checked if voltage and frequency readings via the cursors are correct and they are. Fine by me!


But I need to buy some replacement knobs that I broke.
Also need to find the cause of that Channel 2 drift problem. It seems only Channel 2 is drifting from my close observation.
That will be for another day.

[fivefish]

100% fixed.

The problem with the "long startup" and trace drift (and also volts/time cursor drift) is related to the Power Supply capacitors.

I figure I have nothing to lose so I decided to replace the PSU caps.

That's when I discovered several 180uf/40V, and 250uf/20V caps were leaking electrolyte from the bottom. The tops weren't bulging, so it wasn't obvious. When I removed them, that's when I discovered the leak. Replaced them all with 330uf/50V caps 105C.

Also replaced other caps like the 100uf/25V, and 47uf/25v, 100uf/100V... I already did all the hard work of disassembling and removing the PSU boards so I might as well replace them.

Now the scope is stable, no more long startup times, and the traces don't drift upwards or downwards. Everything is now stable.

Photo showing all the caps that have been replaced in the HV PSU and Regulator board sections.