Tek 465 odd PSU issue

[WaveyDipole]

My Tek 465 seems to have a weird PSU issue. First let me mention that as well as being powered from the mains, this scope has option 7 (DC to AC inverter) included and can be powered from a 24v or 12v DC supply.

The scope normally runs fine, but when running from the mains, the trace disappears after just a few minutes of running. The channel indicator lights are still on, the display light works and the fan is also still running, so there is still power to the equipment. Checking the supply lines when this happens shows that the 15v, 55v, 110v and HT lines are OK, but the -8v line is down (+0.7v) and the -5v line is low at around 3.8v. The common factor is the op amp IC (U1554) which requires the -8v supply, hence both lines are affected. When powered off and then back on again, even relatively quickly, the scope sometimes seems to work OK again for a while. Other times it takes quite a long time or switching to DC running to recover. I have tried swapping the IC (LM1458) for a spare NE5532 that I have, but it made no difference.

The curious thing is that when running from a 12v supply using the option 7 inverter, it takes a lot longer for the fault to occur. I had it running for something like 20-30mins before the problem kicked in. An examination of the inverter circuit and how it is integrated reveals that both the mains AC supply and inverter output are delivered to the same power transformer except via separate windings. All secondary supply lines are run from this same transformer. I therefore don't understand why the behaviour is different when running off the mains?

When the problem occurs, the DMM reads -0.7v across the rectifier/smoothing cap (red probe to +) although around 15v of AC is present. One rather odd observation: it seems to take a couple of seconds for the AC voltage to climb up to its max level when I take the reading. This tends to suggest that the rectifier has failed, but when I turn off power and test with a DMM using the diode test, it tests fine both from end to end and across the individual diodes.

This does seem to look like a thermal breakdown problem, although since the scope seems to recover pretty quickly sometimes, this would seem to run counter to that theory. I'm not sure whether a rectifier can have a failure mode like that? On the other hand if the smoothing cap was shorting to the point where there was only 0.7v across the rectifier, I would have thought there would have been smoke and a bang by now, but it doesn't seem to matter if I keep the scope running for a while, which is why I have plenty of time to take readings with the DMM.

I do have a rectifier to use as a substitute, but unfortunately it is not possible to get access to de-solder the existing part easily. It looks likely to take hours of dis-assembly and re-assembly. The quick way would be to snip it off the top but even this is not an easy proposition as the legs are difficult to get at and the original part would probably need to be crushed in order to be removed. I would therefore like to be sure that it is faulty before embarking on this particular course of action.

[MrBungle]

Oh extracting the A board. Many a fun evening doing that in my life

Might be the filter caps as well. They drag the rails down when they give out. If you have another scope, check the rails as this may be the DMM approximating what is really going on. Also check all the tantalums for shorts - they tend to cause this sort of thing as well and being remote from the power entry they like to burn up traces on the boards which turn into nice low ohm resistors as well so be careful.

Edit: feel around for warm pass transistors too. 0.7v looks like a base/emitter drop.

[RogerEvans]

What frequency does the inverter run at? There is obviously enough iron in the transformer to cope with 50Hz but the designers may have opted for a few hundred Hz in the inverter and that would reduce the ripple current and consequent heating of the reservoir and smoothing capacitors. That could explain the greater delay when running from the inverter.

Roger

[Skywave]

Your described symptoms could be caused by a leaky or low capacitance C1562: 3,000 uF. If you do replace it, whilst you're there, I would replace C1552 (5,000 uF) and possibly C1569 and C 1559 (o/p rail decouplers) just to make the 'visit' worthwhile.

Al.

[WaveyDipole]

Originally Posted by RogerEvans

What frequency does the inverter run at? There is obviously enough iron in the transformer to cope with 50Hz but the designers may have opted for a few hundred Hz in the inverter and that would reduce the ripple current and consequent heating of the reservoir and smoothing capacitors. That could explain the greater delay when running from the inverter.

That's a good question and I couldn't find it specifically mentioned. However the Option 7 maintenance section discusses setting up the inverter balance. If I'm looking at this correctly, then one of the diagrams states the time period for half a cycle which is 1.3ms. A complete cycle would then be 2.6ms. Using an online frequency converter calculator I arrived at around 312hz. If this is correct then evidently a higher frequency is indeed being used, which, as you say, accounts for a different thermal response. I wonder if this also explains why the fan seems to run faster....

Originally Posted by Skywave

Your described symptoms could be caused by a leaky or low capacitance C1562: 3,000 uF. If you do replace it, whilst you're there, I would replace C1552 (5,000 uF) and possibly C1569 and C 1559 (o/p rail decouplers) just to make the 'visit' worthwhile.

C1559 has already been replaced topside as it was shorted. Another wet tant had leaked its electrolyte onto the PCB and was replaced after the gunge was cleaned off. A couple of other wet tants were also replaced for good measure. All was done after preliminary checks and before applying power.

I've had some time to investigate further today and I think you are right about C1569 being the culprit. At first I couldn't remove the solder properly with the temperature controlled iron so I had to resort to the SP25 with its large tip. This worked and I was able to remove the solder cleanly from the centre (+) pin and attach a spare 3300uF cap to the pad without shorting to the pin of the original cap. I've had the scope running for a good half hour and it has not lost its trace so far. The rectifier still runs quite hot, but the cap is cold. The original one does get warm, although being next to the transformer I was not sure where the heat was from. I am really surprised that nothing went up in smoke though!

Obviously this will now require replacement probably along with C1552 and C1542 for good measure. The capacitor I am temporarily using is very much smaller standard 85deg part. Although modern parts do tend to be smaller, I rather suspect that something rather more robust will be required. I have two 4700uF Elna caps that might be good for C1552 and C1542, although C1542 is specified as 5500uF, so I was wondering whether a 6800uF would be better to use in that position and then use the two 4700uF Elna caps to replace both C1552 and the 3000uF C1562. I am open to suggestion as to what type of caps to use?

[RogerEvans]

2.5msec period for the invertor would be 400Hz. A lot of Tek equipment has the supply specified as 40 - 500Hz which included US and European mains and aircraft wiring which used to be 400 or 440Hz. Anyway sounds like you have it fixed now.

Roger

[PJL]

The -0.7V across C1562 can only really be explained by a faulty capacitor or rectifier but its worth noting that Q1564 collector is supplied from the +5V rail so I am not sure the -8V would work if the +5V is not correct.

[WaveyDipole]

Originally Posted by RogerEvans

2.5msec period for the invertor would be 400Hz. A lot of Tek equipment has the supply specified as 40 - 500Hz which included US and European mains and aircraft wiring which used to be 400 or 440Hz. Anyway sounds like you have it fixed now.

Roger

That makes sense as from the manual option 7 appears to have been designed for marine and aircraft use. In fact, I missed a line at the bottom of one of the pages which does state that the frequency of the inverter is approximately 400Hz. I guess the online calculator got it wrong!

Originally Posted by PJL

The -0.7V across C1562 can only really be explained by a faulty capacitor or rectifier but its worth noting that Q1564 collector is supplied from the +5V rail so I am not sure the -8V would work if the +5V is not correct.

Yes, I did notice that, however the regulation of that line depends on U1554A which presumably needs to be correctly fed with -8v.... So this seems a bit of a catch 22 situation. I'm not sure how that is balanced out but clearly there seems to be an interdependence on the circuits controlling these two lines.

[WaveyDipole]

BTW. does anyone know where to get the plug/cable that connects to the 12v/24v input socket on the back of the scope?

[MrBungle]

Try these guys: http://www.sphere.bc.ca/test/index.html

Watch out though because everything ends up super expensive. I got a quote for an attenuator module from them and by the time it got to me and I was mugged for fees, courier, import I'd be down £40.

So I just bought another scope off eBay.

Quite happy the DC input on my Philips scope is a bog standard DC jack.

Actually might be worth retro fitting it with Anderson power poles or something else.

[DaveVoorhis]

Try here: http://qservice.eu/

A bit cheaper than Sphere, if they have what you need.