Tektronix 2213A PSU problem

[torquil]

Hi!

I have a Tektronix 2213A with a dead PSU that I'm trying to fix. After I got the scope, it worked fine for about 15 minutes. Then it emitted a bang/pooof and I could smell that something bad had happened... No visible smoke, though. I expected to see some black residue or something somewhere in the PSU or some other place in the scope, but there was no visible signs of anything bad. I have looked at every component. The caps look fine with no bulging, no burn marks. The fuse is still intact.

I removed the three RIFA caps, and they look fine and measure fine out of circuit. I will replace them later on, but for now I'm testing the scope without them. I guess they could have blown at any time, but I'm pretty sure they were not the problem. Could there be RIFA-type caps in the Schaffner filter in the metal casing on the IEC input plug? In that case I could understand that there was no visible trace of an explosion. The input plug is still working, though, so in that case some other components was damaged as a side effect, because the scope is not working now even though 230 VAC is coming into the scope.

I have check that the 230 VAC is reaching into the scope, past the power switch. I plugged the scope power cable into a power meter. When I connect the scope, the power meter reads 1 W. When I switch the scope on, it still reads 1 W...

When I measure the different voltage rails described in the service manual, I get 0 VDC on all of them, with no ripple (TP961, W968, W960, W956, W954). I have checked both with a DMM and another scope.

I then checked the PSU waveforms (I ran the other scope ungrounded for this, in order to use TP940 and TP950 as reference, but I was very careful when doing this). First I tried to use a two channel scope and use subtraction, but I was not confident that I was doing it right. I'm pretty sure I have done the measurements correctly now with the ungrounded scope).

The result was that waveforms 27, 28, 29, 30, 32, 33, 34, 35) were flat at 0V with no significant ripple.

I did a few measurements on the big metal encapsulated filter cap C906. It gets charged to 320 VDC with no ripple. The resistance across it is many MOhms. Measuring its capacitance gives 95 uF, which is fine.

The smaller cap C925 doesn't seem to get charged. I measure only some 100 mV on it (with and without Q9070 connected in the live circuit operation). Reistance across it is only around 100 Ohm (in circuit).

The DMM diode test on each of the rectifier diodes CR901-904 gives expected results with a forward voltage drop around 0.6 V in one direction and O.L. in the other.

I disconnected the connected W9077 which goes to the IRF710 N-channel MOSFET Q9070 on the heatsink. The FET looks fine visually. Could it "blow" without it being visible afterwards? I did some measurements on it following http://electronicsbeliever.com/how-to-know-if-mosfet-is-defective/ . From the guide there it seems that the FET is defective:

Source-drain diode test on DMM: Positive probe on source and negative probe on drain: O.L. (should be 0.4V - 0.9V according to the web page above for N-channel MOSFETs). Reversing the DMM leads still gives O.L.
Source-drain resistance: O.L. for both polarities.
Source-gate resistance: O.L. for both polarities.
Drain-gate resistance: 94 Ohm for both measurement polarities.

These results look quite suspicious to me. Could this be the cuplrit even though I don't see any "burn marks" or other signs of any violent explosion on it?

I don't want to start replacing components willy nilly, so I'm hoping some of you guys could have a think about what is the most likely suspect.

[bd139]

Q9070 is notorious for blowing up on these. Same as 2215 and 2235. See my thread here: https://www.eevblog.com/forum/testgear/tektronix-2235-repair-thread/

May have blown the crap out of the driver transistor and a diode as well and the SMPS controller if you're really unlucky.

The FET can die without too much visible damage and it goes with one hell of a bang when it does.

[torquil]

Thanks for the answer and the link to the other thread!

I put in an MTP6N60E MOSFET as mentioned by David Hess in the other thread. However, there are some other problems:

1) Since the scope didn't work after fixing Q9070, I looked more closely at nearby components. And sure enough, R909 was cracked open and had left a small black mark on the circuit board. It was open circuit. So I replaced it by what I had on hand, a 47 ohm resistor. I will put in a 39 ohm as the schematic demands when I get hold of one.

2) I lifted one leg of each of CR907 and CR908 and both are broken:

CR907: Resistance measurement gives around 4.4 ohm both ways. Diode voltage drop measurement gives 0.004V and is symmetrical.
CR908: Resistance measurement gives around 44 ohm both ways. Diode voltage drop measurement gives 0.043V and is symmetrical.

Regarding CR907: In your other thread you mentioned using a STTH2R06 ultra-fast rectifier for CR907. Was that a good substitute?
Regarding CR908: The service manual says 1N4152. Is it best to order the exact same is possible?

[bd139]

The STTH2R06 worked nicely.

1N4148 should be fine where the 1N4152 is if you have one lying around else order the same.

If you’ve got a 220 ohm lying around solder that across the 47 ohm one. Parallel combo is about 39 ohms.

[torquil]

Ok, so after receiving the parts I have replaced CR907, CR908, Q908, R909 and Q9070.

But when I powered it on, it blew the fuse. I checked it, and for some reason it was a fast blow fuse. The back cover says to use a 1A slow blow fuse (We have 230VAC here in Norway).

I inserted a proper fuse and disconnected the FET Q9070. When I powered it on, the filter cap C906 was charged as expected. However, when I power it up again with Q9070 connected, the magic smoke was released, seemingly from the A6 filter board, and the fuse blew. At least that is what is looked like. I haven't done any measurements yet. I'm positive that I have installed the diodes with the correct orientation.

After the magic smoke was released, no voltage is now reaching the rectifier diodes CR901-CR904 when I power on with a new fuse. So something is wrong on the filter board. Btw, all the RIFA caps were removed before this, and also the single one the the main board. I guess I need to check the inductors on the filter board? Would the inductors be able to create a puff of smoke when subjected to a high current?

I guess I need a lightbulb current limiter now? I would appreciate and tips on how to debug this without blowing more components... :-)

I know that the scope works after the PSU, because it did so when applying 43VDC to the TP940/950 test points.

[bd139]

I had the same problem with the fuse blowing on my 2225. Had me scratching my head for a while that did then the slow blow went in

Getting there! I'd pull the filter board out and check for shorts.

[torquil]

I made a mistake when claiming that there was a problem on the filter board. It works fine. As long as the fuse is intact, 230 VAC is passed to the main board. The smoke must have been the fuse, I cannot find anything else.

Now I have compared my 2213A with my 2215A which has a functioning identical PSU. In both scopes I disconnected Q9070 to compare, since the 2213A blows fuses when Q9070 is connected.

2215A: This does what is expected when checking the waveforms 27/28 in the service manual on my trusty old Tek 465B: it "hiccups" or pulses periodically (about once every second). Waveforms 27 and 28 appear correctly within each pulse (temporarily). Between each pulse, the responses disspear. The pulses can e.g. be seen on pin 10, the output from U930 which goes towards the Q9070 gate via Q908.

2213A: There are no such hiccups. The voltage on the gate of Q9070 is stable (I think it was a few tenths of a volt if I remember correctly). Pin 14 on U930 which is supposed to be a 5V reference is a stable -30 mVDC... compared to the ground pin 7.

I'm suspecting that U930 is defective. It is close to Q908 and CR908 which were blown and replaced, so it wouldn't be a big surprise. I checked that Q928/Q930 are OK, since they are close to U930.

Coarse of action: I have ordered a replacement TL594 together and a socket. The datasheet says UC494ACN but the actual component was a TL594.

[torquil]

After also replacing U930, the behaviour with Q9070 disconnected was the same as for my 2215a, so seemingly correct.

However, when I tried to again connect the MOSFET, it blew again and is now shorted.

But now powering the scope using 42V on TP940 and TP950 no longer works. When trying it I heard a poofff sound, so this attempt must have caused some new damage.

[bd139]

Have you tried hitting it with a cricket bat? I usually resort to that at this point in time. Sorry not helpful I know

[torquil]

I'm quite determined to make the 2213A work. From before I had more theoretical than practical knowledge of electronics, but while trying to repair these things, I'm learning the importance of practical experience :-) For example, I think the old solder flux I used around the IC socket becomes electrically conductive after heating... And I really need to up my game in the "remove IC's from circuit boards"-department. My 2215A is also sick, so I have plenty of opportunities.

In the mean time I have my trusty Tek 465B to use for troubleshooting.

I just hope I didn't do any serious damage when trying to power it with 42V. From now on I will isolate the PSU and its different sections better when working on it to avoid collateral damage. This will also help me to avoid blowing my budget on fuses.

After working on the SMPS I'm left feeling that I would take a linear PSU any day, even with the extra weight, inefficiency and extra complexity of having a fan.

[bd139]

I’ve fought two of those Tek power supplies now. Both exploded on me at least twice. I’m done 

The design is based on a very borderline iffy preregulator and then a relatively simple switching power supply. The preregulator is what tends to explode. I note that in my Philips PM3217, they just used a linear transformer instead of a preregulator! Makes like much easier so I’m with you on that. I have wondered if it’s possibly to just give up on the preregulator and get a MeanWell 50V supply, turn it down a bit and cable tie it to the back of the unit and totally bypass the preregulator portion. Lazy I know but it may work.

A colleague of mine was once in charge of developing small switching supplies that worked off 28V DC. Usually buck converters. Things used to explode on his bench daily. Comes with the territory. Alas I was only responsible for testing then so nothing blew up when I got it

[tautech]

For you guys fighting with Tek preregs the last couple of pages in this thread will get you on the right path:
https://www.eevblog.com/forum/repair/tektronix-2215-scope-repair/

The FET gate drive is marginal to get beyond millar capacitance but careful choice of a replacement FET and some subtle mods can overcome it.

[torquil]

I found only 17 Ohm between TP940 and TP950. I lifted one leg of each of R912, R907 and CR907 in order to divide the PSU in two for further testing. In any case, it turned out that Q947 is leaking:

Base-collector: 0 Ohm
Emitter-collector: 17 Ohm.

I found some discussions about replacements for these, and it seems that it will be some TIP41C could be a good fit. Do these have the standard pinout BCE from left to right (in the photo below)?

If the pinout is BCE, then there is also something wrong with Q946, because the the BC junction I'm getting a voltage drop both ways when using the DMM diode setting.

The transistors I have in there (which seem to be original) are marked as follows:

Q946:

8339 M
476-02

Q947:

K
M 8439
476-02

The M looks like a Motorola logo and the 476-02 coincides with part of the Tek number (which I found in a Tek 2215A service manual, because Q946/Q947were missing from the 2213A service manual I have).

[bd139]

Full part number is 151-0476-02 which comes up as a TIP31C here. TIP41C is a good substitute. Same pinout, higher maximum collector current.

Looks like these are the Royer converter driver transistors. Might be something on the secondary side of that which caused it to overload. Worth checking the resistance of all the secondary power rails in the scope to make sure there's not a shorted capacitor there as well. Also check that converter transformer has no open windings now.

I think the PSU is the same in this as the one in my 2235 repair thread here - full schematic in thread: https://www.eevblog.com/forum/testgear/tektronix-2235-repair-thread/

[Chris56000]

Hi!

All normal bipolar transistors in TO–220 and the all–plastic ISO–220 package like Q946 & Q947 you've shown, are B–C–E reading left–to–right with the tab connected to C in the middle!

The slightly smaller TO–126, like BD131/2, etc., which has the metal surface on the back and is a rectangular flat plastic package, is E–C–B from left to right, viewed with the metal–back surface downwards!

It's unusual for leakage to be such that a meter indicates a diode–drop in each direction, nonetheless both Q946 & Q947 are unserviceable, one is c/e shorted, the other has excessive b/c leakage!

Begin again with two new TIP41Cs and work from there!

Chris Williams

[torquil]

The strange thing is that underneath the board it is indicated "C-B-E", when going from left to right on the transistors in the transistor photo above. See the board photo below. It is looking towards the transistor along the pins, with the heat sink north of the transistor in the photo.

But the board notation has to be wrong, because it does not seem to fit with the schematic. If the board notation indicates that the Collector on Q946 goes to R946. But the schematic says that it should be the Base that goes to R946. If the board notation is wrong, and it is really B-C-E as you say, then it is the Base which really goes to R946, which fits with the schematic.

Also, if the board was correct, the Base of Q947 goes to pin 1 on the transformer, which is not consistent with the schematic. It should be the Collector, which is correct if the correct order is B-C-E.

So everything points towards C and B being switched on the board notation, and the transistor really has the standard B-C-E pinout.

[torquil]

I replaced Q946/Q947, as well as Q908 and U930, and now the PSU works!

The scope now seems to works perfectly except for one thing: It will not trigger from channel 1.

I have tried to clean the trigger source int-switch with isopropyl alcohol in case it did not make contact when selecting channel 1, but I have not yet actually measured that it makes contact.

Any other likely failure points?

Update: The voltages around U310 are getting dragged towards ground, so U310 is probably broken.

Update2: Scratch that. It seems that the wiper on R309 was not making good contact with the carbon track. After some isopropyl alcohol and pot-twisting it is ok now

[DavidSECT]

My Tek 2213A scope stopped working, the little green light was dark. Troubleshooting showed a lack of dc distribution voltages. Probing in the circuitry feeding inverter transformer T948 showed the pulse width modulator (PWM) IC U930 and FET Q9070 starting up but then stopping about once every second. I think this is the same “chirping” that others here have seen.

Eventually I found that the power switching transistors Q946/947 - or one of them - had failed (my notes aren’t clear on this) so, when the inverter circuit started up upon receiving sufficient supply voltage (+43 volts) from the PWM, the inverter failure to start up cut off the PWM circuit.

This is because Q930 and the PWM FET Q9070 start up on an initial feed from Q928 and Q930 and establish +43VDC for the inverter driver circuitry but then the fully running PWM is supposed to get its voltage supply from the inverter primary via T906 and CR920.

I disconnected the DC power distribution jumpers - W955, etc - looking for a short that might be causing an overcurrent shutdown. Later, for replacements to the failed Q946/947 pair I tried TIP31A, TIP31C and TIP41C because they were in my parts box but each type failed as I replaced the DC distribution jumpers. Finally I installed two NTE379s. The scope has been working for over a year so far.

The Tech Manual’s theory of operation is detailed and very helpful. Even so it took awhile for me to understand the power supply circuitry. Comments found online were helpful too but I had to be careful to separate the wheat from the chaff 

Please beware troubleshooting this power circuitry. The tech manual states that waveform measurement instrumentation “must be connected to its ac power source via a 1:1 isolation transformer”. This is because the inverter is floating above ground, so test point waveforms in the TM that reference TP940 must be measured with an isolated oscilloscope. Please also take the precaution of isolating the oscilloscope’s ac connection through an isolation transformer for your own protection when probing this circuitry.