SMPS with too low and fluctuating output voltages

[jogri]

I am currently trying to repair a Tek TDS 1000 scope (bought it as "for parts/repair") that appears to have a broken power supply: All output voltages are too low (~20V avg on 35V channel, 3.5-4V on 6V channel) and are fluctuating heavily between aprox. 50 and 80% of their nominal output (i am lacking both a differential probe and a variac, so those voltages were measured with a DMM).

What i have found so far:
-the driver IC blew at some point, there are still traces of that left on the board
-one Octocoupler that was connected to that IC was broken (i suspect that the IC took it with him when he blew)
-all electrolytic caps are fine (under 2 ohm ESR, capacity still within tolerances)
-one of the capacitors on the LC filter at the input had a capacity of 50nF (was a 100nF PP cap)
-the ICs i could test (Schottky rectifiers & diodes) are fine

Do you have any idea where the problem might be? I am rapidly running out of ideas which parts might be causing that behaviour. The only other suspect i could think of would be the transformer, but since it outputs a voltage that is roughly in the right range, i don't think that the transformer is faulty.

[EPTech]

Hi there,

U1 looks like a fully integrated switcher ( like a TOPSwitch from PI). Those should be easy to come by. I would desolder it and check for a short or low-resistance path between the output and other terminals. That is the most likely suspect.
Als check the clamping circuit across the primary (D6, D7, C28). With U1 desoldered, carefully check the DC bus voltage across C3. That should be around 285VDC. Carefull, it is lethal!
You can also check wether the control voltage for U1 is present across C5. That should be around 12V. It may be that the feed-back terminal on U1 is blown as well, so U2 may still have survived. You can also check this with U1 desoldered.
If all of that check out, you can safely replace U1 and give the supply another go. I recommend using a lght bulb in series with the live wire as a current limiter. When you power on the bulb will flash momentarily as a result of the inrush caused by C3. After that is should go out almost completely and shine dim when you load the supply. Judging from the size supply, a 20W-40W bulb should suffice.

Happy repairing.

[jogri]

Thanks for the tip with the clamping circuit, i'll check it out. I forgot to mention that i've already replaced U1 (it is a TOP244Y, props for spotting that on that picture). Btw, the IC i desoldered (U1) still looked intact after powering it on (no internal shorts).

Do you by chance know what those diodes(?) labeled SG1 and SG2 are? The three stripes and the "SG" labeling are throwing me off. They have a glas casing and it seems like they consist of a 1mm long semiconductor (they have a resistance over 60 MOhm (both directions) and my DMM doesn't show a forward voltage). Are they some form of overvoltage protection, given that they sit across the spark gap on the live wire?

[EPTech]

Hi,

They are GDTs, glass discharge tubes. The spark gaps are the first line of defense, followed closely by the glass discharge tubes, usually followed by the MOV, metal oxide varistor. I am pretty sure the big yellow disk on the bottom right is a MOV. They should all be open circuit.

[tautech]

2 ohms sounds a bit high for secondary side SMPS low ESR caps.
Also grab the datasheet for the TOP244Y and see if it has a typical application circuit as in practice this is followed often.
IME the VCC smoothing cap for the TOP244Y should also be checked carefully to ensure the IC is getting a clean supply. Also ensure the HV DC dropper resistor/s measure correctly for proper startup of the IC. The primary side feedback diodes that supply the running VCC could be faulty/leaky too.

[jogri]

Pretty much all electrolytic caps that i didn't replace (yet) are around 1.3-1.7 Ohm (the smoothing cap is at 1.7), is that still okay for a power supply? Btw, they are all either 105°C Nichicon, Chemi-Con or Rubycon caps, looks like the manufacturer didn't cheap out.

It looks like C28 (the small ceramic cap under the transformer) blew out, the plastic case was ripped open at one leg and it had ~50pF instead of the nominal 47nF.

This is the datasheet for the TOP244, the schematic at page 23 is a pretty accurate representation of my power supply (R5 [on the schematic] is 330 Ohm, a few caps have a different capacitance and it has an additional floating output for the LCD): https://www.power.com/sites/default/files/product-docs/top242.250.pdf

[tautech]

The bulk HV DC cap @ 1.7 Ohms is fine. As you go up through the cap voltage ratings the ESR is always higher.
OTOH the secondary side caps that are trying to smooth say 30 KHz ripple need good low ESR.

Despite being good brands used these TDS1k models are getting a fair age now so they might need a good hard looking at.

[jogri]

Well, this power supply just doesn't want to work... I've replaced all parts that looked faulty to me, but the problems remain or even get worse: C28 (picture attached) had the marking "47k 1kV" on it, so i replaced it with a ceramic 47nF cap. When i do this, the output voltages drop to 0.4V (yes, i have no load attached but the Tektronix manual clearly states that you can/should test the output voltages w/o load).

I assumed that "47k" means 47.000 pF or 47 nF, but the "1kV" left me guessing: a 1 kilovolt rating seems too high given that the leads are only 3mm apart and the "1K" voltage rating for film caps (80V) seems a bit too low for a cap on the primary side of a power supply.

(the cap has 50pF capacitance, but it looks damaged so i don't know if that is its nominal capacitance)

[mikerj]

47=47pF, K=10%, 1kV=1000v maximum.

[jogri]

Well, that explains why it refuses to power up now... Changed the cap, and the scope powered up. It still shows errors for the line trigger on both channels, but apart from that everything works.

Since the line trigger pin on the power supply only outputs a DC voltage without any AC on top of it, i think that i've found the source of the problem.

[Twoflower]

It looks like that U1 isn't attached to the heatsink at all. Either the screw is missed, or a clap holding it to the heatsink.

On the other side the diodes D12 for example are also laking a screw/clamp and they are attached correctly. I think they used thermal conductive glue to keep the devices attached to the heatsink. But for U1 it looks like that there's an air gap.

[jogri]

Yes, U1 had a clamp but that picture was taken shortly after i replaced it. Getting the clamp into place is a PITA, so i did not put it back on while i was repairing it.

It seems like U1 originally had a clamp and thermal adhesive but since somebody replaced U1 before i bought the scope the adhesive was completely dried up.