Help Requested with diagnosing issue with Farnell Tops3d Bench PSU

[danielhall]

Hi All,

Long time lurker here, I am a hobbyist at best with most of my experience being with microcontrollers.  I am trying to build out my lab since getting some space and I have an old "Farnell Triple Output Power Supply - TOPS3D"  It's not super accurate but I guess I could callibrate it later.

The Issue I have is that the overload warning led is constantly lit for the +17v rail.  The -17v and the 6v rails are both fine.  The overload led stays on for a good minute after powering off, so im presuming it's getting fed by a cap somewhere.

I managed to get the schematics and so I started taking measurements.  I was initially looking around U1 (Op Amp), VT5 NPN BJT, D12 1n4003 and C6 27p Electrolytic.

Ive attached the PDF with the measurements taken in red.  For ground I used the ground output on the front panel (the one for the +/-17v rails (im a bit new to dual rail supplies).

Also attatched an image of the PSU (not mine but identical).

Looking at the measurements, the 0.41v going into pin2 of the op amp ("In-" according to https://www.ti.com/lit/ds/symlink/ua741.pdf?ts=1612518239742&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FUA741) stands out due to it being so far off the 7.97V of the Pin 3 (In+).

Could anyone advise on where I should be focussing?  I would really appreciate any pointers/guidance.

Thanks

[CJay]

Either R21, R22, P5 or IC1

Try adjusting P5 to see if you can make the 0.4V change

[danielhall]

Hi, Thanks for taking a look.

Adjusting P5 to either extremity has no effect on the 0.41v so something must be pulling it low (mostly)?

I guess next step would be to de-solder the R21, R22 and check with an ohm meter, same for P5?

If either are miles off the schematic value then replace, otherwise order a replacement IC1?

Does that sound like the right order?

I don't keep anything near those values (or wattage most likely) so will have to order whatever I need.  I guess it makes sense to test what I can test after de-soldering.

Ordering the Op-Amp is no problem, any idea on the likely wattage of those resistors?  1/2W?, i guess it would be similar for the variable resistor?  Ive included an image of the components

Many Thanks

[Ian.M]

In most cases you can initially check resistors without having to desolder them.  If the resistance is lower than expected the cause is often a silicon junction shunting the resistor and reversing the DMM leads will get you close to its actual value.  If its higher than expected you know the resistor is bad no matter what's shunting it.  Desoldering is only needed when you aren't sure whether the resistor has gone low or the external circuit is shunting it, and for through hole parts, you can often get away with only desoldering one pin, if you can be reasonably confident its got enough hole clearance not to short to its pad while you're probing it.

[danielhall]

Thanks Guys,

I got values lower than those listed on the schematic, but to be sure I desoldered and tested them off of the board.

R21 got 83k (82k on schematic), R22 got 221k (220k on schematic) and the P5 pot gave a value from 0.01k to 49.8k at it's extremities (single turn pot)

Put them all back on, and confirmed I was getting that 0.41v on pin 2 of the op-amp still.

So I guess next step is to order a replacement ua741, anything else worth checking/ordering while im waiting for the postman?

Thanks for all your help with this, it's a good learning opportunity for me

[CJay]

I'd also check D12 before placing any orders, wouldn't worry about the resistors/preset being slightly off, they'll be fine.

[danielhall]

Ah, so I have never used the diode test function on my cheapo multimeter (UNI-T UT33C+) but it is showing as OL in both directions after de-soldering it.  Im not sure how much I trust it so as a sanity check I threw it on a breadboard with an led, 220R and 5v.  In the normal direction (silver ring cathode) in series before the led.  Led lights up as expected. 

Reverse just the diode, and led no longer lights up, so appears to be working.  It shows about 0.07v at the cathode (of the diode)  in this arrangement.

What I did find wierd was testing it via the ohm meter.  It gives a different value depending on what range im set at (440 at 2M range, 184 at 220k Range and 18 at 20k range).  But the resistance dances about and eventually goes to "OL" after a few seconds on all ranges other than 2M.

Im using those spring test lead clips so it's not moving during all of this.

But it seems to work as a diode.  Maybe I should order a replacement anyway?

[Ian.M]

The circuit does *NOT* make sense.  For it to make any sense, the OPAMP (IC1: uA741) must have a reference voltage at its -in.  D12 cant be a 1N4003.  For the circuit to work it must be a 12V Zener!  Please post a good closeup of D12's markings.

Was there any evidence of previous bodgery in that area?

Also, disconnect pin 2 of the OPAMP.  if the voltage on the track goes up significantly, the OPAMP is probably bad.

Edit: see below

[danielhall]

Hi, Close-ups attatched.

The markings on the diode are "1n4003 8325" and it's a 1n4003 on the schematic.  Not sure if the schematic is original but is from 1981.

There does not appear to much in the way of bodgery, solder joints all look similar (old), but it has clearly been opened before (odd missing screw, seals broken etc).

I'll have a go with the op-amp.  I guess it's best to de-solder, bend the pin up and re-solder it to test.

Thanks

[Ian.M]

Ahh I see what's going on now.  I missed that the tail of the potential divider R21, P5, R22 goes to the -17V rail, so provides the reference.  1/4 of Vout+17V is compared with Vout/2.    My comments about it being impossible are incorrect - that circuit should work correctly if all the parts are good and D12 is fitted with its cathode (band) towards pin 2. 

I'm currently suspecting R18 (1 ohm) has gone open circuit.  Check that before going any further.

If R18 is OK, then you may wish to remove the OPAMP, though I'd check VT7 first - an out of circuit diode check both ways on each junction should give you a fair idea of its condition.  If its OK (and reconnected), with the OPAMP removed, you should get 0V out on the +17V terminal, 0V on its pin 3 pad and approx -0.7V on its pin 2 pad, clamped by D12.  If not, the pass transistors must be leaky.  Also, while its removed,  if you apply a voltage (with respect to local 0V) via an extra 470R resistor to the pin 6 pad, you should get approx 3V to 4V less at the output than at pin 6, if the voltages on the diagram are to be believed.  You could use 2x PP3 batteries to test that.  If  you get an output and the overload LED goes out, the OPAMP is bad.

[danielhall]

Ok, so I got R18 out (it's huge with the pins soldered into these tubes) and it appears ok, shows 1.1ohms on the DMM.

Im now on hold as ive somehow misplaced the 1n4003 diode so need to order some.  Once that turns up (or I find it), I will try VT7 next, otherwise remove the op amp.

Thanks

[Ian.M]

D12 is non-critical as its only there to stop the OPAMP -in being dragged negative of its Vee (-V supply) pin by more than one diode drop.   Any small signal silicon diode or <200V <1A silicon rectifier diode will work there.   Don't be afraid to substitute a 1N4148 or similar in the D12 position in this specific circuit, as it only has to handle a few mA at under 30V.

I'd have measured R18 in situ and only desoldered it if I got a reading over about 1.5 ohms or under 0.7 ohms.  See my comments reply #3 above.  To get reasonably good at repair work you *need* to develop a feel for what can be checked in circuit without desoldering, and at least for through-hole parts, the skills to wherever possible desolder a single pin to isolate it for testing without having to actually remove the part.

If a part cant be checked in-circuit, you need to assess the risks of PCB damage if you remove it for testing, balanced against the odds that replacing it will solve the problem, and always consider if there is a less invasive test or check that should be done first.

[danielhall]

Thanks, thats something I will take on board, certainly easier than de-soldering everything, though was pretty quick with an smd rework heatgun.

Ive taken VT7 off however, and all possible combinations simply give "OL" when using the diode test on the DMM.  Also OL when checking resistance.

For the diode test I did:
B+ E-
B+ C-
B- E+
B- C+
E+ C-
E- C+

The part is a 2N3053 and it's an NPN so the first two tests should have shown a voltage drop?

Looks like we found the dodgy part?

[Ian.M]

Yes, if VT7 has failed, the OPAMP would work an order of magnitude harder trying to bring up the output voltage via its driver VT6 and visa versa if VT6 had failed, in both cases turning on the overload LED.  However the schematic shows VT7 as a 2N3055 which is a big two pin TO-3 package transistor which will most likely be bolted to a heatink or the chassis on a mica washer or silpad to isolate its can, (which is the collector) and a solder tag or other ring terminal on one of its mounting bolts for its collector wire.  Its *NOT* a  2N3053 in a TO-39 8mm three lead metal can package - that would be its driver VT6.  To test the 2N3055 I would advise isolating its wiring at the PCB end^* rather than dismounting it. 

I'm not confident in your multimeter's diode test mode as you reported problems with it earlier when testing D12.   Please try a functional test of the transistor in your breadboard, extending its leads as required.  Use the LED + series resistor, fed from 5V, in series with its collector, ground its emitter then apply base current via a 1K resistor from +5V.  The LED shouldn't light at all till you connect the base resistor, then should light at full brightness and the transistor should then have under 0.3V drop across it from collector to emitter and about 0.6V to 0.7V Vbe. It would also be worth practicing with your DMM's diode test mode on a selection of known good diodes and different colored LEDs to become familiar with its limitations.

* Exception - if you find solid core wire wirewrapped on a square pin without solder *NEVER* disturb that type of connection unnecessarily as they cant be remade without a special wrap tool and will need the old end cut off and the wire stripped back to expose enough to remake the joint.  If there isn't enough slack, you may have to replace the wire.  Its possible to bodge it by cleaning the wire and the pin, manually wrapping a couple of turns and soldering it but it always looks fugly, sticking out like trailer-trash in a Walmart dress at a Paris fashion show!

[danielhall]

You are of course right, and damn I feel stupid.  I must have go turned around as I was struggling to find VT7 on or near the board, so I was like "hmm, VT7, nope not there, ah there it is VT6".

I believe I may have found VT7, there are 3 of them of that package size, need to trace which is which.  Weirdly it is outside of the case, nestled in the huge heatsink on the back, picture attached for giggles.  They are connected to the PCB by wires , they look like strange connections, is this what you meant? (another image attatched)

I have to agree with the Diode test, as it's a new (and cheap) multimeter, It is yet to earn my trust.

I may as well test VT6 on a breadboard before tackling VT7. 

[danielhall]

Ok, So VT6 works on the breadboard, no LED lit until the Base is connected to +5v via 1k resistor.  I am showing 0.490v at both the collector and the emitter (with the other end at +5v).

Diode test on multimeter is seemingly useless.  tired with a standard 3mm led, 5mm led and a 1206 led.  Just get OL in both directions.  I need to order me a set of normal diodes.

Im working on building Ben Eaters 8-Bit breadboard computer, so I only have the components I needed for that.  I unfortunately gave everything away that I had previously when I moved house and gave up the hobby, so slowly building it back up.

Again, thank you for your patience

[Ian.M]

I'd expect red LEDs to read as diodes on a DMM diode test range as their Vf at low current is almost invariably under 2V. (Some cheap DMMs will only read Vf up to 1.999V)  Other colors typically have higher Vf, possibly over 3V for some types of blue and white LEDs, so will only show up on DMMs that support a higher test voltage on their diode range.

As its failed to read a diode and a transistor that functionally tested good, and displays OL i.e. open circuit even when the junction is known to be working and forward biassed, I therefore believe your DMM diode test range is totally borked.  Check the DC current ranges with a power supply and a known resistance load in series with the meter, calculating the expected current in advance so you don't overload it, as depending on the DMM's design, its possible a blown current range fuse may also have taken out the diode test range.

[danielhall]

Yeah just had a quick look, the 200ma 250v fuse inside the multimeter has blown.  Which is surprising as ive had it less than a week and didn't think I had put much through it at all.  Will order some replacements and give it a good test.  A proper DMM is on my future shopping list.

[danielhall]

Ok, I have found another two 2n3055's hidden inside the outer rear heatsink (with the 1st two sandwiched between the two heatsinks).  Out of the two new ones, i have identified which is VT7 as it's the only one that connects to VT6 and R15.

Im not sure how best to go about disconnecting it as the two pins have those weird rings on the pins (same as the image I posted).  The base is only connected via a small nut/bolt which I could easily disconnect.  Would that take it out of the circuit enough to test (once I get a fuse)?

Thanks

[Ian.M]

Your meter doesn't have a separate jack for low amps ranges, so its very easy to blow that fuse if you turn the knob to or through an inappropriate range while connected across any voltage source that is low enough impedance to deliver >250mA.   Get in the habit of *THINKING* before turning the range knob past any of the white dividing lines round it, and whenever possible avoid changing ranges while the probes are connected to a powered circuit, as even if you don't go through a bad range, doing so when probing significant voltage or current can decrease the life of the range switch, which as a PCB switch integral to the meter PCB isn't really reparable.

N.B. Multimeter fuses are pretty special and usually expensive.  Do not substitute ordinary glass fuses as if you are probing a high energy circuit, and make a mistake that shorts it, a glass fuse can explode violently enough to rupture the meter case and also fail to quench the resulting arc.   At best you'll need new underwear and a week to stop jittering, at worst you can suffer a life changing injury or even die!

If it has ceramic body fuses in it replace like for like.  If it has glass fuses I strongly recommend getting fast acting (F) HRC fuses of identical current rating and the same or higher voltage rating.  HRC fuses are usually ceramic bodied and sand filled to adsorb the vaporized fusewire and quench the arc.  Specialist HRC fuses for more up-market multimeter applications are usually uncommon sizes to reduce the temptation to make inappropriate substitutions, and melamine bodied to be even better at containing explosive arcs than ceramic bodied fuses.

Re: VT7,  The bolt + tag will be the collector not the base.  Can't you disconnect the PCB end of its wiring?  The wires should slide off those PCB terminals fairly easily if you heat the terminal with a heavy bit in a decent soldering iron, then be possible to coax back on with small  pliers so you can resolder it.  If possible avoid disturbing VT7' mechanical mountings.

[danielhall]

Thanks, the fuse inside is a 5x20mm 200ma 250v white ceramic job.  Ive ordered one from the manufacturer.  It's an £8 multimeter, im guessing they did not put anything too special in there, but figured I would get the right one from them.

I'll give the VT7 a go at the board side and see where I get to, otherwise onto the op-amp I guess.

This exercise is showing me I need to spend a lot more time getting back into things and just taking my time in general, aswell as holding a selection of common parts and non-junk equipment. 

Thanks

[danielhall]

Hi, So a little bit of an update.

I received the replacement fuses for the Multimeter, fitted them but it's still useless, still shows OL on both sides of a selection of new 1N4003's.  I was able to borrow an old Mastech multimeter from work thats seen better days, but it shows a voltage drop of 0.65v in one direction.

Used that to test the (out of circuit) 2N3055, and it shows a 0.675v drop when Anode is connected to base, and the cathode is connected to either of the Collector or Emitter, all other configurations show nothing.  So that looks all good?

I was also able to obtain a replacement Op-Amp, swapped it out and still has the overload LED lit. 

One thing I did notice, and it may be unrelated, but when I turn it on, it seems to take a long time for that channel to get to the target voltage, like it takes over a minute to get upto 18V from cold.  After that it's fine.

This only effects the rail with the overload issue.  Both the -17V and the 6V rails reach their target almost instantly.

Could this be a Cap issue?

Thanks