Big Farnell PSU SCR control circuit problem

[Chris Wilson]

Once again my old Farnell 60V @ 50 Amps linear bench supply has given me trouble! This time (and just possibly this issue could be behind previous other issues...) it's the SCR control circuitry that controls the 240V 50Hz mains input to the huge transformer's primary. I am getting little or no voltage on the secondary or the primary of the mains transformer, so after much thinking what would be safe I have fond there's virtually NO rigger waveform to one of the two big SCR's. Disconnecting the gate of the one that has no trigger (measured at the control board) made no difference, nor did swapping in a known good transformer T1 one from a similar supply. I see no short to ground between the board connector strip (board removed) and the cut right by the SCR gate wire. I am now at a loss as to why I get a very healthy 50 Hz sine wave on one of T1's secondaries, the one that feeds pins 8 and 9, but virtually nothing on pins 6 and 7. Can anyone point me to other possible issues please? When probing SCR control board terminals 8 and 9 the PSU will suddenly give some output, which immediately decays. The 150 Ohm R27 gets mad hot when it's run, even with the gate of SCR2 disconnected. Thanks. The schematics relevant are at

http://www.gatesgarth.com/H60001.jpg

and

http://www.gatesgarth.co/H60003.jpg

[dave cox]

I would expect to see 100Hz 'pulses' aligned with zero crossing on the outputs of T1. For each 1/2 cycle only one of the SCR's will fire (other reverse biased). The stream of pulses will stop when the series regulator has enough voltage accross it to stop it dropping out. Strange there is no flyback diode on T1 primary ?

dc

[Chris Wilson]

I have to admit I didn't do things in the correct order and now, looking at the voltages on the IC's I see 24V across pins 4 and 8 of IC1 Eeek! I have replacements for both IC's but need to know WHY this voltage is so high. Measuring across Z1 and Z2 I am seeing 15V and 10V respectively. Thanks Dave!

[cmjones01]

24V across the power pins of IC1 should be fine, and looks consistent with the circuit diagram. IC1 is a 4558 (or similar) dual op-amp, so 24V is just the equivalent of +/-12V: no sweat for an op-amp of that era.

If you're getting no trigger pulses at pins 6 and 7 even when T1 has been replaced, I'd expect that either D8 is open circuit or the SCR connected to those pins has gone short-circuit gate-to-cathode.

Chris

[dave cox]

Z3 @ 5V1 open circuit ?

[Chris Wilson]

No pulse either side of D8, and no pulse with that SCR gate disconnected either. the pulse i see on the other gate is at 50hz and looks like a sine wave, and I can post a capture. I see the same waveform on R21 too, and other pins of IC2. Thanks will check Z3 later!

[dave cox]

A painful grounding problem in there! I would disconnect both SCR's (all 4 wires) - the connection to mains is scary !

I would think about putting the scope ground on pin 13 Should see +15v, -10v and +5v on IC2. i don't see anything else nailed to ground but please check!

dc

[Chris Wilson]

I have been concerned about grounding and mainly have used a battery powered hand held scope. But I will disconnect both SCR's, which mechanically is easier said then done. someone who i inherited the schematics from shows a saw tooth pattern at, I think, pin 5 of IC2. Am I safe going looking for that with a "normal" none isolated inputs mains scope, or my USB scope from which I can capture and post patterns Dave? Thanks for the heads up!

[Chris Wilson]

Originally Posted by dave cox

I would expect to see 100Hz 'pulses' aligned with zero crossing on the outputs of T1. For each 1/2 cycle only one of the SCR's will fire (other reverse biased). The stream of pulses will stop when the series regulator has enough voltage accross it to stop it dropping out. Strange there is no flyback diode on T1 primary ?

dc

Is there any reason at all, other than a faulty transformer or a short to ground on the gate wire of one SCR, for only one secondary to show a waveform? I can't get my head around that... Thanks

This shows the waveform from pin 8 on IC2 with the scope ground to chassis ground, maybe that is my problem, the scope ground??

http://www.gatesgarth.com/R21-to-IC2-pin8.jpg

[dave cox]

It looks like everything is floating WRT chassis ground.
The SCR control board center rail is 'almost' at V+ output (less what's dropped accross the current sense resistor R98). The main output looks to be floating so it 'should' be OK to put a scope ground anywhere - BUT that's one very big PSU is you get it wrong !

The SCR control board creates an 'almost' split rail supply from the small transformer but using 1/2 wave with center tap connected to the -ve rail. I guess they did this so VT1 / VT2 take the 100Hz unsmoothed dc and, via IC1, generate the SCR firing pulses at the right moment. This can be inhibited via the other 1/2 of IC1 (monitoring the voltage drop accross the series regulator). A second inhibit is VT6, which can crowbar IC2's power when the fuse blows. Probably

dc

[Chris Wilson]

OK, I am seeing how it works a bit better now, thanks Dave. Do you think I could put a separate variable bench supply across the (disconnected) terminals 5 and 13 to mimic the drop across the pass transistor block to see if the SCR board responds correctly? Once I get a sensible waveform from VT2? Thanks for the help!

[dave cox]

Mmmm, normally a dummy load would be a good choice, so it pulls enough current to get the regulator working, do you a any 1KW resistors ?

Failing that a 9V battery / pot between pin 5 and 13 might be OK, after isolating.

dc

[Chris Wilson]

Well, some progress, but it brings more and different questions, fixing this thing up is becoming obsessional, but I need it and it's a learning curve, just so long as the more technically astute folk on here don't get too bored! I couldn't find any real issues with the board after doing all the tests suggested, so I put in a working sister H series supply, a 60V @ 25 Amps one, and the supply seemed to work just fine! BUT, when I swapped in the control board from the troublesome supply the volts meter immediately went off scale, even with, I think, both coarse and fine volt knobs on zero. So it looks like the control board from the 50 Amp supply is at fault.


What I just cannot get my head around is why, on the 50 Amp machine, I see no waveform at all on one of the secondaries of T1. I may check what can be seen on the smaller working supply. Is there any way it is correct for only one secondary to show a waveform? What I will do next is look at the working supply. In the faulty one the wire wound R27 still gets mad hot, and the base of both VT4 and VT5 show around 0.7 V or more on a multimeter, but whether it's pulsing I don't know, maybe I should scope it? With more correct probing methods I DO see a saw tooth wave form on the collector of VT2, which I guess is something.

On the main general schematic why does only SCR2 have a resistor and cap across it and not both? I assume it's a Zobel network? What if C25 was shorted? It's SCR2 that doesn't seem to get a signal to the gate, but there's still no apparent signal on either pins 6 or 7 of the SCR board output connectors even with the gate disconnected at the SCR2? I believe BOTH SCR's should fire, never just one?

Thanks for all the help! Swapping boards is OK, but I have a real fear of ending up with two knackered PSU's by trying those tricks. The boards are not very good quality, so removing components or lifting legs usually results in a track coming off the substrate. The control board for the troublesome supply is already patched with fine insulated leads to replace some tracks that were blown when I got the thing, years ago, and it's getting worse with every de-soldering job

The Control Board schematic is at http://www.gatesgarth.com/H60002.jpg

[Chris Wilson]

I see now the Zobel network or whatever R99 and C25 are, actually sit across BOTH SCR's.

[dave cox]

Both secondaries should show something similar, assuming no load. No ratios for the TX are given but it seems sensible the 2 secondaries are near identical.

If only 1 SCR was getting fired then the PSU might still work up to a point - depending on the type of transformer (a torroid would likely just blow the fuse) but this is another area where my knowledge is rather lacking The SCR's will fire alternately on + / - half cycles andwill block on the other half of the waveform (but see below). I think the SCR's will only fire near zero-crossing rather than switch later for a kind of PWM.

The complication on all this is the control system. The SCR's will ONLY fire as needed so if there is no load on the PSU then they will fire VERY infrequently to top up the main filter cap to a suitable voltage. At max voltage / max current I imaging they will be pretty much conducting on every cycle !

dc

[Chris Wilson]

When not playing up even with no load, the V control ramps the meter voltage up and down, as does the smaller sister supply, but now I have measured the voltage on the primary and secondary windings of the main transformer I see just mV. Yet I see about 10V on the voltmeter of the supply. High frequency switching of the primary? To test your last suggestion I will put a small load on the faulty supply, but I suspect I am now looking at an issue with the control board PCB. I'm slowly getting a much better grasp of how it works, but sadly my diagnostic techniques, especially with something with so much floating, is failing me. thanks again Dave!

I spent 30 minutes not having my morning cup of tea, but annotating the Control Board schematic to help me and anyone else understand the interconnects. It's at http://www.gatesgarth.com/H60002.jpg You may need to refresh your browser to not show the previous version! It must be doing "something" to stop the SCR board firing properly, as the Control Board from the other machine gets things working, albeit, probably because it has some component value changes for the 25 Amp one, the max voltage won't go very high.

[dave cox]

The two control boards do interact.

As the main control board demands more voltage the voltage drop across the series regulator (big trannies) falls - this turns on the SCR control board and it connects mains to the transformer primary. As this propagates through the secondary / rectifier / main filter cap the voltage across the series regulator increases until the SCR control board switches off. I don’t think the SCR controller does anything clever other than turning the mains on/off in multiples of 10mS – that is ½ cycles of mains.

I have seen a strange consequence of this in similar switching systems. At a critical amount of load current you can end up with massive asymmetry of switching but slowly changing form one ½ cycle to the other. I think I posted a capture of this on a storage scope …
dc

Memory not quite as bad as I remember :
here

[Chris Wilson]

Grabbed a bit more time on it this morning. If I put the identical Control Amplifiers Board in the faulty PS it works, but the most volts it will give is 3%V even with the fine and coarse V controls fully advanced. Now what I did do is replace all the pass transistors and the two pass driver transistors. IO went from the original all 2N3055 devices which are 60V @ 20A to all MFJ15003 devices which are the same body type but 140V @ 20A. I was told they would work as a direct replacement. But I am now wondering if they need more drive to the bases than the 2N3055's and the supply can't turn them on hard enough? Have I messed up swapping them, the 2N3055 seeme marginal in a supply I usually run at the full 60V output. Thanks.

[See_Mos]

No problems fitting the MFJ that I can see, the HFE is higher than the 3055 so it will require less driving current. I'll have a look at the other information to see if I can come up with something.

[Chris Wilson]

Thanks for the reply. I have now tried putting 3 known good boards in and whist it now works the max V out is still about 35 to 38V. I have wracked my brains to think what else I unsoldered or fiddled with changing the pass transistors and suddenly I remembered a stud diode on a small heat sink that holds the two MFJ driver transistors. It is supposed to be in parallel with the main pass transistors, I will check I haven't screwed up with that, but can you tell me what its function is please?

It is shown on the overall schematic as D23, but doesn't appear on the pass transistor, pass driver transistor schematic. it *IS* present in the PS though, but conveivably i may have wired it back up incorrectly somehow. I have to now assume that although functional with three different boards in it that the low output voltage is not related to the boards but something wrong in the main body of the supply.

Overall schematic; http://www.gatesgarth.com/H60001.jpg

Pass transistor block schematic: http://www.gatesgarth.com/H60005.jpg