TTi EX354T Power supply - CRACKING!

[weejag]

I have spent days trying to get a TTi EX354T Triple Power supply working.  I bought it with an obviously blown NTC (negative temperature coefficient) thermistor (RT1).  I discovered that the main 'boost' Power MOSFET (Q5) was open circuit, and that D54 boost diode was also blown. 
Despite many hours of searching I have found no other faulty components (although I have really only focused on the primary side of the main transformer) .  However, when I apply power loud 'cracks' are heard at a frequency of about 1-2Hz.  It sounds like a capacitor discharging a high voltage but I can not find the cause.  I have done most of the testing with the board not connected to the front panel but when it is then one circular dot/led lights up on each section of the display in time with the crack.  I only persist for a couple of seconds then cut the power as I am afraid that other components are likely to be adversely affected.  This makes it difficult to get any more clues.

Has anyone else seen this problem? I would be grateful for some guidance as I have just about reached the limits of my abilities.
Thanks.

Regards,
weejag 

[TomKatt]

Have you taken any voltage measurements?  Or looked with a scope?  I follow pretty basic troubleshooting, checking levels and ripple first.  I also wonder if you’d see anything that aligned with the sound you hear.

[Martin72]

Hi,

I discovered that the main 'boost' Power MOSFET (Q5) was open circuit, and that D54 boost diode was also blown.

How did you find out about Q5, you can't test a MOS-Fet like a transistor.
In fact, IC8, Q5 and D54 are components of the PFC circuit.
This makes it somewhat difficult to pinpoint the fault, as everything is interdependent.
But there is the circuit diagram, I'll look over it again later, maybe I'll come up with something.

[nctnico]

I'd start by applying power to the auxilary supply voltage which feeds the primary part using a PSU. That way you can measure signals in the primary part WITHOUT connecting it to mains. The next step is to put a low voltage like 30V DC to the high voltage DC and see if something appears at the secondary side.

This power supply looks like it is a forward converter with Q4 and Q7 driving the transformer. The use of magamps at the secondary side is a clear sign.

I think the PSU is in hick-up mode. Either there is a short / overload at one of the secondary output voltages or an overvoltage condition. Q6 and Q10 seem to form some kind of undervoltage detection circuit.

PS: I'd fire the person who drew the schematic 

[nctnico]

Hi,

I discovered that the main 'boost' Power MOSFET (Q5) was open circuit, and that D54 boost diode was also blown.

How did you find out about Q5, you can't test a MOS-Fet like a transistor.

You can but differently. A MOSFET has an internal body diode which you should be able to measure between drain & source. The gate is OK if it has no continuity to source and drain.

[weejag]

I used a DCA75 (PEAK) to test MOSFET, and I'm sorry, it was short circuit so was easy to find with multimeter.  I replaced all 3 components mentioned (DCA75 correctly identified new MOSFET and D54).

It is difficult to take measurements due to my fear of further damage.  However, I found that using a VARIAC at 120v (rather than the usual 240v here in the UK) the crack was not as disconcerting.  Putting in an incandscent bulb current limiter stopped the noise and allowed some measurements to be taken but I'm not sure how relevent they are.  The rectifier produced an appropriate 100Hz fully rectified unsmoothed trace at appropriate voltage.  There was a 15-20v 'ramped' voltage going to Vcc of IC8 and the output to the MOSFET gate (pin20) did show some control pulses but not clean? I was surprised to see that the Vref produced square waves with about a 10% duty cycle.  Having checked many other things I suspected that IC8 may be faulty so I found a replacement but it made no difference.  I put a socket in for this chip and later for the 2 logic chips (replaced with known good ones even if originals were ok - yet to test).  The 7815 only put out 1v with the current limiter in circuit but about 15v when 'cracking' was allowed (although it was a very rough output).  I replaced the 7815 anyway.
I have checked all resistors, some capacitors, and the diodes.  I have removed and tested Q4, Q7, Q2, Q3, Q22, Q23, Q6, Q10, L3, L9, L10, TX1, D67, SC1 all of which are fine.
Really pulling my hair out now!

[weejag]

Ah, right. Thanks NCTNICO.
That sounds like a good plan.  Unless management has plans for the weekend I will try your strategy.  I suspect I may not be allowed to play today though 

[Martin72]

PS: I'd fire the person who drew the schematic 

You can underline that, you get a seizure after a short time.
Apart from that, nctnico has already said it, I wouldn't do it any differently.
Apply an auxiliary voltage, check whether switching signals are present at the gates(Q4,Q7) slowly run up a rail voltage (feed in behind D54 to suppress the PFC chopper for the time being), see what happens at the output.

The gate is OK if it has no continuity to source and drain.

Or it's completely open, as we've had before.
But he tested it with the DCR75.

I was surprised to see that the Vref produced square waves with about a 10% duty cycle.

Some PFC ICs have an “undervoltage lockout” circuit, which could explain the interruptions in the Vref.

[asis]

Hi,

When the PSU is turned on, through the filter L3A/L3B (and, in fact, the associated inductance), during transient processes, capacitor C15 (+V) is charged.
This voltage should be enough to reach the operating mode of IC8 (+Vcc), if the triac SC1 did not work, i.e. mounting "OR" from optons (IC12; IC13; IC14) of overvoltage protection did not open the triac.
Next, IC8 is powered from the auxiliary winding TX2-F (pin18;19) D6;C93* !;D56.

[AVGresponding]

I have spent days trying to get a TTi EX354T Triple Power supply working.  I bought it with an obviously blown NTC (negative temperature coefficient) thermistor (RT1).  I discovered that the main 'boost' Power MOSFET (Q5) was open circuit, and that D54 boost diode was also blown. 
Despite many hours of searching I have found no other faulty components (although I have really only focused on the primary side of the main transformer) .  However, when I apply power loud 'cracks' are heard at a frequency of about 1-2Hz.  It sounds like a capacitor discharging a high voltage but I can not find the cause.  I have done most of the testing with the board not connected to the front panel but when it is then one circular dot/led lights up on each section of the display in time with the crack.  I only persist for a couple of seconds then cut the power as I am afraid that other components are likely to be adversely affected.  This makes it difficult to get any more clues.

Has anyone else seen this problem? I would be grateful for some guidance as I have just about reached the limits of my abilities.
Thanks.

Regards,
weejag

Sometimes it's the simplest of things...

I recommend you have a close look at the main power switch; I had to replace the one on mine when the contacts eroded to the point they were audibly arcing when I switched the unit on and off.

[weejag]

Well, this has been interesting.
I put 18v between the two diodes D6 and D58 to power IC 8 and the 15v regulator.  This worked well and I got appropriate signals on the output to drive Q5.  IC1 put out an 83kHz (supposed to be 100kHz).

Measuring the HT line nothing was seen until the suggested voltage was applied to Vrect then the circuit worked.  I limited the current on this line and as this was increased so did the HT voltage up to about 180v. The Q6 / Q10 detection circuit shuts down IC8 if the supply voltage falls below 85v so this was disabled by removing Q1.

As I increased the current limit on the Vrect supply (and the supplied voltage also increased) I thought I was on to a winner as the HT voltage increased. I noticed that two resistors R120 and R121 suddenly became white hot and blew. I had forgotten to current limit the low voltage supply which was could put out 4 amps.  This was quite fortuitous as it made me realise that the only way that both these would blow is if the the thyristor SC1 kicked in.  This suggested the problem was on the secondary side and also explained the hiccups.  I replaced the resistors and current limited the line to 500mA.  Later I realised that having attained a certain HT voltage this supply could be switched off as the supply from TX2-F was able to sustain IC8.

FB8, FB5, FS2, and MA1 and MA2 were removed to try and remove any problem in the secondary side.  Unfortunately the hiccup still persisted so I looked to the transformer. I thought that TX2-G may be short circuit as it measured only 0.0005ohms ( in circuit).  I removed the transformer, cleaned it up and retested. This time it was 0.01ohms but the other ranged from 0.17ohms to 0.05ohms and considering that TX2-G seemed to have the thickest wires I thought this was probably reasonable.  If it had really been 0.0005 and was now 0.01ohm it may have resolved the problem.  It seemed much better when I took the HT up to 340v before it tripped.  Stupidly I decided to give it a proper go and replaced the missing components and plug in the mains supply to see if anything had changed.  The big 470microF 400v capacitor went bang!  I have a replacement on order.  Unfortunately, I still don’t know the answer but thanks to NCTNICO I have had a great deal of fun trying to get there.  What a day!
I’m now thinking that the OPTE must still be working despite the removal of the MAGAMP coils (and that It was not really logical to look to the transformer).  Next move is to remove Q14,15,16 and Q11,12,13 and test them individually but that’s for another day.

Any further suggestions?

[asis]

Due to the fact that during the experiments you used your own Lab PSU and powered the circuit with V+ (20~30V), even with the FS2 fuse removed, voltage on optocoupler IC14 (pin1) could arise through the circuit:
V+ ->R60; D35; Q16; D10; R30.
Keep in mind that Q16 - STP60NE06, has an internal diode between S(3) & D(2), connected in the forward direction and taking into account the total component of the switching thresholds of the zener diodes D35(18V) + D10(6.8V), the voltage level could enough to turn on optocoupler IC14.

[weejag]

Thanks ASIS.
I'd missed that V+ was there at all so I guess that I need to look at those components first.
Getting there!

[weejag]

Actually looking at V+ on the secondary side it is derived from TX2-G rectified by D8 and smoothed by C29/39 so in fact this is disabled by removing FS2.
However, I only realised this after having checked every component in this part of the circuit.  I was only left with the LM358N so again I socketed this chip and replaced it (as I had one available).
I received my new 470uF 450v capacitor and stuck it in.  Having not really discovered any obvious faults I thought that I would have another go with NCTNICO's powering up option.  I got the HT up to 380v and there were no hiccups.  I guessed I might be there. I replaced Q1 to monitor undervoltage, and powered up with the variac.
Eureka!  At 120v I got an HT of about 390v which stayed steady when I went to full 240v AC input.  Sorted!
As expected, when I connected the front panel all went well and it now sits beautifully on a shelf in the workshop.

I didn't expect such a journey but I have learned loads from the experience.  As it was, the main culprit seems to have been the big capacitor C88.  I had gone down many rabbit holes having believed that testing this cap with an LCR meter and getting a good capacitance value and good ESR meant that it was OK.  Clearly this was not the case and I am thinking of revisiting Paul Carlsons Capacitance Leakage Tester project.

Thanks for all those willing to help.  TTFN

[Hydron]

Congrats and enjoy the supply - I also resurrected one of these in the past (free from work due to being faulty) and it served me well for years. In my case one of the logic chips went bad so a reasonably easy fix given the service manual availability (allegedly it was damaged by something falling down into the supply through the vents, though I'm dubious it would have caused that failure mode).