Mastech HY5020E 50v 20a Bench Supply: Economical to repair?

[Andy Watson]

As above I've always been puzzled why Q401 just looked to be a linear reg with no pre-regulation.

Would it make more sense if the label at the bottom of HY012 read "U+". I can't see where S4 connects to, but all the other sections appear to be referenced to U+.

[albert22]

I dont know where S4 is. In the schematic U301 is referenced to U- and the resistor divider R302/R304 should set the output of the switching regulator to a fixed 61V approximately.
So I think that referencing it to U+ will not work with those values of voltage FB.
As grumpydoc said the power at the MOSFET will be huge (1100W). No way this can work, a pre regulator is required.  I suspect that the provided schematic is purposely wrong.
I also would like to know why a full bridge/forward converter is a bad idea and please what is the function of L201 ?
 
 

[oldway]

I dont know where S4 is. In the schematic U301 is referenced to U- and the resistor divider R302/R304 should set the output of the switching regulator to a fixed 61V approximately.
So I think that referencing it to U+ will not work with those values of voltage FB.
As grumpydoc said the power at the MOSFET will be huge (1100W). No way this can work, a pre regulator is required.  I suspect that the provided schematic is purposely wrong.
I also would like to know why a full bridge/forward converter is a bad idea and please what is the function of L201 ?

I agree, it seems there are serious mistakes in this schematic.
I have never seen an electrolytic capacitor directly connected to the secondary of a switching power supply as C106 in H2 of auxiliary power supply.
It obviously lacks as a connection point between the rectified + (+ of the capacitor C007A / B) and Q203B and T101 connection.

[Theobald]

Same for H1, C107 is wrong and the H-bridge supply (VCC) should be connected to the D105 cathode.
Funny.

Theo
Edit: In fact they're twins chemicals: C106 & C106' and C107 & C107'. In the real world they're not across the windings. I hope.

[Andy Watson]

I dont know where S4 is.

S4 is to the left of U301. It shows where the "GND" for U301 is connected, but the other end of S4 only appears  to connect through to "HY022 on PCB", which I can't see anywhere!
I was suggesting that "U-" was a typo and it should say "U+" under U301.

In the schematic U301 is referenced to U- and the resistor divider R302/R304 should set the output of the switching regulator to a fixed 61V approximately.

I make it 3.9V.

[oldway]

GROUND (GND) of U301 must be referenced to U+ and not to U-

So SMPS would act as a preregulator maintaining a fixed voltage drop on Q401 of 3.91V

[albert22]

So that was S4  . Sorry. Certainly connecting u301 gnd to U+ will work as a pre regulator but I dont understand how do you get 3.9V with the values for R302 and R304 shown in the schematic.  Another error in the schematic or I am not seeing something obvious (again)?

Please someone comment on L201. never saw that on a forward converter. I guess that it should be in series between C209 and C205 like L202 but on the + side.

[Andy Watson]

how do you get 3.9V with the values for R302 and R304 shown in the schematic.

U301 produces Vref of 5V on pin 14 (according to datasheet). This is the source of potential divider R309 and R305 (100k, 3k3) producing 160mV. This drives the negative input of error amplifier, pin 2, which compares it against the feedback voltage on pin 1. 159mV x 49k / 2k = 3.9V.

[richfiles]

Desoldered all the large semiconductors and blown/burned parts, cleaned up the board, and took some more photos. That fan header... shorted with a solder bridge, and the fan is soldered directly to the PC board elsewhere. Q403 was unpopulated. R204 has 4 parallel shunt wires, while the schematic says 3. Many of the transformer's labels show different part numbers than what the schematic shows as well. Schematic also Labels U+ and U- as 30v. This model is supped to do 50 volts. I'm wondering if they took a general schematic and altered it for different models... and maybe missed some details.   

I also gave that thing a bath! Isopropyl alcohol and a brush, top and bottom, then let it sit on the oven rack for a bit at the warm setting, to bake out anything that remained.
(No wife... I'm allowed to do that with the oven!) 

[EPTech]

Hi There,

This is a very interesting design. The U- is a floating reference. Notice that the TL494 is also referenced to U-. The TL494 will actually keep the voltage before Q401 slightly above U+ to maintain a current flow that results in the set voltage voltage across the load. This supply actually works in CC mode. The supply actually requires a minimal load, in this case R203.
So the SMPS part regulates the voltage so that Q401 can maintain and regulate a current trough the load with minimal power dissipation.

If you plan on repairing this supply, I would recommend you put some 15V transient suppressors or zener diodes across the primary side fets to protect their gates. Because the pulse transformer probably killed them. The 15 ohm resistors are not sufficient for dampening that.

Good luck.

[albert22]

how do you get 3.9V with the values for R302 and R304 shown in the schematic.

U301 produces Vref of 5V on pin 14 (according to datasheet). This is the source of potential divider R309 and R305 (100k, 3k3) producing 160mV. This drives the negative input of error amplifier, pin 2, which compares it against the feedback voltage on pin 1. 159mV x 49k / 2k = 3.9V.

  .  Fool mistake. And I cannot blame, coffee, beer or wine for it. I took for granted 2.5v on p1 because that was used in a similar circuit. Shame.
Thanks

[edpalmer42]

Which components actually blew up - as in let the smoke out?  There are probably more than failed silently, but you can get useful info by tracing the smoke.

Ed

[Seekonk]

Just do it!   It is a simple H bridge transformer coupled.  Replace all the FET and gate resistors.  Before replacements installed power up to see if other supply is still working and that driver transistors were not damaged.

[richfiles]

 :(R206A, R206B, R208A, R208B, and Q202A, Q202B, Q203A, Q203B all blew up. Violently. Only one FET managed to not physically detonate it's die. The others all had an out of body experience.  

R207A, R207B, R209A, R209B all test okay, at 27 ohms, but have some significant charing, that goes into their housings. I'm replacing them regardless of them passing.

R002, R003, R004, and R202 Passed and cleaned up just fine (R202 was clean, the others just had a light covering of soot. R002 will need a new piece of heatshrink around it).

C206A, C206B, C207A, C207B measure 2.1 nf consistently. They all appear to pass, but were pretty toasted in soot. I'll likely replace them to play it safe.

C003 and C004 are the caps across the mains. C001 and C002 are between mains and earth. I can replace these with your recommendations. If any other caps are sitting in critical places, and ought to be upgraded, like anything across the primary and secondary, I'm all ears.

D1-4 (actually 8 diodes, as they are in parallel pairs on my unit... something the schematic does not show) all tested okay.

C006A, C006B, C007A, C007B measure 765-767 µf consistently, well within the tolerance of these caps, but the two closest to the heatsink had ejected material from the resistors and MOSFETs melt the vinyl label. The two furthest from the heatsink had soot that could be wiped off easily. These big caps are all Nippon Chemicon caps. You can see on the PC board where one of the resistors that blew ejected material at the two closest caps. I usually order from Digikey, since I live in the same state as their location. Even cheap shipping shows up in a day or two. It'd be about $16 to replace these with equivalent Panasonic or $28 to go with Nichicon. Both options are uprated (2000 and 5000 hours @ 105°C, respectively). Honestly, I'll probably just go with the Panasonic caps. I've never had trouble with them, and it's already double the rating of the old ones. I figure I might as well hit all four caps.

C203A and C203B loved very clean. I left them on the PC board and tested them in parallel. they were 4.34 µf in circuit. since they are 2.2 µf caps, that's about dead on, well within tolerance.

I tested Q201, which is a half bridge diode pack. It appears to pass.

I did not do any significant testing of Q401 or SCR001. Q401 was a 5N2008, and SCR001 was an S6055M. S6055MTP is the replacement for S6055M (which is obsolete). Newark and Digikey both have no stock and a 250 part minimum. Mouser has no stock and a 10 week lead time, but will sell singles for $4.98... I have no idea how to test an SCR. I'll have to look that up, and see if I have what's necessary. Might have to order from Mouser if I need it. The 5N2008 is a different story. The 75NF75 listed on the thematic is a much smaller package component than the one in my unit. There isn't much showing up for 5N2008 either. Looks like a China special... Alibaba has s some, and Ebay seems to have ONE seller, who thinks it's an SCR.

I hope those last two are okay, but I'm not certain how to properly test either part. They aren't shorted, at least.

[Richard Head]

This circuit is an utter joke.
I'll list the errors I can see.

1. There is a full bridge switching arrangement on the primary side of the HF transformer but only a single ended forward converter output topology. This means that the transformer is being used uni-directionally but with all the complexity of a double ended topology. Why not just use a forward converter? NOT TRUE. Actually a current doubler topology. That's why there's L201.
2. Auxiliary PSU output H1 connects directly to C107 with no rectifier at all! Can't work.
3. The voltage feedback resistors (R302/R304) of the pre-regulator produce a voltage at the error amp input of 160mV as reported earlier! Why so low? (I believe the pre-regulator is meant to referenced to U-)
4. Purpose of L201 is unknown to me. IGNORE
5. How does U301 get its Vcc? Missing connection.
6. What's the purpose of D? Is it a panel meter? If so it'll indicate the duty cycle of the pre-regulator. Why?

I bet there are plenty of other serious issues that I haven't noticed also.

Feel free to add to the list.

[richfiles]

(D) took me forever to figure out too, but I think it's the fan. The fan is powered by it's own transformer winding, and has a diode, cap, resistor and inductor isolated from the entire rest of the circuit. It has absolutely nothing to do with the header marked "fan"... That header is shorted on the bottom of the board... 

I don't understand the fine workings of switching supplies well enough to follow this circuit. I had no reservations that this wasn't gonna be some crap Chinese thing. The two questions are obviously first, can it be fixed. I think that's not going to be hard to do. The other question is can it be fixed? I mean, can dumb stuff be corrected, so this will be reliable in the long term? Who knows? I got it for free, so if I can fix it cheap, that's awesome. It's SO OVERKILL though for anything I actually have plans to do.

If anything, I seem to have given the forum a nice challenge to rip a new one dig into! 

[cs.dk]

I suspect a lot of designs would run off a 60V current limited DC supply, so if you have a 2-channel 30V supply you could use that to power the thing. Much safer than a "dim bulb" tester, if it works

Sorry for offtopic;
But why on earth is it so difficult to find a "LAB PSU" that outputs 230VAC with current limiter and other features we know from all our DC PSU's?

[TurboTom]

It's not only the schematic that's questionable, the PCB layout is the same nightmare. Just look at the creepage distance of the negative rectified primary voltage to one pad of the secondary shunt array. Is it substantially more than a millimeter? On the control supply plug-in board, it's even worse. Adjacent terminals of the transformer are life and isolated side without any additional insulation, even with flux residues in between. An arc-over is only a quastion of time. If I'd been given that thing, I would have dumped it immediately. Maybe kept the enclosure with the displays for some project of my own but I definitely wouldn't use it in its original configuration as a lab power supply!

Btw, the power ooutput rectification is symmetrical and not single-ended as Richard mentioned. It's drawn a little awkward in the schematic but actually, a "virtual center tap" of the power transformer secondary is produced by L201 and L202 each of which carries half the DC current but basically blocks the AC from the power transformer's secondary. Quite an unusual design but probably nothing wrong with it.

Cheers,
Thomas

[Richard Head]

Turbo Tom
You are quite correct about the output inductor. It's the so called current doubler configuration. I didn't recognize it the way it was drawn. It's actually very effective for high current outputs.
I'll have to update my comments.

[Andy Watson]

This circuit is an utter joke.

Too right It's hard to imagine how anything useful could be produced from such a circuit - I wonder if it's been deliberately obfuscated?

2. Auxiliary PSU output H1 connects directly to C107 with no rectifier at all! Can't work.

There is something very wrong with this part of the circuit. Perhaps there should be a note that says "arrange these components to form a rectifier and filter."  If you do that you will also answer this:

5. How does U301 get its Vcc? Missing connection.

[CJay]

With the dire warning that if it's got auto input voltage sensing then using it on 220V with a series light bulb could just trigger the rapid, violent release of more smoke.

I've not read much about this occurring, have you got any instances or references where this has happened and how a good SMPS failed from it? I can only seem to find it mentioned on the repairfaq.org site. Would the same occur using a variac?

Personal experience from a couple of years of my life when I worked in a PC repair company, some power supplies aren't 'wide input range' but have a switch to alter the smoothing capacitor configuration (turns it into a doubler configuration and yes, they used to be expensive enough to be worth repairing).

That switch can either be a physical panel mounted one or an electronic one. The electronic ones can get latched into 120V operation and winding the input voltage up higher can have dire consequences for the PSU.

Which is quite frustrating when you've spent an hour drilling out rivets, cleaning up soot and replacing a couple of dozen components.

Some PSUs were (may still be) too sensitive to dips and spikes on the mains supply and latched to 120V in normal operation, we used to call them one shot brownout detectors.

A supply that has a single smoothing capacitor and is rated for 100-264V input is a wide range supply, a switched supply will have series connected smoothing capacitors and the 'centre' connection between them is where the input voltage switch is connected.

[richfiles]

Confirming, this supply has a manual 120/240 switch.

Hearing of the issues in the design instills me with such "Damn it, China!".

[MrSlack]

It's the Chinese design ethic. Steal someone else's well thought out design, go on a parts minimalism trip removing every bit of protection and reliability so it works long enough to sell it, slap times new Roman on it and whip an army of cheap labour to knock it out.

Like Sinclair but with less black and red and a different typeface.

[CJay]

Confirming, this supply has a manual 120/240 switch.

Hearing of the issues in the design instills me with such "Damn it, China!".

A proper switch, excellent, one thing less to worry about.

Looks to me that you should be able to test the various boards in there one at a time and build it up as you verify each one.

I think you can ignore Q401 for the moment, leave it out of circuit, it's being used as a linear pass regulator.

HY023 is an independent power supply, used for soft switching the main supply, powering the PWM controller, linear regulator drive and logic/display so that should power up if you can feed it rectified and smoothed line voltage or a suitable alternative.

HY012 will produce dual, complementary PWM modulated signals on V01 and V02 if you give it, I suspect, 12V or thereabouts on VCC, you should be able to vary the pulsewidth by varying the voltage on V1 (VI?).

Make sure the PWM drive is making it through the drive transformer to the gates of Q202, 203 A and B

Definitely add some 15V zeners to the gates of the primary MOSFETS to ground as someone else suggested, they won't save anything if it goes bang again but they'll help stop it going bang.

Once you get to this point you should be able to power up the main supply and get a non variable output on the drain of Q401. I'd guess somewhere around 60V.

You'll need to test HY013 and HY014 as one unit but again, you should be able to do that without the rest of the supply running. I'm guessing they're not going to be faulty but you don't need much more than 12V to do that.

Front panel meters are powered by HY023, they're real simple.

[richfiles]

Thanks CJay for that finely detailed troubleshooting information!

I won't be able to get to it immediately. I had the other day off, so I was unusually active with this, and I also just got an order for some cables that I build (the cable provides power and commutation for a brushless motor in a surgical tool. The cable is built to be able to withstand being autoclaved repeatedly to sterilize it). I'll have to do those first. Hopefully, by the time I finish those, my dual tracking power supply that I bought will have shown up, and I can use that for the testing...

Heck... maybe i should open THAT ONE up and have a look to see if it needs any safety updates. It's a used unit, from a liquidation sale. All I could afford at the moment. Just got SO TIRED of current limiting my Tektronix 500 series supplies. 400 mA is NOT reasonable for a bench supply to max out at. Fine for breadboarding a few logic chips or an op-amp on a breadboard, but not anything more!

I think I can follow most of what you said. I tend to work visually, hence why I do stuff like annotate pictures of the PC board and stuff. You broke this down into a nice bullet point list of tasks, and that's really awesome! 

My boss had a Sorensen DCS 300-3.5 power supply on hand that he uses in place of this unit. He has another Sorensen supply that's 3 or 4 times thicker. I don't recall what that model was It's a Sorensen Elgar SGA400/12C-0AAA (Input 187-242V 3 phase, Output 0-400 Volts DC @ 0-12 Amps). It's a beast of a machine, probably similar to the size of four of these units. Both are rack mount units. As far as I know, he's not even using the big one yet (He's not, he wants to sell it). Needless to say, with those beastly power supplies, I don't think he has a need for this sad little Mastech unit, especially in a production setting, if it's gonna just up and pop again someday. I kinda wanna fix it, and I kinda don't even have a clue what to do with it if I succeed! 

Can always do it for the challenge. 

What are thoughts on the 4 big caps? Obviously, they were in circuit at the time that things blew, and the vinyl labels suffered some cosmetic damage... Given that all 4 of them measure within 2 µf of each other and are within the +/- 20% tolerance for their value, can I just leave them be for now? Replacing them will cost $16-28, depending on the grade I buy (and not downgrading to cheap craps... I mean caps). I do not have an ESR meter, but I should be able to do it with a function gen and scope, right? If they are fine, I'd rather leave them be to keep the repair cost down. If they are borderline though, I would replace them. It just seemed interesting to me that they'd all test so consistently.

Another question, regarding the creepage distances of some parts. Obviously, the design is... less than stellar. My question, is will a thorough isopropyl alcohol cleaning, a bake to dry out, and a conformal coating help with this, or is the issue of an arc over something that would be a potential hazard regardless? I only occasionally work with 120 volt stuff, and have never serviced a switching supply before. About all I've done have been linear supplies and voltage doublers (to run nixie tubes). I've always conformal coated my high voltage stuff... But is that actually good to do?