Xantrex XKW 33-33 Power Supply - Successful Repair!

[edpalmer42]

I just received this power supply.  I think that it suffered an internal failure, then a heavy-handed, feeble, unsuccessful repair attempt, and finally the corpse was desecrated by having parts removed.  I'm really glad I got it cheap!

It's future is in serious doubt.  There are no schematics available.  So, since I occasionally like to tackle impossible jobs, I'm begging for some help from some kind soul who has, ideally, the same or a similar model.  It appears that Xantrex sold their power supply line to Sorensen so that might be another source for similar models.  The Sorensen DCS33-33 looks like it's the same as mine.

I need some part identification info for the following missing bits and pieces.

1.  Make & model of the blower that should be in the lower left corner.
2.  Part numbers for the four Mosfets that should be on the black heat sink.
3.  Part number for U3 - a socketed 8 pin DIP.
4.  Wiring info or photo of the 110/220 jumper that plugs into the 4 pin Molex connector near the top left.
5.  Value of R93 - large black resistor just to the left of the very cool current shunt assembly near the center of the unit.
6.  Value of R98 - small black(?) resistor just above the small transformer near the center of the unit.
7.  Part number & rating for the circuit breaker that's supposed to be on the front panel.

Yeah, I know.  It's 'Beyond Economic Repair'.  But I'm still going to play with it.  It has the advantage (from a repair point of view) that it's all through-hole.  I can *see* the parts!!    No stupid grains of sand!  And no processors, just nice little ICs that have datasheets and everything!  Should be easy - what could go wrong?

Thanks,
Ed

[fzabkar]

Have you looked at these?

https://elektrotanya.com/showresult?what=sorensen&kategoria=&kat2=all
https://elektrotanya.com/sorensen_dcs-series_3kw_power_supply_old_sch.7z/download.html
https://elektrotanya.com/sorensen_dcs-series_3kw_psu_rev-l_1999_full_sm.pdf/download.html

[edpalmer42]

Hi, fzabkar.

I do know about electrotanya, but I didn't check there.  The models you link to are the 3KW versions which are a lot different from mine which is a 1KW unit.  The manuals for the 1KW units have a circuit description but aren't distributed with schematics.  Apparently you have to phone Sorensen and give them your serial number to get access to the schematics.  So close.....

But what you've given me has resulted in a few more clues and ideas which is always welcome.

Thanks,
Ed

[coppercone2]

i destroyed a partially working one of these 30v30a by shorting some shit out probing it in a hurry, blew a fuse and made a horrible noise, I have no trouble shooted it since. It's a big job you probobly need to solder wires to it and stuff

It's a RE job, you need to follow the control signal to the transistors if its still functional. otherwise get rails working first

for unknown power supplies I start at potentiometer and look into control conditioning and scaling circuitry and it leads to the power devices typically. Unless its some weird shit like potentiometer being read digitally (dear lord).

I think I figured out (at the time) the limiting circuit, I thought something was getting stuck near a rail, but when I probed the LDO or whatever I think I put unregulated voltage into the board, multiple parts replacement likely
beyond economical repair    (jk, no such thing especially when you are employed)

[MagicSmoker]

I have a Glassman 80-13 which looks very similar (but not identical) inside to yours (the PCB is labelled Xantrex, in fact) with fried output rectifiers that I started to fix but then got sidetracked and it's sat on a shelf ever since. It looks like the bridge MOSFETs are missing and they are part number IRFP450 in mine, though a more modern part with similar or better Rds[on], voltage and, especially, gate charge specifications in a TO-247 package should drop in just fine.

Getting at the output rectifiers is a real pain in mine as they are buried under a PCB with lots of heavily soldered connections; yours doesn't seem to have the same PCB (likely for snubbing/noise-reduction) so should be easier to access.

Oh, and make sure the blower still works - mine died which likely contributed to, if not outright caused, the rectifiers to fail, and IIRC, I didn't find a mechanically-compatible replacement at Mouser, Digikey, etc. That was one of the reasons I ended up shelving the repair job (the other being I have lots of power supplies).


EDIT - added strikeout for not reading all of the OP...

[MagicSmoker]

1.  Make & model of the blower that should be in the lower left corner.
2.  Part numbers for the four Mosfets that should be on the black heat sink.
3.  Part number for U3 - a socketed 8 pin DIP.
4.  Wiring info or photo of the 110/220 jumper that plugs into the 4 pin Molex connector near the top left.
5.  Value of R93 - large black resistor just to the left of the very cool current shunt assembly near the center of the unit.
6.  Value of R98 - small black(?) resistor just above the small transformer near the center of the unit.
7.  Part number & rating for the circuit breaker that's supposed to be on the front panel.

Sorry, skipped right over your wish list. I think mine mainly differs from yours in that it has the IEEE 488 bus option installed (as well as output voltage/current rating, of course). I'll take a photo of mine later when it is socially acceptable to fire up an air compressor to blow off all the dust, in the meantime:

1. Unknown.
2. IRFP450
3. LM358AN
4. Jumper pins 1 & 4 for 230V; or pins 1 & 3 and 2 & 4 for 115V (pin 1 closest to front panel).
5. Mine is 2R/5W, vitreous enamel wirewound
6. 10R/0.5W, carbon film
7. 20A/250VAC standard single pole "push-to-reset" thermal breaker; part number illegible on mine.

[MagicSmoker]

...
5. Mine is 2R/5W, vitreous enamel wirewound
...

This resistor is part of the output current sensing so may very well be different on your model. Otherwise, this post and the next will have pictures of mine in clockwise fashion starting with the top right. Note that the IEEE 488 interface board occupies the topmost portion of the top right and left photos, but since that isn't installed on your board I didn't take any care to capture it all.

[MagicSmoker]

And the bottom left and top left:

[edpalmer42]

Magic, that you very very much!  This is great!  I knew the eevblog hive mind wouldn't let me down.

1.  Make & model of the blower that should be in the lower left corner.
2.  Part numbers for the four Mosfets that should be on the black heat sink.
3.  Part number for U3 - a socketed 8 pin DIP.
4.  Wiring info or photo of the 110/220 jumper that plugs into the 4 pin Molex connector near the top left.
5.  Value of R93 - large black resistor just to the left of the very cool current shunt assembly near the center of the unit.
6.  Value of R98 - small black(?) resistor just above the small transformer near the center of the unit.
7.  Part number & rating for the circuit breaker that's supposed to be on the front panel.

Sorry, skipped right over your wish list. I think mine mainly differs from yours in that it has the IEEE 488 bus option installed (as well as output voltage/current rating, of course). I'll take a photo of mine later when it is socially acceptable to fire up an air compressor to blow off all the dust, in the meantime:

1. Unknown.
2. IRFP450
3. LM358AN
4. Jumper pins 1 & 4 for 230V; or pins 1 & 3 and 2 & 4 for 115V (pin 1 closest to front panel).
5. Mine is 2R/5W, vitreous enamel wirewound
6. 10R/0.5W, carbon film
7. 20A/250VAC standard single pole "push-to-reset" thermal breaker; part number illegible on mine.

Your board looks identical to mine although mine has some differences in the high power components installed.  I suspected that would be the case.  Use the same basic circuitry and change the switching transformer and output components as necessary.

My junkbox should have most of the missing pieces except maybe for the IRFP450.  I did see an IRFP460, but coming up with 4 seems unlikely.  I know where there's a much bigger junkbox so this could turn out to be a really cheap repair.

In your next message, you say that R93 is part of the current sensing.  Are you sure about that?  In the circuit description for the Sorensen DCS33-33, another clone of this unit, it says that Q17 and R93 make up the overvoltage protection.  But Q17 is only rated for 18 - 10 amps depending on the junction temperature.  In mine, it looks like R93 didn't burn, it exploded and Q17 is now shorted.  I confirmed that R93 and Q17 are connected in series, and are across the output terminals.  I think that whatever fried R98 (the small blown resistor) killed the regulation, drove the output to maximum and when the protection triggered, R93 exploded.  The circuit description describes it's purpose as discharging the output after the main drive has shut down.

That missing circuit breaker baffles me.  What's the point of putting a 20A breaker in series with a 20A slow blow main fuse?  I'm replacing it with a 15A breaker, but the whole idea of a breaker in series with a fuse seems silly.

Okay, I've got some work to do.  I'll have to trace out the schematic around R98 and figure out what killed it.  Based on its location, it looks like it's in the driver/regulation area of the circuit.  I've seen a few similar units on ebay that also had that resistor burned so it looks like it's a common fault.  Then I'll try powering up the control side and see what happens there.

I just checked the main output rectifiers and they pass the multimeter test so that's good enough for now.  It'll take some more work before I can do a smoke test on the high power side!  I'll also have to source a 10A HRC fuse to replace the one the scavengers took.  Grrrr.  A pox on their house!

Ed

[duak]

Ed, I've got one of these too.  The fan looks like it's hard to get to - mine has a cardboard shroud to direct air around the hot stuff but It's more than I want to deal with right now.  The fan is an axial flow unit about 80 x 80 x 35 mm.  I think it's a 24 VDC unit and the voltage is varied to control the speed.  I measure a current of 150 mA at 18 V.

I'm in Vancouver - are you anywhere nearby?

[edpalmer42]

Ed, I've got one of these too.  The fan looks like it's hard to get to - mine has a cardboard shroud to direct air around the hot stuff but It's more than I want to deal with right now.  The fan is an axial flow unit about 80 x 80 x 35 mm.  I think it's a 24 VDC unit and the voltage is varied to control the speed.  I measure a current of 150 mA at 18 V.

I'm in Vancouver - are you anywhere nearby?

Duak, it looks like yours is a different style - maybe newer.  Mine has a dark front panel and a I found a few 1997 - 1998 date codes.  The power for my fan is 110AC.  I've traced one lead directly to the neutral side of the line and the other to the center point of the input capacitor bank.  I've seen this configuration before where they use a 110VAC fan in equipment that has strapping for 110/220.

Rather than an axial fan, mine appears to be a tangential/centrifugal blower.  I think I've identified it as a Rotron Biscuit Blower model BT2A1.  There's a picture on ebay of a Sorensen DCS33-33 with the top removed (see attached).  It looks a bit different than mine, but it's basically the same unit.  I can't quite read what's on the blower in the picture, but when you compare it to a Biscuit Blower, it's pretty obvious.  I've got the dimensions for the Biscuit blower, but I still have to do a few calculations and measurements to confirm that it's the right unit.

I don't know if I'll try and buy a Biscuit blower or maybe reengineer a cooling solution based on the type of fan used in a 1U rackmount server.  Maybe include a thermal sensor for fan speed control.  Those 1U servers can be really noisy!

I'm in Regina so no, I don't think I'll be popping over for a visit, but thanks for responding.     (Regina and Vancouver are about 1330 km (826 mi.) apart.)

Ed

[MagicSmoker]

...
My junkbox should have most of the missing pieces except maybe for the IRFP450.  I did see an IRFP460, but coming up with 4 seems unlikely.  I know where there's a much bigger junkbox so this could turn out to be a really cheap repair.

Any MOSFET rated for 500V or higher with less than 0.4R Rds[on] and 150nC max gate charge (Qg) in a TO-247 package will work, but make sure they are all the same so the bridge remains balanced. For example, the first three that come up in this presorted list - IPW60R180P7XKSA1, STW28N65M2 or R6020KNZ1 - will all work fine: https://www.mouser.com/Semiconductors/Discrete-Semiconductors/Transistors/MOSFET/_/N-ax1sfZ1yzvvqx?P=1yzxdnwZ1z0xv21Z1yzt659Z1yzt658Z1y8ps0hZ1y9o2dsZ1yuk7vzZ1z0z63xZ1z0y3zrZ1yw76gdZ1yw76cyZ1yw76gpZ1yw76ej&Rl=ax1sfZgjdhozZ1yw78diZ1yw76fbSGTax1sfZgjdhp2Z1yw76fyZ1yw76fuSGTax1sfZgjdhp1Z1yw8purZ1yw8pjxSGT&Ns=Pricing%7c0

In your next message, you say that R93 is part of the current sensing.  Are you sure about that?

Nope, I just eyeballed the board and made an assumption about its purpose. I don't have the schematic and haven't gotten around to reverse engineering it.

In the circuit description for the Sorensen DCS33-33, another clone of this unit, it says that Q17 and R93 make up the overvoltage protection.  But Q17 is only rated for 18 - 10 amps depending on the junction temperature.  In mine, it looks like R93 didn't burn, it exploded and Q17 is now shorted.

Hmm, if that is the case then R93 is still unlikely to be the same value in your model as in mine.

What's the point of putting a 20A breaker in series with a 20A slow blow main fuse?  I'm replacing it with a 15A breaker, but the whole idea of a breaker in series with a fuse seems silly.

Thermal (bimetallic strip) breakers don't open too quickly regardless of overload, even compared to a slo-blo fuse, so likely it's a belt-and-suspenders type of thing. I agree it's a bit much, though.

[coppercone2]

jesus why does everyone hate protection systems, its so you can repair the fucking board without scraping 2 oz of char off it

its a sorsensen supply that retailed for 1500$ or so. I have seen them used to run physics experiments and lasers that cost 10-100x the price. You put a few of them and some vacuum pump in a little rack that is wheeled around a REAL laboratory.

its called quality  . its not supposed to be sold at dollar tree

p.s. they still suck because xantrax likes MTA connectors which are shit (on their little box supplies you often need to poke the ribbon cable with a screw driver to reconnect the front panel.)

i think they are sold cheap because their noise angers scientists

if the new model was made by cheapskates in this thread
https://i.chzbgr.com/full/5672551936/h942269E2/

[volvo_nut_v70]

Xantrex OEM'd supplies to Sorenson, Elgar, Glassman and Chroma. Sorenson bought Xantrex when they sold off their power supply business, they are a local "hometown" business.

[edpalmer42]

Status Update



Yes, it's alive but, like Frankenstein's monster, it has issues.

By raiding various cemetaries (while avoiding local peasants with axes and pitchforks) I was able to obtain enough parts to complete the corpse.

Initial attempts at resusitation were promising but ultimately unsuccessful.  The corpse was connected by an umbilical cable to an external power supply that replaced an internal source with an external, current-limited source.  This was done to protect the newly replaced R98 (10R0,0.5W) and help diagnose why it burned.  Both the corpse and the external supply were turned on together via a switched power bar.  With this arrangement, I was able to run the output voltage up to its maximum of 33V.  AC current drain is ~200 ma.  No load testing was performed.

I then removed the umbilical and reconnected the internal supply.  Now, the OVP (Over Voltage Protection) turns on at power-on and there is no output.  Remember that originally, R93 and Q17 were blown.  These are part of the OVP circuit so that might be part of the original fault.

There are two items that I will be investigating next.  I checked the internal power supply voltages and all are good except for the 6V2 'reference' which is sitting at 6V08.  Although the difference is small, it could have shifted some trigger point just enough to cause issues.  Second, without doing too much research, I replaced the missing U3 (LM358) with an RC4558.  A quick review of the specs suggests this should be acceptable, but maybe not.

More technical details will come in the next few days.  I've discovered some fascinating details regarding R98 and its surrounding circuitry that hints at some very smart design work by Xantrex which doesn't seem to have translated to Sorensen when they took over the product.


I may have to wait for the next thunderstorm to complete the project.   

Ed

[edpalmer42]

Success!  Now it really is alive!

Originally, U3 was missing.  It was supposed to be an LM358A.  I installed an RC4558 and it triggered the OVP on powerup.  I dug deeper into my boneyard and came up with an LM358 which solved the problem.

Not a full load test (yet), but still impressive.  The 'charging panel' is actually a 1 ohm 300W potentiometer (!).  The power supply is displaying 20V0 @ 18A9 or 378 watts.  Yes, the pot is overloaded, but it didn't mind.  The power supply efficiency is ~65% and the power factor is, coincidentally ~0.65.

Ed

[edpalmer42]

The Strange Failure of R98

While repairing my Xantrex XKW 33-33 power supply, I came across a puzzling failure.  I'm going to go into some detail on my thoughts and results because I've seen pictures of two of the Sorensen clones of this model (DCS33-33E) that suffered more serious damage than mine did.  Hopefully, this info will be useful to owners of either the Xantrex XKW or Sorensen DCS power supplies

When I received my unit, R98 was burned so badly that it was open circuit and there was no way to determine its value.  Luckily, MagicSmoker had a Glassman 80-13 power supply which is another clone from the same family as mine so I was able to get the value.  Since I can't find a schematic, I traced out the circuitry around R98 (see attached ).  The power supply shown as +AUX is the unregulated output of a full-wave bridge from a 16V transformer that is then filtered with a capacitor.  It forms the source for the +5V and +12V supplies used elsewhere.  C67 and C68 go to other parts of the circuitry.  Q7 and Q10 are two of the six fets that drive the main switching transistors through T2.

Why put an 18V zener on a 19V, unregulated supply and why use a 5W device?  Since all the other components shown test good, how could you blow R98?  Looking at the board, R98, CR31, R104, and R106 are all spaced away from the board.  I don't remember seeing 0.25W or 0.5W resistors mounted like that.  Usually that's done with higher power devices to prevent the heat from degrading the board.

The only thing I can conclude is that the unit experienced a significant overvoltage condition on the incoming AC that drove +AUX higher for a significant time.  It could have been a power line fault.  It could also have been a user mistake where it was strapped for 110V and then plugged into 220V.  If that happened, I'm surprised at how little damage there was, but there's no way to tell how much of that voltage made it through the transformer.  Nothing else shows signs of heat or voltage stress.

After repairs were complete I did some temperature testing while the unit was outputing ~90W with some eye-opening results.

R98:  48C
CR31:  35C
R104 & R106:  80C   

Now I know why R104 and R106 are spaced away from the board.  But why not just make them larger so they can better handle the power?  I wonder if R98, R104, and R106 are sacrificial components that are intended to blow under fault conditions while saving the board and the other components.  Could R98 be a fusible resistor?  When I look around the unit, R37 (1K, 0.5W) and R103A (15R, 0.5W) are also spaced away from the board.  R37 is a normal-looking resistor, but R103A has a reddish-brown ceramic body which makes me think 'fusible / flameproof resistor'.

In the pictures of blown Sorensen supplies, it looks like larger components were mounted tight to the board.  The boards appear to be badly damaged by the failure of R98 so it obviously wasn't a fusible resistor.

In summary, if you come across one of these power supplies with a blown R98, check the components in the partial schematic.  If they all test good, the fault was probably caused by external forces.  Replace R98 and see what happens!

Ed

[duak]

Good-O! for keeping another piece of electronics out of the dump for a few years.

I've seen 1/4 W or 1/2 W resistors in recent equipment that seemed different in shape and/or finish from what I've seen before.  I remember using some Philips resistors that had an unusual brown and granular coating instead of the usual tan and smooth coating.  These resistors were rated for 2 W although they were about the size of 1 Watters.  Regarding the parts in the Xantrex supply,  I wonder if this downsizing was applied to lower power resistors in their line or if they were rated for a higher voltage. 

[edpalmer42]

Here's the shot you've been waiting for.  Full load test - 33V at 33A!  The two large resistors on the left are each 10 ohms @ 500W.  Power is fed into the centers and all the ends are shorted together giving 1R25 @ 1000W.  The 300W pot on the right plus a couple more resistors behind it make up the rest.

Yes, everything's sitting on my kitchen table.  The kitchen is the only place where I've got circuits that only have one outlet!

Power drain from the wall was interesting.  My Kill-a-Watt was complaining that the current draw was 15A3.  A current transformer fed into a DMM said it was 11A5.  Another Kill-a-Watt clone said 12A1.  I think my official Kill-a-Watt is defective for reasons other than this test.  In any case, it looks like I can run it at full load as long as there's nothing else on the circuit.  I live in Canada where our mains is 115V and the breaker is 15A.

Ed

[edpalmer42]

Good-O! for keeping another piece of electronics out of the dump for a few years.

I've seen 1/4 W or 1/2 W resistors in recent equipment that seemed different in shape and/or finish from what I've seen before.  I remember using some Philips resistors that had an unusual brown and granular coating instead of the usual tan and smooth coating.  These resistors were rated for 2 W although they were about the size of 1 Watters.  Regarding the parts in the Xantrex supply,  I wonder if this downsizing was applied to lower power resistors in their line or if they were rated for a higher voltage.

I'm not sure I understand your point.  Xantrex originated the design so, if anything, Sorensen upsized the resistors.  Voltage rating isn't usually an issue with typical resistors because they hit a power limit long before a voltage limit.

I suspect that Sorensen didn't fully understand the purpose of the 'undersized' resistors or why they were mounted above the board so they used standard engineering logic and increased the power rating and used standard mounting rules.

Ed

[duak]

Sorry Ed, I should have been clearer.  I mean to say that the component manufacturer may have offered downsized versions of their lower power parts (say 1/2 W) by using more refractory materials allowing them to run hotter.  It complicates finding replacements because what may appear as a 1/2W part might be 1 W or, as you say, may be a fusible part or just sized to fail open in case of a problem, ie. fail-safe.

I just helped out a friend with a CNC lathe.  The operator smelled something burning and opened the electrical cabinet.  A bit of smoke and a lot of heat of came out.  He thought he saw something glowing red so he shut it off pronto.  I came out and confirmed that a 250 W ceramic resistor was glowing red hot.  This is a braking resistor for a servo drive and the power transistor driving it had shorted out causing it to dissipate 2 KW.  The resistor was mounted vertically about 2" from the servo amp.  All of the nylon Molex style connector housings facing the resistor were charred and melted.  Funny thing, the servo still worked, didn't throw an alarm and didn't trip the internal circuit breaker.  The servo is from the early 80's and has very few protective devices.  I think I see an over temperature switch on the heat sink but nothing on the power resistor.  Things like this work until they don't.  Eventually the resistor would have failed and probably damage the assemblies underneath it.  Or something else would have gone west in the servo drive.

Xantrex manufacturing used to be about 3 km east of me.  I have a dual lab supply from them when they were called Anatek.  I have no idea if they do anything technical around here anymore.  When I looked at their website I see their canadian office is in the building that I used to work in.  This building is kind of futuristic in styling and is sited in a park like area.  It was sometimes used as a set to shoot various TV shows such as McGyver & Airwolf.  In addition to the entertainment from watching the filming, the production company would feed us from the catering trucks - a double win for us.

[edpalmer42]

Okay, now I've got it.  Yes, if they used resistors that looked smaller than their actual ratings it could be impossible to tell.  Too bad the resistor color code didn't include a wattage rating.

Ed

[volvo_nut_v70]

Found on the web, this might help in your work.....

[edpalmer42]

Thanks Volvo!  I looked high and low, but only found the standard User Manual.  Well done!

Of course, the damn thing only shows up after the repair is complete.        Well, at least I'll be able to confirm if my substitutions were correct.  Or maybe not.  A quick check shows that some of the parts are only identified with internal part numbers.   

For anyone interested, I extracted the schematic and stitched it together into a single sheet.  Kind of busy, but at least it's legible.

Ed

[volvo_nut_v70]

Thanks for stitching it together! I've also uploaded it to BAMA and KO4bb.

Congratulations in your hard work and postings!