Xantrex/Sorensen XHR 40-25 schematic?

[coppercone2]

lol where is waldo

watch that fixes it!

[ifonlyeverything]

Hello, you maybe have problem in two components in corner?

Looks a little toasty, I'll have to check it. Unfortunately haven't had much time to work on this due to demands of my day job. I also need to order a new set of output capacitors and desolder/individually check the input capacitors and FETs.

[coppercone2]

a burned out part is like a really strong guess that it might just work if replaced. and whatever that resistor is connected to also.

[NikiG]

 

[ifonlyeverything]

4 months later and I finally have a little free time to dick around with this. I replaced the capacitor that NikiG identified and fired it up. The relay clicked a bunch of times, the display flickered, the SMPS fired up for a second, and then it went back to flickering the display. I power cycled it a few times. No more relay clicks and the display is flickering. It's not powering up fully as I can't hear the fan run.

I can measure 170V DC across P2 so I know everything upstream is good. U21 (UC3842 SMPS IC) has an unstable VCC-GND voltage, fluctuating between 12V and 13V DC rapidly. I measured the voltage across C38 and I'm getting between 0.8V and 1.0V rather than the expected 18V. I measured the voltage across C35 and I'm seeing 7-8VAC at 60 hz which makes no sense.

This C35 output, I think it's supposed to be 18V, feeds the UC3854 PFC IC and the UC3524 full bridge IC. The UC3842 VREF output feeds the U64 ripple counter, which feeds the UC3524 full bridge IC. I measured this VREF and it's 0.8V to 1V, so nowhere near its intended 5V.

I'm not an expert with SMPS here but I assume something in this UC3842 section is having an issue here. Any suggestions on what I should measure next? I have a scope but I'm a little paranoid of using it since I don't have differential probes.

[coppercone2]

how are you measuring the flickering voltage?

It bothers me. But its also a SMPSU IC. I don't know if those can make voltage flicker maybe if its switching something rapidly. I think that there is a problem here, but pulse loads can do weird things.

This usually means there is a bad capacitor. I assume its a linear regulator powering this IC?

If you see that much AC on a filter capacitor it might mean the capacitor is bad. Of course someone that works with that  chip might know for sure.

8VAC on top of a digital chip power supply rail seems like way too much


Fix that because it can make other things go really screwy. Like everything on that rail is suspect if its behaving like that. Or at least verify its normal some how 

[ifonlyeverything]

how are you measuring the flickering voltage?

It bothers me. But its also a SMPSU IC. I don't know if those can make voltage flicker maybe if its switching something rapidly. I think that there is a problem here, but pulse loads can do weird things.

This usually means there is a bad capacitor. I assume its a linear regulator powering this IC?

If you see that much AC on a filter capacitor it might mean the capacitor is bad. Of course someone that works with that  chip might know for sure.

8VAC on top of a digital chip power supply rail seems like way too much


Fix that because it can make other things go really screwy. Like everything on that rail is suspect if its behaving like that. Or at least verify its normal some how 

I measured the 12-13V DC between pin 7 (VCC) and pin 5 (GND) of the UC3842. There's no linear regulator. The P2 connector is seeing 170V DC, which is fed through R14 to the VCC pin 7 of the UC3842. The auxiliary winding presumably takes over and feeds rectified voltage (diode CR16 and capacitor C22) to the IC, once it successfully starts up.

This is all further complicated by U64 which is acting as the oscillator for the UC3842 (pin 7 of U64 to pin 4 of the UC3842)... as well as acting for the oscillator for the UC3854 PFC IC and the UC3524 full bridge IC. I barely understand how U64 works but I think it implements a startup delay in addition to acting as the oscillator... so I have no idea how there is synchronization between UC3842 pulsing on with 170V DC, the auxiliary winding taking over, and the UC3842 somehow outputting a VREF stable enough to power U64.... which then takes some time to finally generate the oscillator pulse that feeds UC3842 RT/CT inputs... which it presumably needs BEFORE the auxiliary winding can fire? 

Maybe I'll try probing the Q11 gate and the Q11 sense resistor, to see if the switching waveform look normal at any point, and to see if excess current (from a short somewhere) is causing the IC to repeatedly loop through overcurrent shutdown and start up.

[coppercone2]

But you said that is 6VAC on a digital rail that goes to UC3842.


it might have a wide range but VDD pins are usually DC voltages with not much AC component.

Even if its just a transformer secondary thats regulated it should be smoothish I think, not 30% ripple. if I read that right 30% ripple + 1VDC out of a VDC reference regulator means there is a problem. I would start with 30% ripple being a problem

[ifonlyeverything]

But you said that is 6VAC on a digital rail that goes to UC3842.


it might have a wide range but VDD pins are usually DC voltages with not much AC component.

Even if its just a transformer secondary thats regulated it should be smoothish I think, not 30% ripple. if I read that right 30% ripple + 1VDC out of a VDC reference regulator means there is a problem. I would start with 30% ripple being a problem

The low AC voltage I measured is across the smoothing capacitor C35 of the rectified auxiliary winding, which feeds the UC3842... as well as the PFC IC, full bridge IC, full bridge gate driver ICs, and several op amp power rails. Maybe I'll measure this with a scope as well. I initially measured it with my DMM, which refused to read anything on VDC but showed 7-8VAC at 60 hz on VAC mode. It has to be an artifact from things hiccuping and my DMM not being able to read fast enough. I have no idea why else I would be seeing 60 hz AC unless something is seriously fucked up.

[coppercone2]

yeah, the capacitor !!! electrolytic capacitor not smoothing. news at 11


very common for caps to die and not do their job

The only thing that can cause a functioning cap to do this, is if the load is way greater then what that value capacitor can do. Capacitor value in a smoothing circuit is proportional to load. More load = bigger cap to get same ripple

unloaded it should be dead smooth

or the diode

i won't say that this is the only possible explaination, I think there might be some other possibility, but this is a VERY COMMON occurrence if its working as I described

[ifonlyeverything]

Pulled every electrolytic capacitor on the main PCB and they all tested fine on ESR meter: C22, C28, C35, C37, C39, C41, C77, C90, C100, C108, C289. Pulled Q11 and tested it on my transistor tester, looked fine.

I think my next step is to pull every diode, resistor, and non-electrolytic in this UC3842 circuit block and test for dead shorts. If nothing is bad then I'll pull just replace UC3842 and hope that works, because I have no clue what else it could be other than the transformer being bad (unlikely) or something pulling down the +18P rail (a ton of shit.)

Or maybe I could just unplug it from mains and inject +18VDC from my bench power supply at CR17 and see what kind of current draw I'm getting.

[ifonlyeverything]

Pulled CR17 and F1, swapped R14 with a 10K resistor temporarily, and powered the downstream size of F1 with +30VDC with a 1A CC limit from my bench power supply. All voltages are looking nice and stable. Plugged in the secondary control board and the front panel, voltages were fluctuating and the front panel was visibly flickering. Unplugged front panel from secondary control board and voltages were still fluctuating. Looks like the problem is isolated to the secondary control board. 

[coppercone2]

ohhh I see its a totally different form factor then mine. i thought it was the single unit high pizza box

the schematics look very similar. but mine had 0 capability for isolation unless you want to cut traces. All I could do is unplug the GPIB card. I did not think of that trouble shooting technique


I forgot to mention, do you have crappy sockets in there for op amps? Because one of mine had a bad socket for a op amp. they are not machine pin, it had some really dodgy ass leaf socket

pic1 confirms you do have the shitty sockets.




Just thinking about it, flickering is sometimes caused by bad contacts. On those cable harness too. But I can't think of how it would cause the vref voltage to drop down to 20%

if you fix it you might wanna replace them all for 10$ before you put it back together, so you don't explode something later on.



I had the linear one in that form factor, the only thing I renember about it before I threw it out
1) it had something like 5 transistors working together on the output stage for the big power transistor, that is a bunch of 'passives' (do hardly any work with current). One of them broke. I replaced all of them and it started to work. I over tightened a chassis screw and caused a bolt to fry one of them again because a spacer was missing or not engineered in. it fried something again. Then traces started coming off during rework and i threw it out. if the power transistor breaks the ones around it break too
2) one of the wires on the IDC cable (insulation displacement cable) broke and came out of the harness. Clever method to weld the wire to the connector but I don't like it. you can't really poke it back in there, whole cable needs to be replaced or a solder fix has to be made with a external jumper wire. the crimp tool causes it to cold weld to the pin fork. once that breaks it done. its like wire wrap, not a crimp. the crimp tool is expensive

[ifonlyeverything]

Fed the secondary control board with my bench power supply, the 5V and 9V sections were fine. 18V and -18V had shorts, replaced the linear regulators U238 and U252 and the short went away.

Found out that main A3 board and the secondary A2 can have connection issues because the box headers aren't properly sized, meaning the pins will shift over 1 row... when this happens it will trip the auxiliary supply like before. Not sure if this caused the U238/U252 regulators to blow for the previous owner, but they did not blow for me.

I reassembled A2, A3, and the front panel and made sure A2-A3 connectors mated properly. Fired up the auxiliary power supply with +18VDC from my bench supply and it worked perfectly. All of the buttons on the front panel worked as expected, holding in V/I to set up a commanded V/I worked fine -- no drift in the numbers.

Re-soldered CR17, put F1 back in, swapped R14 for the 68K resistor. Reassembled everything. It fires up, front panel lights up, fans run, and there is an output voltage... but it's not working right.

I click in STANDBY and hold in V/I to set the V and I and the numbers will drift. They did not drift when I was powering it off +18VDC bench supply. I click out STANDBY to enable the output and I can hear a buzzing sound. Not sure if this is expected. Voltage refuses to go over 20V and it jumps around. Current limit is also acting strangely. It looks like the CC and CV limits are being obeyed but it just drifts around and won't stabilize. I'm using resistive loads.

Rear panel SW1 is in default configuration and I have J2 SNS- wired to the negative bus bar and SNS+ wired to the positive bus bar.

So... back to troubleshooting!  Will be sure to double check ALL of the sockets.

[coppercone2]

ok but if the regulators were shorted it might mean that there was a shorted load connected to them and that is why they maybe over heated and died after a while. like one of the op amps. or a cap shorted out and went open.

idk how to check the sockets though, replace. I had this bullshit where I pressed it in and it would work. then after i reassembly chassis it stop working. I open it up and push on chip it works again.

its like playing peek a boo with a troll

if you think you fixed anything by messing with the socket just replace the damn thing, its not reliable, i think you can get 30V randomly if the wrong socket goes up. meaning bye bye load (and a reason to fuse your prototypes even if your on a lab suply). i bought like 20 fuse holders and a whole set of fuses so I don't have to worry about malfunctioning opamps burning things

but thankfully when my op amp socket acted up, it dropped down to 3V max. idk how it decide on which set point to fall down to. maybe its built so it goes down to a low level if it fails, but I can't verify this... also it did once go to 17V when set to 15V I think, which caused a fan to burn out (otherwise tolerance circuit)



if your setpoint is acting up, that is either a reference drifting (it all based on reference), resistor divider or op amp malfunctioning.

or it might be nooise from a broken capacitor causing circuits not to settle (filter went haywire)


intermittant connector might cause things to dance around too. 14x+ magnification and a healthy dose of paranoia help find this problem


I think that really bad connector develop sensitivity to current load (resistance changes with load) and probobly temperature, which causes very minor motion due to expansion, makes it ACT up. *


*could it be a bad pin on a op amp making a very high impedance that ends up effecting the setpoint of a voltage circuit?


when your done fixing it, hook it up to a high current load (i.e. big ass resistor) and leave it on for a while to make sure it won't malfunction after 2 minutes and make danger



I have a feeling these power supplies probobly have distributed damage (lots of minor wear and tear) and that replacing stuff generally helps them. i think they basically get generally clapped out


when you screw in the pCB it gets flexed and stretched a little bit. might work when its out then stop working when you put it in. this is mechanical damage



the other thing you can do, recommended, is to reheat every solder joint on it, but it needs to be well fluxed. this means you need to do a big ass clean up after wards. I would do it for a 1kW supply though. but renember it needs to be well fluxed, and that means big cleanup. if you just do it dry to make it easy to clean it might not help things. like probobly a few hours of scrubbing to get it really nice, and you possibly loose some text and component indicators if your not careful. if you carefully apply flux, on both sides, and clean as you go, its not much worse then other penal labor . And heat can reveal problems, so you need to grid the board and test in between doing squares if you want the most benefit and highest success rate from this method to say.. maybe identify a bad bond wire or something from heat. where it goes from intermitant to broken. if you do it all at once you can get a fix, but you won't know where.

[ifonlyeverything]

You were right about something killing the regulators. I think the -15V regulator is blown (it's outputting roughly -18V instead of -15V) and the 15V regulator is running super hot because something is pulling it down to 8-9V. Auxiliary power supply output is still fine. Will dick around with this later but I'm guessing the OP07 block is screwing me over, since that's where the blown open tantalum capacitor was found.

[coppercone2]

thermal imager might come in handy here . seek thermal

or probing stuff (its fucking annoying though)

[ifonlyeverything]

thermal imager might come in handy here . seek thermal

or probing stuff (its fucking annoying though)

Thermal camera didn't reveal anything unexpected. Only thing heating up were the linear regulators and a couple small value resistors on the 6.2V zener and the linear regulator. The zener was regulating fine. I replaced the linear regulators and powered it off my bench supply, didn't pull above like 0.05 amps at idle (due to those small value resistors.) Connected everything back up same problem with U238 being pulled down and U252 instantly shorting out. One of the output capacitors (new Nippon Chemicons from LCSC) blew the fuck up too. 

[ifonlyeverything]

I'm not sure why it didn't pop up on the thermal camera before but C239 was hitting 50C so I swapped it with a new through-hole ceramic cap. I replaced the output capacitor that blew and replaced the linear regulators with brand name versions because I realized my stock were counterfeits I bought from eBay years ago.

I did some light load testing and I believe it is working. Horrible transformer buzzing is gone. Linear regulators U238 and U252 are cool to the touch and are outputting the correct voltages. U257 is running quite hot at 80C case temperature but it's not heatsinked as per design. It feeds the fan and I measured the fan pulling 100 ma, so I guess it's just engineered to run hot.

I need to mess around with some of the trimmers because the output voltage is half a volt more than the CV limit, but everything looks stable at a few hundred watts load. Just need to come up with a way to test it at its rated 40V and 25A.

In summary: C151 visibly blown; C239 shorting causing U238 and U252 linear regulators to short or blow; U257 linear regulator replaced preventatively; output capacitors C137, C140, C141, C143, and C162 replaced preventatively; new C137 capacitor blew so it was replaced again.

[coppercone2]

nice I had to take mine apart to wash it the other day because coolant spilled on it because it was bein used as a table for a pump (non conductive  but still)


are all those caps electrolytic or other types?

[ifonlyeverything]

C151 and C239 were tantalum, replaced both with ceramics from my eBay TH ceramic kit. The others I replaced were electrolytics.

[ifonlyeverything]

The problem of a buzzing transformer and fluctuating output voltage is back -- except this time the linear regulators are still working fine! Applying pressure to the top of the unit apparently causes this problem. It looks like this flexes the super shitty IDC box headers they used to join the A2 and A3 PCB. I can't easily Ohm this out so I guess I'll just replace it. The PCBs all have thick insulating paper underneath and they definitely are not cracked, so I don't think it's mechanical pressure inducing a short or open circuit or anything like that.

[coppercone2]

yeah xanatrax sorenesn IDC breaks I seen it pop out before but that is a cold welded cable (it cold welds the wire to the fork via the crimp tool), its like a crimp spot weld.

some people touch it up with solder but it might be difficult

the bond area for those IDC cables is basically the area of two prongs of a sheet metal fork (the thin inside part of the fork, like between fingers (bronze I think, maybe steel), once it goes it goes

just reseating the wires with a screw drivers is a temporary solution, your not remaking the cold weld. its the same technology as this
https://telcodepot.com/premium-punchdown-tool-1328

[ifonlyeverything]

yeah xanatrax sorenesn IDC breaks I seen it pop out before but that is a cold welded cable (it cold welds the wire to the fork via the crimp tool), its like a crimp spot weld.

some people touch it up with solder but it might be difficult

the bond area for those IDC cables is basically the area of two prongs of a sheet metal fork (the thin inside part of the fork, like between fingers (bronze I think, maybe steel), once it goes it goes

just reseating the wires with a screw drivers is a temporary solution, your not remaking the cold weld. its the same technology as this
https://telcodepot.com/premium-punchdown-tool-1328

The A2-A3 board connection is done with a box header to box header connection -- there is no IDC cable. Anyways, I was targeting the wrong area. It was the damn tab-mount snubber on the diode being slightly bent, just enough to ever so slightly contact the case of the power supply when mechanical pressure was applied! I bent it downward, toward the PCB, and the problem is now gone.

Now to get some Nichrome wire for a load test and to figure out how to trim out the error between CV and measured output voltage. The error is 0.1V for CV <9V; 1.2V for CV of ~24V; and 0.6V for CV of 36V. So that is definitely strange.

[coppercone2]

well that cable is insulation displacement (IDC) did you see how the crimps are made? Not sure if it works exactly like a MTA crimp (individual press in), but it does cut through ribbon cable insulation. you don't strip that

there is a tiny fork that cuts the insulation and then I assume cold welds itself to the wire in a tiny point pressure point.



to make those cables you just lay it in unstripped...

I never saw a soldered version of it in any test equipment, definately not in any xantrax or sorensen I worked on (numerous)