Stunning 3000A, 2000A, 3000T power supply discovery - MUST READ IF YOU HAVE ONE

[EE-digger]

After spending some time on the repair thread for these scope power supplies, I did further research and also prepped a Murata supply which I had on hand, to replace the original Cherokee/GE CCH125 in the scope.

So today, I took the factory supply out of my pristine, dare I say virgin, DSOX3104T.  The scope is super clean inside and out.  Hadn't really looked at the supply much because ... it's a supply.

Also, haven't used the scope in weeks but it's plugged in a day at a time.  First thing I noticed was that the ac mains cap was hot as hell.  (thermo and IR camera in the days to come)

Back of the board looks like crap.  Not blackened but clearly looked like it was sitting in an oven at high temp. 

I'll cut this short with more to come later.

I've designed SMPS from low level to high level modulars into many medical products but do not profess to be an expert in topologies.

So, need some help here.  The 420v mains cap is sitting at 397 volts dc      This is on a 120v mains.

This is why these scopes are cooking.  Might this supply be using doubling up front?  You just can't do it with these bleed resistors.  They will be running red hot, and right below the cap with thermal coupling via the cap leads.  With very little thermal decoupling, these 1W resistors are running at 0.4W, and the thermal rise is increased due to their close proximity.

The new Murata is sitting on the bench, most parts cool to touch and the voltage across the cap sitting at ..... drumroll ..... 168vdc.  I could hear it saying "take me", "take me", "take me".

Ok little Murata, as soon as the connectors arrive.

It seems like these scopes are a ticking bomb, almost guaranteed to blow on European and other 230v mains, although I don't know what their design would do there.

[perdrix]

It's quite normal to run 120V setting through a doubler (for switchers that convert mains input to DC as the first step).   When set for 240V input the doubler isn't used.

In the absence of other faults, I would suspect that the cap in question has gone high ESR, and therefore gets seriously toasty.

If there are bleed resistors across the cap, you could consider increasing their value - its not uncommon for the value of the bleed resistor to be unnecessarily low.

[EE-digger]

Thanks for that. 

I'm going to refurb several of these with a higher value bleed resistor and an 8000 hour cap of identical size.  I just realized that in the powered off scope, this supply has aluminum chassis on one side which will conduct heat if close enough (not so sure it is) and the bright plated shield on the other side, which may be closing the "oven door" so to speak.

Open on the bench, the resistors get up to 100C and the cap to 60+C.  I may emulate or just put it back in the scope to test the "oven" effect.  But that +397v cuts it too close to the 420v rating for my taste.

And as they say, the proof is in the pudding.  For a very low hours scope, this supply looks "cooked", just short of board blackening / burning.

But for my scopes, this baby is gone

[TheSteve]

You've got me curious so I will open my scope this weekend and have a peak. It has been connected to power 100% of the time for many years.

[EE-digger]

Think of it less as a scope and more as a crypt.      Can you post a pic of the board bottom?

I've installed new cap and resistors.  Excellent improvement and thermal images posted below.

It's a design flaw.  Someone missed the +397 volts sitting across the series Rs.

Resistors were at 100C, now at 64C, capacitor pins are about 10C below that  (new R value = 68K each,  bleed time to 50V = 24s)
There are also hot resistors lower in the image but they are only pushing 70C+

Capacitor top was pushing 70C, now at 46C  (what looks like a hot ring around the capacitor is the emissivity of the shrink wrap tricking the camera)

[TheSteve]

Opened up my scope. My cap is sitting at 398 VDC with the scope off - ouch
Top side of the cap is 60 C
Mine has been powered for 6-7 years.
Can't say I'm impressed they use a doubler in it - explains the rather poor efficiency this thing has.
My cap measures ~82 uF and 1.5 ohms ESR. It is time for replacement. I will be ordering a new cap and higher value bleed resistors.

You can see some discoloration of the PCB, but nothing horrid.

[2N3055]

Would somebody, please,  be so kind and say what are resistor case size and what capacitor size is?
What is resistor value now? What are you going to replace them with?
I would like to order them up front, before opening the scope...
Thank you..

[TheSteve]

Original Capacitor is 420 volts @ 100 uF
I will be replacing it with a 450 volt @ 100 uF
Diameter is 25mm
lead spacing 10mm
Original height is pretty short at 23mm. I will be replacing it with a taller capacitor as there is plenty of room below the metal shield.

I was thinking of maybe a ESCOS-TDK B43644B5107M000
I'm open to other suggestions, but can really only consider parts from Digikey.

Original resistors are 22K, size 2512
I will replace with 68-100K each.
Perhaps Bournes 75K CRM2512-FX-7502ELF

[2N3055]

Thanks a lot, much appreciated!

[EE-digger]

B43644B5107M000 is the same cap I used.  It's 25 x 25 and is fully below the module height.  Digikey has stock.

Removal is a royal PITA.  I first sliced the silicone beads around the cap with a thin blade.

After vacuum sucker and solder wick couldn't finish the job, I ended up using some Chipquick low temp alloy, then heated and pressed each lead downward with a stainless steel pin, alternating leads until it dropped.

[TheSteve]

Rework done.

Installed the ESCOS-TDK B43644B5107M000 capacitor and Bournes 75K CRM2512-FX-7502ELF resistors.

PCB was a little toasted under the original 22K resistors, the glue that held the parts on was very crispy and burnt.

Good for another 10 years I hope.

[EE-digger]

Looks good and the 75k resistors will keep it comfortable for a long time.  My new Murata is hissing at me so the cleaned up original may go back into the scope.

[tautech]

Seems an unusual SMPS design needing bleed resistors on the HV DC cap. 

Normal modern SMPS design has the controller remain working until the HV DC cap gets discharged to low and safe levels.

[TheSteve]

Seems an unusual SMPS design needing bleed resistors on the HV DC cap. 

Normal modern SMPS design has the controller remain working until the HV DC cap gets discharged to low and safe levels.

The datasheet tells the tale - it has very poor efficiency(but good power factor). I guess the design is minimum 15 + years old but even then it is pretty weak.

[EE-digger]

I have a second Murata coming in at noon.  If it is quiet, it's going in and the cleaned up original goes on the shelf until I need to use my service agreement.  Then the old supply goes back in, etc. 

And Steve, you can't say these supplies haven't been burned in 

[SeanB]

One capacitor only says this has active PFC, so very likely the active PFC is running all the time. those 22k resistors are likely a start up supply for the main controller, which then will use the main transformer to generate it's own supply, which will also drive the boost converter doing active PFC.

would also sufggest checking for dry joints on the transformer, and that join, likely a heatsink, above those new resistors does look rather suspect with a ring around it.

[HighVoltage]

Thanks for pointing this out.
Time to open all my scopes, I guess.

I had a similar problem on a 7000 series Agilent scope.

[TheSteve]

One capacitor only says this has active PFC, so very likely the active PFC is running all the time. those 22k resistors are likely a start up supply for the main controller, which then will use the main transformer to generate it's own supply, which will also drive the boost converter doing active PFC.

would also sufggest checking for dry joints on the transformer, and that join, likely a heatsink, above those new resistors does look rather suspect with a ring around it.

Yes it does have active PFC and as you go higher on the resistor values the minimum start voltages rises.

[tautech]

One capacitor only says this has active PFC, so very likely the active PFC is running all the time. those 22k resistors are likely a start up supply for the main controller, which then will use the main transformer to generate it's own supply, which will also drive the boost converter doing active PFC.

would also sufggest checking for dry joints on the transformer, and that join, likely a heatsink, above those new resistors does look rather suspect with a ring around it.

Yes it does have active PFC and as you go higher on the resistor values the minimum start voltages rises.

So they're not HV DC cap bleed resistors then ?
And instead just the ordinary dropper resistors found in most any SMPS to allow it to charge the controller Vcc cap to exceed the startup threshold ?

[SeanB]

The resistors go to a trace not ground, likely with a small capacitor on the other side of the board. Higher value just means longer time to bootstrap the PSU, as most of the controller IC's will have undervolt lockout that holds them off till the capacitor charges up to a high enough voltage.

you can see the trace leading off the bottom to the daughter board, and the capacitor, and as well on the other side of the daughter board the SOT diode that feeds it in operation, with another diode isolating the feedback and power pin so the controller can sense the transformer state for feedback.

[tautech]

The resistors go to a trace not ground, likely with a small capacitor on the other side of the board. Higher value just means longer time to bootstrap the PSU, as most of the controller IC's will have undervolt lockout that holds them off till the capacitor charges up to a high enough voltage.

you can see the trace leading off the bottom to the daughter board, and the capacitor, and as well on the other side of the daughter board the SOT diode that feeds it in operation, with another diode isolating the feedback and power pin so the controller can sense the transformer state for feedback.

These are another common failure point in SMPS when this cap goes high ESR.

[tautech]

The resistors go to a trace not ground, likely with a small capacitor on the other side of the board. Higher value just means longer time to bootstrap the PSU, as most of the controller IC's will have undervolt lockout that holds them off till the capacitor charges up to a high enough voltage.

you can see the trace leading off the bottom to the daughter board, and the capacitor, and as well on the other side of the daughter board the SOT diode that feeds it in operation, with another diode isolating the feedback and power pin so the controller can sense the transformer state for feedback.

Do you know the SMPS controller pt# Sean ?

[SeanB]

No, it can be any number of them, you find them almost to be generic, hundreds of part numbers, all doing roughly the same work. Same for the PFC, which will also be on this board.

[mikeselectricstuff]

Four 1206 resistors fit pretty well in the original 2512 footprint

[Someone]

Higher value just means longer time to bootstrap the PSU, as most of the controller IC's will have undervolt lockout that holds them off till the capacitor charges up to a high enough voltage.

Put 100k parts in place of the 22k, works fine!