Maker Pro
Maker Pro

Need help repair dead switching power supply

Hi all,
I got this piece of dead china made switching power supply AVF brand labelled as 350w from a friend who previouly used on an old pc. When i opened the case nothing seemed burned and no sign of bulged caps. To prevent further damage, i connect a 100 watt light bulb(our country main supply of AC is 220 V) to replace the 5A glass fuse. I also use a piece wire to short the green ps on to black ground on the pc power connector.
When i switch on the mains, the fan not spinning nor the light bulb on. How do i go on further test to diagnose which component is at fault? On hand i have a digital multimeter,digital capacitor meter,and have a 40watt soldering iron.
Any help and suggesstion to repair this power supply is very much appreciated.
Thanking in advance.
 

Attachments

  • IMG_20151004_154928.jpg
    IMG_20151004_154928.jpg
    183.2 KB · Views: 412
  • IMG_20151004_155100.jpg
    IMG_20151004_155100.jpg
    205.9 KB · Views: 986
  • IMG_20151004_155745.jpg
    IMG_20151004_155745.jpg
    182.8 KB · Views: 484
  • IMG_20151004_160211.jpg
    IMG_20151004_160211.jpg
    230.5 KB · Views: 519
  • IMG_20151004_160714.jpg
    IMG_20151004_160714.jpg
    220.1 KB · Views: 705
Last edited by a moderator:
.


Sir mikey5791

I HEAVILY suspect . . . .
With no parts blown and fuse intact. . . . .

Get your self a BRIGT light or go outdoors for sunlight along with
a STROOOOOOOOONG magnifier and confirm that these wires/terminals
are not FLOATING within their solder joints.
Plus you need to inspect about another 100 or so connection / joints in the
high current paths.(WIDE foil)
Also, you can only be POSITIVE on some extremely BADLY oxidized situations,
until you pull a component and scrape its leads clean down to the bare copper and
then liberally rosin flux and then heavily re-tin and reinstall the unit.

Pee Ess . . . . .BTW
A wimpy little 40 watter is being very borderline inadequate on most of these joints thermal mass
that will be involved in trying to get up to proper reliable solder re flow heat levels.

Your Working photo marked up:

Ax4GUsc.jpg




73's de Edd
 
Hi Sir 73's de Edd,

I am a bit flattered by your formal salutation but later after seeing your country icon I can understand. Thank you for your suggestions and quick reply.

May I confirm with your statement saying "Get your self a BRIGT light or go outdoors for sunlight along with a STROOOOOOOOONG magnifier and confirm that these wires/terminals are not FLOATING within their solder joints. Is FLOATING meaning a break or crack in the solder pad, wire/terminals not fully in contact to the PCB pad?

First, I will relook at those 15 marked up solder points as you suggested to pull the component (unsolder),then clean it to bare copper, retin and then reinstall. I will do the suggested and report back soonest.

Have a fine day then.
 
Sir Michael . . . . . .

Indeed your perceptions are being correct . . . . .

May I confirm with your statement saying :


"Get your self a "BRIGHT" light or go outdoors for sunlight along with a STROOOOOOOOONG magnifier and confirm that these wires/terminals are not FLOATING within their solder joints.

Is FLOATING meaning a break or crack in the solder pad, wire/terminals not fully in contact to the PCB pad?

Exactly . . . . some times you can find a component that is so loose, and has solder that has eroded away so bad, that you can grasp it from one side of the board and watch it move on the other solder side of the board.

Sometimes the contact area has decreased so much that an arcing has occurred and those are the ones in such bad shape that they sometimes require removal to get them tinned and able to be soldered in to get good connectivity again. OR a new part could be installed.

The reason I suggested a magnifier is that is so apparent that a portion of a connection is not having good solder coverage.

The most of those joints will need a drop of rosin flux and then a HOT soldering iron put to them until you see a thorough flowing of solder and the making of a solid connection . Some will need some solder added.

I merely pointed the joints that SEEMED to be floating or POSSIBLY floating, your sharp eye will have to confirm for sure, as there must actually be a 100 connections on that secondary area fed from the switch mode power transformer just above.

ALso your POWER FET circuitry that drives the primary of the transformer is being up and across the transformer from my top 3 YELLOW ovals that I have marked up.

The left connection of /or/ near near CR103 also looks suspect, CR101 area also. Plus there is also more HEAVY foil circuitry all over that half of the board also.

You may just need to find ONE connection for that PS to be working again, but finding other marginal ones NOW, will be one more failure that will not occur TOMORROW.


I am a bit flattered by your formal salutation but later after seeing your country icon I can understand.

I would have thought that assigned titling would only have been of any REAL significance . . .to an Englishman.

But didn't Abdul Halim of Kedah REALLY use the more distinguished . . . . Sir Michael ?


73's de Edd



.
 
Last edited by a moderator:
Hi Sir 73's de Edd,

Upon your recommendations, I went outside the house under strong sunlight (which we have plenty of it here) and examine the solder as you had marked with a magnifier. Also I moved/jiggled each component on the top pcb to check for loose fittings on the other side. But so far I have not find any to be suspect yet. Hope to find that faulty connection some how to get the ps working.

I try to rule out the suspect one component at a time.From my attached pictures, I had desolder six(6) electrolytic caps from secondary side,then use my caps meter to check each of the caps,but the UF reading seems within 5% range as stated on caps. With power off, I checked resistance reading on the green thermistor SCK054,on scale 200ohm, the thermistor reads 6 ohm. This should still be good.

Then, I use diode range to check the four diodes next to the thermistor, each diode reads OL(over limit)on one side and about 594 ohm on other side. All four diodes should still be good. Next I try check transistors on primary heatsink but no idea how to do it. There are three transistors, my noob understanding is the left pin i assume is Gate(G),middle pin is Drain(D) and right pin is Source(S). I will use my digital multimeter on diode/continuity setting, first place the negative(-) black probe to Drain(D) while the positive red probe (+) on Source(S),next step is place the positive probe on Drain(D) and negative probe on Source(S),last step is place negative
probe on Drain(D) and positive probe on Gate(G).

On first transistor label C5353 2C, I got these readings:
-D&+S=739
+D&-S=1(over limit)
-D&+G=524
On second transistor label E13007 F2, I got these readings:
-D&+S=496
+D&-S=1(over limit)
-D&+G=540
On third transistor label E13007 F2, I got these readings:
-D&+S=490
+D&-S=1(over limit)
-D&+G=542

Does the above readings says anything important about the transistors, whether they are still good.

Hope my above simple tests can assist to further narrow down the ps fault.
Thank you for kind feedback.
 

Attachments

  • IMG_1002-caps removed.JPG
    IMG_1002-caps removed.JPG
    152.8 KB · Views: 281
  • IMG_1005-secondary caps position.JPG
    IMG_1005-secondary caps position.JPG
    194 KB · Views: 263
  • IMG_1006-caps removed.JPG
    IMG_1006-caps removed.JPG
    164.9 KB · Views: 329
  • IMG_1008-primary heatsink.JPG
    IMG_1008-primary heatsink.JPG
    164.6 KB · Views: 316
  • IMG_1004-secondary heatsink.JPG
    IMG_1004-secondary heatsink.JPG
    171.4 KB · Views: 254
Both those types of transistors are appear to be NPN TO-220.
With the same pinouts of 1.Base, 2.Collector and 3.Emitter. While looking at the front (Marked side).

So put your meter to diode test or continuity and put the pos lead to 1.Base and leave it there. Check collector and emitter pins with the negative lead. Should read as a diode.
Now swap the leads and the same. Should read OL.
I also check collector to emitter too. OL.

You will have to take them out of circuit to test them..

Martin
 
Good point Martin, it should also be noted that while in diode check mode, the readout is in volts (not ohms)
The expected readout is about .7v, which is the voltage dropped across the diode or transistor junction.
 
.

Sir Michael . . . . .


Yes indeed . . . you need to reverse track your technology back one step to bipolar transistors and exchange your DGS's for CBE's.
As per . . . . our Most Distinguished,Honorable and Esteemed . . . . Sirs . . .Martaine. . . et . . .Tha fios agaibh.


For this switch-mode application, that Motorola MJE13004 thru 9 series was absolutely just about as good as it EVER got, up to the era of the introduction of FETS to replace them.
They had fantastic switching rise and fall times, handled inductive transient spikes well and their innovative semi die design gave them a phenomenally low conduction resistance.

BUT with their showing of forward conductance voltages of ~594 Mv . . .you have good units, at least they are not hard crashed from C-E.

I see your tested PS caps resting upon the case/cover and all of the left ones, seem to be resting flat upon it, but do check on the far right two to see if any slight . .or . . .more . . .upward doming of their aluminum-i-ninny-yum-yum case tops is being evident .

I don't know your photo medium, but the pic of the caps is great with light,and depth of field. It's having the writing on the caps in fine focus and even on back to your meter front and only blurring when you get afar to the distance of the function switch.
(Therewith your illumination lighting was adequate to auto initiate a small iris closing, high F-stop equivalency and still producing a good sharp pic at depth. )

The other ones, of poorer definition were probably dependent on a minimal function flash.

See if you can dupe the lighting and conditions that produced the caps pic, and get me a top close up of the 14-16 pin IC that is to the left of the most massive silver alum heat sink.

I am interested in it and its nearby surrounding peripheral support components.
I suspicion that IC to be "someones" own frontal xx numbering but with it having a final 394.

Back when you had the caps in circuit, did you once try a power up and have the incandescent lamp in the background, being series connected as the fuse.

If so was there an initial very slight glow as the two largest AC line derived supply capacitors*** took on their initial charge, and then . . .no glow . . . . and nothing ?

*** 1st photo . . .the 2 largest black electrolytics at the very top of the pic.

(From just above . . .How comes "spell checks" always assumes that I don't be smarts e'nuffs to spels "electrolytics",
when in actuality, THEY apparently don'ts be ables to spels ELECTROLYTICS ! themselves)
( Maybe they needs mo' books larnin'.)

Standing by to get your feedback . . . . .

73's de Edd

.
 
Hi Martaine2005, Tha fios agaibh, Sir 73's de Edd,

Very much appreciate all your points. As suggested, I refer my earlier pic (img1006) on the two PS caps on the far right and physically inspect for slight dome. Noticed esp on caps labelled D.S.1000uF10V is a bit bulging on the bottom black rubber filling although caps meter reading still ok. I will replace the suspect cap when doing the final solder back in pcb.

Attached img 1009 being top close up of 16 pin IC as requested (left of most massive silver alum heatsink). Is this the secondary heatsink? FYI, the almost unreadable label of the 16 pin IC is E150A (on first row) KA7500B (on second row).

Sir 73's de Edd quoted "Back when you had the caps in circuit, did you once try a power up and have the incandescent lamp in the background, being series connected as the fuse." Yes,about a few days ago I had try a power up but didn't notice any glow. Last night, out of curiousity I replaced the 100 watt bulb with a 60 watt (the six caps still off circuit and short wire connecting the green wire to ground on ATX connector removed) and try a power up in a dark room.Voila! I got SLIGHT GLOW lasting like 2 seconds then no glow later. Is this a good sign? My noob knowledge is the series light will glow with full brightness if there is a short. It is a silly mistake before this I was working under normal light making it hard to notice slight glow. (LOL)

Hoping for further guidance and patience in tracing the fault. I am enjoying learning new skill each day.
Thank you all for kind reply.
 

Attachments

  • IMG_1009-16 pin IC.JPG
    IMG_1009-16 pin IC.JPG
    214.9 KB · Views: 572
Mikey, good call replacing the bulging caps. A cap can read normal on a normal dvm but show bad on an esr meter.

I wouldn't be alarmed with the initial glow of your 60w lamp as it sounds like normal power up to me.
Do make sure you have a load on the dc side as well because circuitry may inhibit an output without it.
 
Hi Martaine2005, Tha fios agaibh, Sir 73's de Edd, Dorke

With my limited knowledge, how do I test if the 16 pin IC KA7500B (Fairchild SMPS Controller) is good?

Yesterday out of curiousity, I tested two black diodes (next to a 100ohm resistor) below that small transformer "MTREE-19-1" and found 1 diode labelled CT(first row) SB240(second row) shorted. I used diode range and it read 001 at one end and swap probes and get 001 the other way. Is this a schottky diode 40 Volt 2.0A type? The nearest diode stock that i have is a LT116(first row)2A05(second row) which gives a peak repititive reverse voltage of 600V and average rectified output current of 2.0A. Can I use this to replace the shorted CT SB240 ?

I have not resolder the six removed caps due to waiting replacement of this shorted diode. Once this diode is replaced and the six caps soldered on pcb and put a load(maybe a 12V DC fan will do ?) and the series bulb still connected, is it time to power up to see if the PS will work?

Appreciate all the feedback and suggested points on the troubleshoot.
Thank you all.
 
You are correct about the SB240 a "schottky diode 40 Volt 2.0A".
You need a schottky diode for replacement.

Don't know what LT116 is.
The 2A05 (si diode)is not good for replacement because:

1. Vf to high
2. si, too slow
 
.



.




Sir Michael . . . .

Sorry for the time lag in response, but seems like that I am now being involved with 6 or 7 other assists.

Your dim lamp test is telling us, at the least, that the raw dc power supply is properly charging up those two
high capacitance electrolytics and that their FWB rectifiers supply section is OK.
This is being from a stone cold start of the supply. (The dormant filters have dissipated all their charges.)
Now the 13007's are just not getting their base drive to activate the units main power supply portion.

BTW . . . the revealing of the last photo being a Fairchild KA7500B, which is being same-same as the
common as dirt TL494. . . . .which is what I was expecting .

Since you could find no poor solder connections and the dim bulb test showed the main raw DC power supply
apparently operating fine.
Lets get all of the electrolytics reinstalled and in their proper places and make NO reverse installation of
their + and -'s.

STOP ON THE VERY NEXT ACTON SINCE I SEE THAT YOU HAVE FOUND A DEAD SHORTED PS DIODE
. . . . .SO READ ON


On a power up of the supply thru the incandescent lamp . . .100 watt preferred.
That KA7500B /TL494 then should have +10-12 volts or so being on its pin 12 when using pin 7 as the meter negative ground.
That will let us know if the units separate standby power supply is working.
If not, that is what we need to analyze and test next.
And then . . . . .AND THEN !

You said:

Yesterday out of curiosity, I tested two black diodes (next to a 100ohm resistor) below that small
transformer "MTREE-19-1" and found 1 diode labelled CT(first row) SB240(second row) shorted.
I used diode range and it read 001 at one end and swap probes and get 001 the other way.


Looks like you have just stumbled on a STRONGLY possible cause of the PS's failure to start up.

This unit has a simple supply that is called the auxiliary or standby supply, The smaller "MTREE-19-1" transformer is its transformer
and is used to produce a 5V standby supply and MORE importantly, the B+ which feeds your KA7500B /TL494, if it no workee your 13007 's
receive no base drive to make THEM workee.

If you have an electronics JUNQUE salvage /collection of parts from tv's other computer power supplies , etc . . .if so look at all of their diodes in this size of casing.
In doing your testing, and using the diode test mode, a Schottky
( "spell check" is now insisting that it is a Scottish diode . .hee hee )
type of diode will read a forward junction conducting reading of approx 240 mv instead of the 500-700mv readings that you typically have been encountering on other diodes.

Now this sets particular design, in this portion, is not needing anywhere near that SB240's full 2 amp capability, with but a mere 100 ma or so being needed in our situation.
Additionally, trace down the cathode of that SB240 along its foil path and you should find an electrolytic capacitor connected in, which is doing the filtering for its supply line.
Any chance that it is one that you have already tested ?, if not, pull and test it. I wish that you ALSO had an ESR meter, since the high frequency square wavelets hitting these
capacitor functions really HAMMER them !

But looks like we will just have to hope for the best.

I have placed a relevant portion a schematic of a standby circuit that have some portions being applicable to your set.
(Starting with its " T3 EE-19A " power transformer.)
Its Q3 power transistor is your one that shares the 13007 transistors heat sink, your limited photos have not shown me if a second small signal transistor is involved like this units Q4.
But from my additional photo *** below, looks like it might be the transistor located just below the NEC 2504 optical isolator. at the edge of your PCB between the HOT/COLD
chassis division, it may be used like this units U1.

On this unit, I see the D7 providing for a low voltage supply and its higher capacitance 1000 ufd and 470ufd filters.
The unmarked diode, which is series connected with R73 . . . 100 resistor . . .provides voltage for your KA7500B /TL494 and it is using a lower 47 ufd of filtering. And like I said,
I expect this supply line to be connected to pin 12 of the KA7500B /TL494 BUT it is now receiving RAW AC, because your diode is shorted and that could have damaged the companion
electrolytic filter on its line.

Now check out how much of this info is being relevant to your units situation.

*** Reference:

sxnDs9d.jpg





Partial Schematic . . . . Ach Himmell . . . . .icht bin der Russkies ! . . .







sy-300atx.gif









73's de Edd


.







.


.
 
Last edited:
It may be a trick of the light, but in that first pic in post #14 the two lower pins (i.e pins 1 and 2) of the NEC2501 (opto-isolator) look blackened. If so, check if there's a dead diode between those pins.
 
Hi Alec_t,

Checked the lower two pins of the NEC2501 (opto-isolator) indeed is blackened.
How do I check if there's a dead diode between pin 1 & 2? Which part is anode and which part is cathode ? Sorry my knowledge on this is limited.

Thanking in advance for the steps.

Hi Sir 73's de Edd,

Your prompt reply and assistance really encouraged me to continue on looking for the faulty component. As you informed earlier, I traced cathode of diode SB240 along foil to a electrolytic cap rated at 220uF16V, the cap meter reads within the cap range. Sorry my limited budget still makes an esr meter a luxury item as of now.

My previous find from a tv junk board nearest to the roasted diode SB240 is diode LT116 2A05 (600V 2.0A) which is a higher rating than the original. For the purpose of testing, I used the LT116 2A05 in place of SB240. I also replaced all the six electrolytic caps on board (four are new caps)-refer image 1020. On power up AC through 60watt bulb in series, and set my multimeter on DC range,I put black probe on pin 7 of KA7500B, red probe on pin 12 give a reading of 4.8V DC. Is this a good sign? I think this should be good.

Hoping for next step on diagnosing further and getting the PS to working condition soon. Thank you for the patience.
 

Attachments

  • IMG_1020-old vs new caps.JPG
    IMG_1020-old vs new caps.JPG
    100.1 KB · Views: 287
Hi Alec_t,

On carrying out the diode test on NEC2501 (opto isolator) two lower pins,I put black probe on pin 1 while red probe on pin 2 got reading 1(over limit). Then i swapped the probes and got reading 704. This I believe is a good component.

One question though. Do I need to test on the upper two pins? Sorry for noob questions as I am inexperience on PS troubleshooting.

Thanks for the simple test. Any other suggestions on getting the dead PS to work again is appreciated.:)
 
So it looks like the blackened half is ok. The pins 3 and 4 usually connect internally to the collector and emitter of a transistor. A simple diode test won't help in testing that half.
 
.


Sir Michael . . . .


Most honorable apologies for the time lag between my responses. . . .
Your newly acquired level of supply voltage is being less than is normally required for the units KA7500B /TL494.
Maybe I can now have you perform a test which will enlighten you on the switching time . . . .time on / time off . . .of diodes.

One of the diodes in the standby power supply secondary is proper in its being a SB240 and the other is having your replaced it with a common slower switching diode, like you would be using in an AC line supply with its s l o w
50 or 60 cycles frequency.

The current SB240 is just fine in its opearating at the expected upwards to 100,000 cycles of frequency.

In AC applications , when a diode is instantaneously switched from a forward conducting state , to a non conducting state it NEEDS some time to return to a NON-CONDUCTIVE state, so it behaves leaky / quasi short circuited for a little time period in reverse direction.
Your SB240 should recover from having any potential "backwards" conducton within ~55 nanoseconds.
Your 2A05 would need in the order of 1250 nanoseconds.
I'm going by its twin brother the 1N4007, since I did not see a Reverse Recovery switching spec being applied on a common diode.
BECAUSE, once while testing in a mock up jig for hand picking some select matched diodes, I just threw in a 1N4007 and found that units recovery time to be ~ 1250 nanoseconds.
So it looks like a SB240 can recover from any partial reverse conduction about 22 time faster than a 2A05 could.

I don't think that I saw if you are still using a test lamp in series with the AC power or in the fuse holder. Somehow I think that you were, so lets test in the same condition that you read that supply voltage to the KA7500B /TL494.
Get the set operating in that same prior testing condition, and then test for the DC voltage at the cathode of the SB240 and log it down.
Then test for the voltage at the cathode of the 2A05 and log it.

NOW you shut down any AC power input, until the PS caps bleed down and then swap the positons of the two diodes and test for voltages and log their voltages as before.
See what the voltages differences were with the 2A05, as we know that the SB240 should have been optimal for use at either positon.

Question?
Was the standby schematic close enough to work with ?, as I see others that are having a variant in their using a LM7805 for the 5V standby supply.

Instead, on this schematic, it is using an adjustable TL431 reference and a control feedback loop thru the 2501 / 817 optical isolator.

BTW . . .with such low voltage involved, on the cold side of the NEC2501 circuitry, I'm not visualizing any problem, it may have just been the lighting angle, as the photo transistor sides pins, on the unit, looks good . . .or dirty board..

How about info now on the different diodes voltage readings.

If a series test lamp was used, of what wattage and if in series with the main ac line or being across the pulled fuse's clip.

I'm surprise that you didn't find a fast switching diode in your used TV " Junque" parts. One can expect a fast switching diode to be used in the vertical circuit power supply and the audio power supply IF they are using small, separate windings on the horizontal output transformer to get those voltages.

Oh yes, was the 100 ohm series diode associated with the original SB240 diode or the other diode ?

73's de Edd


.
 
Top