Identifying weird resistor in faulty SMPS

[badgerthing]

Hello!

I restore a lot of vintage computers, as well as make new parts for them.

I am trying to repair a Compaq SLT 386s/20's charger brick.
The output capacitors (2 x 330uF 35V) had leaked. The input capacitor seems fine (150uF 450V).
I've replaced all the caps except the input cap, as I don't have one, but will get one in my next order.
I've tested many parts in-circuit, and by removing and testing alone, including primary side switching MOSFET (BUK456), switching diode (MUR615CT), and SCR (BT151).
I've replaced the Unitrode UC80669 on the primary side. I assume this is a switching power supply controller, but can't find datasheet.
There are some SMD parts on the bottom. Passives and transistors on the primary side, and some smarts on the secondary (op-amp, 5V linear regulator with error signal (2951ACM), 74HC00 (replaced it, and lifted a track, which I've since fixed), and an unmarked SO-8 package.

I can't just replace the power brick, as it is a 6-pin+shield thing with smarts, and I don't know the pinout (I could work that out if I had a working one).
This SMPS has an AC LED and Battery LED. They both pulse at about 2Hz when connected to mains, with a very faint tick noise. Connecting the SMPS to the laptop itself, makes the LEDs dimmer and changes the flashing rate. The voltage at the output of the 2951ACM never reaches its intended 5V, but pulses up and down between about 1-3V.

While testing the parts, I found a resistor that is confusing. It seems to measure right on 1 ohm, but I don't have the best equipment to measure low ohms.
But the colour code is definitely not 1 ohm, unless some kind of alternate colour code for a shunt or something?
It appears to be Violet-Red-Green-White-Brown, which is 72.5G 1% - definitely not that.
Other way around is 19.5k 0.1% - seems more plausible.

In the circuit, it is in parallel with a 12V TVS diode (SA11). 1 ohm doesn't seem like it should be across a diode. This diode has cathode connected to Source of main switching transistor, and anode connected to one of the primary windings and the negative side of the input capacitor (150uF 450V).

Is there a reason a 1 ohm resistor would be there? Any why is it colour coded like that?

[George Edmonds]

Hi

It sounds like this is the total load sensing  resistor so about or under 1 Ohm is normal. a ticking SMPSU is normally a sign that there is a short on the secondary side.  Be aware that the electrolyte that leaked from the capacitor/s is both highly conductive and corrosive, it MUST be totally removed/cleaned away.

George G6HIG Dover UK

[badgerthing]

Hi

It sounds like this is the total load sensing  resistor so about or under 1 Ohm is normal. a ticking SMPSU is normally a sign that there is a short on the secondary side.  Be aware that the electrolyte that leaked from the capacitor/s is both highly conductive and corrosive, it MUST be totally removed/cleaned away.

George G6HIG Dover UK

Thanks.
I cleaned the electrolyte thoroughly. It was only on the top side underneath the caps themselves. They were the old brown or orange Nippon Chemicon that are always leaking.

Any ideas regarding that weird colour code on the 1 ohm resistor?

I've traced out some of the circuit as best I can.

[drussell]

In the circuit, it is in parallel with a 12V TVS diode (SA11). 1 ohm doesn't seem like it should be across a diode. This diode has cathode connected to Source of main switching transistor, and anode connected to one of the primary windings and the negative side of the input capacitor (150uF 450V).

Is there a reason a 1 ohm resistor would be there? Any why is it colour coded like that?

Are you sure it is a resistor, rather than an inductor?  That would probably measure about one ohm when you try to measure the resistance of it.  Edit: I just dl'd and looked closer at the photo and on the board it is indeed marked R501, so I guess that's not it.  I guess I will have to study the photos closer when I have a moment. 

Have you thoroughly checked for things like shorted diodes on the secondary side?

Ticking is indeed, usually a short....  or faulty capacitors...  (sometimes the bootstrap reservoir cap)
That, or some other fault in the bootstrap supply for the controller in general, or lack of proper auxiliary winding powerback or other method for the main supply to get power back to the controller and take over supply duties (if it is that style,) where the controller ends up getting just enough power to just start to run but then dies because it doesn't get main power.  (Which is usually due to a fault like a short on the secondary side, of course...)

[badgerthing]

Thanks.

I've checked all the diodes. Various through-hole ones, and a few SOT-23 single and dual diodes, removing from circuit where necessary.
The SCR was tested by powering an LED and triggering with 5V - not too thorough. I could probe to see if that is activating (which I believe it shouldn't, as it's a crowbar).

[drussell]

Thanks.

I've checked all the diodes. Various through-hole ones, and a few SOT-23 single and dual diodes, removing from circuit where necessary.
The SCR was tested by powering an LED and triggering with 5V - not too thorough. I could probe to see if that is activating (which I believe it shouldn't, as it's a crowbar).

That should be enough for an SCR test... 
They usually are either working properly, completely shorted, or won't trigger at all, so it is probably fine.

I won't have a chance to look closely at the photos and trace it all out until at least later today, but I think those color bands could well be indicating 0.952 ohms, although 953 is the usual value when it's a 3 significant digit tolerance and that really does look like red in the photos rather than orange...  The brown is a very thin band, and often these small value, especially flame-proof series resistors, etc. have kind of wacky color band schemes where the extra bands mean things like series type (flameproof, fusible, etc.) and temperature, etc. but it is probably supposed to be about .95 ohms, (white, green, red/orange, with a ÷100 silver divider band) ignoring the thin brown band since you usually start at the thick band....

Is there any chance that is actually orange, rather than red, in real life rather than camera photo? 

0.953 is a standard value

[drussell]

I could probe to see if that is activating (which I believe it shouldn't, as it's a crowbar).

Oh, and if it really is there for an over-voltage-protection crowbar, you could temporarily leave it disconnected for testing, (just for testing,) to be sure it isn't being triggered prematurely and causing all your woes due to faulty protection circuitry somehow. 

That would at least help narrow down your list of potential problem sources if it at least came up and ran without the SCR, then you could run it for a quick moment to see if the voltage really was soaring too high or if you're hunting flakey OVP triggering circuitry or whatnot....

[badgerthing]

Thanks.

I've checked all the diodes. Various through-hole ones, and a few SOT-23 single and dual diodes, removing from circuit where necessary.
The SCR was tested by powering an LED and triggering with 5V - not too thorough. I could probe to see if that is activating (which I believe it shouldn't, as it's a crowbar).

That should be enough for an SCR test... 
They usually are either working properly, completely shorted, or won't trigger at all, so it is probably fine.

I won't have a chance to look closely at the photos and trace it all out until at least later today, but I think those color bands could well be indicating 0.952 ohms, although 953 is the usual value when it's a 3 significant digit tolerance and that really does look like red in the photos rather than orange...  The brown is a very thin band, and often these small value, especially flame-proof series resistors, etc. have kind of wacky color band schemes where the extra bands mean things like series type (flameproof, fusible, etc.) and temperature, etc. but it is probably supposed to be about .95 ohms, (white, green, red/orange, with a ÷100 silver divider band) ignoring the thin brown band since you usually start at the thick band....

Is there any chance that is actually orange, rather than red, in real life rather than camera photo? 

0.953 is a standard value

It definitely looks red to me. There is a large ~2W 300ohm on the board with same-coloured green casing, and that is definitely orange.

I couldn't see pulses to the SCR with multimeter, but if they were fast maybe it wouldn't catch. I can get the scope out if worth it.

One of the annoyances is that there is an unmarked SO-8 in the secondary circuit, with some passives around it. One of its pins connects to the base of an NPN transistor, which then powers the LED inside the optocoupler.
The open-collector ERROR output of the linear regulator, also connects to this transistor base, which is interesting as it seems to be an NPN transistor, not PNP. (U7 marking = BSR13)
If the SO-8 were a standard dual or single comparator, that pin (7) would be an output.

I could also probe for optocoupler triggering. Do you think the optocoupler is for protection purposes?

[drussell]

Does the power to the controller IC collapse at the flash/tick rate?

[badgerthing]

Does the power to the controller IC collapse at the flash/tick rate?

Primary side controller IC? I don't know the pinout. It's a Unitrode UC80669.
Secondary side controller IC? I can't see one unless the unmarked SO-8 is it, which I also don't know the pinout of.
Pin 8 of this SO-8 connects to the output of the linear regulator. None of the pins connect to the linear regulator's GND.
The linear regulator output and GND power the 74HC00.
The linear regulator pulses up and down between 2 and 3V on the multimeter as the ticks happen. I will get the scope out now and get a capture.
I am just trying to make sense of the linear regulator's circuit now. Being the SO-8 version, I don't think it's actually 5V, but adjustable. I am seeing where it connects so I can determine what the correct output voltage is, and if it produces that out-of-circuit.

Also, there is some irregularity.
When I first got this PSU and laptop a few weeks ago, it had this problem. Then it actually worked once (when it worked, it would only give me solid LEDs when connected to the laptop, otherwise, ticks). Then never again since. The laptop works fine powered from the internal battery connector with a lab PSU. I don't know the pinout of the connection between brick and laptop, so I can't test it in that regard.
Then I opened it and replaced the caps to no effect.
Sometimes when I turn it on, the battery LED is off instead of blinking along with the AC LED, and I can't hear ticks.
I know that in normal operation, the AC LED should simply be on.

[drussell]

Primary side controller IC? I don't know the pinout. It's a Unitrode UC80669.

Yeah, but you can deduce the important stuff....
Like, it appears that C8 (SMT, on the bottom) is the filter capacitor for the controller supply, which likely will be receiving the bootstrap power from wherever that comes from.  Are you sure it's good?  Not leaky?

You can essentially just probe all the pins and see what they're doing and deduce...
You don't really need the datasheet, it's an 8-pin SMPS controller... 

[badgerthing]

Primary side controller IC? I don't know the pinout. It's a Unitrode UC80669.

Yeah, but you can deduce the important stuff....
Like, it appears that C8 (SMT, on the bottom) is the filter capacitor for the controller supply, which likely will be receiving the bootstrap power from wherever that comes from.  Are you sure it's good?  Not leaky?

You can essentially just probe all the pins and see what they're doing and deduce...
You don't really need the datasheet, it's an 8-pin SMPS controller... 

Oh for sure, but I don't know very much about how SMPS work. Learning as I go. Thanks for your help!

C8 measures 1.2nF out of circuit. I don't have the tools to measure leakage. Would you expect 1.2nF for supply filter for a SMPS IC? Seems low compared to logic ICs.

I don't get a stable voltage across pins 4,5; 1,8; or 4,8. I didn't test other combinations for now (it's a bit hard without the right probes, ensuring i don't short anything [currently shopping for a better electronics multimeter along with better probes, but that's for another post]).

I can probe what you've likely deduced as the power supply pins of the UC80669 with a scope. I would need to use my isolated step-down transformer.

[badgerthing]

From looking at other DIP-8 SMPS ICs, and the tracks for this one, I think pins 5 and 7 are the power pins.
The multimeter shows it oscillating between about 11.5 and 13.8V at about 0.5Hz. That sounds more correct.

[badgerthing]

Here's a video of the blinking: https://photos.app.goo.gl/7AEjuz4ASLDfZw2n8

When connected to the powered-off laptop, the battery LED comes on and both are brighter, blinking at same rate. Perhaps one of the 6 pins in the cable are for "power on" or something.

[drussell]

I'm not sure I would be plugging a faulty power supply into a working laptop, risking frying the internal PSU.   

That pulsing power and tick-tick-tick is definitely not how it normally behaves when not plugged into the laptop.  Personally, I don't think I'd plug it into the laptop until I fixed the power cycling.  If you want to simulate having extra capacitance on the output, just add a capacitor at the power brick's output terminals.

It is interesting that the SLT/286 power supply is a completely different build, despite both the SLT/286 and SLT 386s/20 being covered by the same service manual, using the same internal power supply and having the same specifications for input voltage, current, etc. (10 - 18.2 VDC, 18W steady, 23W peak) and both AC adapters being rated 20W steady, 30W peak.  (I highly suspect that they are actually completely interchangeable, especially since you can find P/N 110353-001 supplies which supposedly replace either, supposedly in stock for $56 USD, though who knows if that is NOS or USED, etc.  I expect that it is just that the SLT/286 would have usually shipped with the earlier, less-surface-mounty version.)

It sounds like the problems with the SLT/286 model brick (110353-001) are usually either the RIFA X/Y capacitors letting the magic smoke out, (as they do,) or, even more commonly, shorted tantalums, of which there are a bunch in the SLT/286, but they moved to surface mount on the SLT 386s/20 PSU (118460-001) instead of the big daughter-boards.

That yellow cap in yours by L2 looks like a multi-layer ceramic, rather than a tantalum, so I expect that if it is a similar design all the tantalums' functionality were replaced with surface mount ceramics on the bottom.  I wish the thing were here in front of me, "virtual probing at a distance" is rather difficult.   

Just for reference, here's what the inside of the 110353-001 for the SLT/286 looks like:

[drussell]

C8 measures 1.2nF out of circuit. I don't have the tools to measure leakage. Would you expect 1.2nF for supply filter for a SMPS IC? Seems low compared to logic ICs.

No, that's probably for setting the timing/frequency or something then...

I'll have to "BigClive" it later (vertical flip one of the layers and print out) to reverse-engineer and wrap my head around what it's probably doing since we've no data for the UC80669 to be able to easily tell why it's either power-cycling itself due to supply issues or constantly being forced into shutdown by an external trigger.

[drussell]

From looking at other DIP-8 SMPS ICs, and the tracks for this one, I think pins 5 and 7 are the power pins.
The multimeter shows it oscillating between about 11.5 and 13.8V at about 0.5Hz. That sounds more correct.

Edit: You could try holding it up at 13.8V with an external supply and see if it stays running.  Note that it may have an internal shunt regulator that will only handle, say 10ma, so you probably need to limit your potential input current from an external 14-15V supply if you try this on some of these little 8-pin chips.

MANY TVs that I've fixed have been issues with bootstrapping the SMPS controller ICs, usually due to (electrically) leaky capacitors (often un-labelled cheap blue MLCC ceramics if it's a Samsung-built piece of crap) not allowing the Vcc to come up high enough, quickly enough before the main logic board locks out the high voltages or faulty controller ICs.  (I'm looking at you, FAN6755U...  )

I would also try freeze spray. 
(...or stick the whole thing in the freezer for a while... and if that has no effect, bake lightly in toaster oven... )

[Labrat101]

Any ideas regarding that weird colour code on the 1 ohm resistor?

Hi i saw that   resistor . Its NOT a 1ohm
it sits parallel with the diode .
it is a high 9.2 meg resistor  if its reading 1 ohm its bad 
its common to do this across a diode .
If its reading that low it will make the crow bar circuit jump in and out .
Replace with high meg  resistor or it will kill the BUK456

Its a very cheap PSU you would be best to buy a Universal power
supply if your not familiar with these units .
 I repaired 2 today about an hours work . also check that the
control voltage is 2.48 volts .
measure the pulses on a dummy load before plugging it to a laptop or what ever .
The ramp should be flat with no load and increase to a full ramp at about 2 amps . if the pulse is not symmetrical the main inductor has a shorted turn or the control IGB is failing .
That's assuming you have checked the large 400v capacitor and it has been replaced .

 

 

[drussell]

Hi i saw that   resistor . Its NOT a 1ohm
it sits parallel with the diode .
it is a high 9.2 meg resistor  if its reading 1 ohm its bad 
its common to do this across a diode .
If its reading that low it will make the crow bar circuit jump in and out .
Replace with high meg  resistor or it will kill the BUK456

That is entirely possible, so the OP should look into this.  I haven't had time to look at the circuitry at all.

It doesn't really explain why it worked once, but if it's a position that would keep falsely activating the crowbar, that would certainly cause the thing to cycle without actually locking out on a real fault....

Edit: I'm not sure how you're reading the color code as 9.2 meg, though, and that seems rather high (although again, I haven't actually traced the circuitry,) but 19.5k 10% would make sense to me if it were brown white green red silver but it looks more like grey to me than silver and 19.5K is not a standard value.

I honestly have no idea at this moment... 

[badgerthing]

I'm not sure I would be plugging a faulty power supply into a working laptop, risking frying the internal PSU.   

That pulsing power and tick-tick-tick is definitely not how it normally behaves when not plugged into the laptop.  Personally, I don't think I'd plug it into the laptop until I fixed the power cycling.  If you want to simulate having extra capacitance on the output, just add a capacitor at the power brick's output terminals.

It is interesting that the SLT/286 power supply is a completely different build, despite both the SLT/286 and SLT 386s/20 being covered by the same service manual, using the same internal power supply and having the same specifications for input voltage, current, etc. (10 - 18.2 VDC, 18W steady, 23W peak) and both AC adapters being rated 20W steady, 30W peak.  (I highly suspect that they are actually completely interchangeable, especially since you can find P/N 110353-001 supplies which supposedly replace either, supposedly in stock for $56 USD, though who knows if that is NOS or USED, etc.  I expect that it is just that the SLT/286 would have usually shipped with the earlier, less-surface-mounty version.)

It sounds like the problems with the SLT/286 model brick (110353-001) are usually either the RIFA X/Y capacitors letting the magic smoke out, (as they do,) or, even more commonly, shorted tantalums, of which there are a bunch in the SLT/286, but they moved to surface mount on the SLT 386s/20 PSU (118460-001) instead of the big daughter-boards.

That yellow cap in yours by L2 looks like a multi-layer ceramic, rather than a tantalum, so I expect that if it is a similar design all the tantalums' functionality were replaced with surface mount ceramics on the bottom.  I wish the thing were here in front of me, "virtual probing at a distance" is rather difficult.   

Just for reference, here's what the inside of the 110353-001 for the SLT/286 looks like:

The yellow cap near L2 is a 100nF MLCC. It measures 88nF, and is between output capacitor +, and earth. Some kind of EMI thing I guess.
The two yellow caps beside the output caps are 1uF MLCC, and measure bang on (with a multimeter at least).

That picture you shared of the SLT/286 PSU, was created by me a while back, before I replaced the caps in it. The NCC SXF cap was leaking. It worked perfectly after replacement.
The tantalums were fine, but I replaced those anyway.
I have since regretfully sold that machine. I should have figured out the pinout while I had it.

The cable from PSU to laptop is hardwired in the SLT/286 PSU, but can be removed in the 386 one. The cable connectors are the same on both, so I imagine they are compatible.

Those IT parts sites always show stock for things they don't have. They're mostly just dropshippers, who will look for your part when you ask for it. They always turn up nothing for things you actually need.

Yea helping from a distance is hard.

Any ideas regarding that weird colour code on the 1 ohm resistor?

Hi i saw that   resistor . Its NOT a 1ohm
it sits parallel with the diode .
it is a high 9.2 meg resistor  if its reading 1 ohm its bad 
its common to do this across a diode .
If its reading that low it will make the crow bar circuit jump in and out .
Replace with high meg  resistor or it will kill the BUK456

Its a very cheap PSU you would be best to buy a Universal power
supply if your not familiar with these units .
 I repaired 2 today about an hours work . also check that the
control voltage is 2.48 volts .
measure the pulses on a dummy load before plugging it to a laptop or what ever .
The ramp should be flat with no load and increase to a full ramp at about 2 amps . if the pulse is not symmetrical the main inductor has a shorted turn or the control IGB is failing .
That's assuming you have checked the large 400v capacitor and it has been replaced .

How are you getting 9.2M ?

You can't use a universal brick. It's not a 'dumb' PSU. There are lines not used for power, and we don't know the pinout due to not having a working PSU.

Control voltage where?

Control IGB?

The input cap tests fine, and maintains 327VDC - the problem is further on. I will be replacing that cap either way, but I don't have them in my junk bin.

Thanks!

[George Edmonds]

Hi

With a color code of Brown, White, Green, Red, Silver the resistor is actually 19K5 at 10% 

George G6HIG

[drussell]

With a color code of Brown, White, Green, Red, Silver the resistor is actually 19K5 at 10% 

A 19.5KΩ doesn't seem to make any sense there to me.  Something like 1Ω does, though.

In the circuit, it is in parallel with a 12V TVS diode (SA11). 1 ohm doesn't seem like it should be across a diode. This diode has cathode connected to Source of main switching transistor, and anode connected to one of the primary windings and the negative side of the input capacitor (150uF 450V).

Is there a reason a 1 ohm resistor would be there? Any why is it colour coded like that?

Yeah, it actually does make sense, since it is the current sense resistor.  (There's a tongue twister!)   

It appears to be the current sense resistor between the MOSFET source and the 0 VDC (side of the +370 VDC input,) with a 12V TVS across it to prevent any spikes from exceeding 12V heading back towards the current sense input on the SMPS controller IC:



BTW, sorry, I meant to have flipped the top side to make it easier to compare to the bottom of the real board, I accidentally flipped the wrong image when I was pasting them together....

From looking at other DIP-8 SMPS ICs, and the tracks for this one, I think pins 5 and 7 are the power pins.
The multimeter shows it oscillating between about 11.5 and 13.8V at about 0.5Hz. That sounds more correct.

Yeah, it seems to have the same pinout as some other common 8-pin SMPS controller ICs like the UC3842/3/4/5.  Those particular ones happen to have a minimum startup voltage of about 16V, and lock out below 10V.

C8 measures 1.2nF out of circuit. I don't have the tools to measure leakage. Would you expect 1.2nF for supply filter for a SMPS IC? Seems low compared to logic ICs.
...
I can probe what you've likely deduced as the power supply pins of the UC80669 with a scope. I would need to use my isolated step-down transformer.

Yeah, C8 is the timing capacitor, it and a resistor sets the operating frequency via an RC constant.

I take it you already double checked that you didn't get a capacitor in backwards or something? 

I would take a close look at the power coming from the auxiliary winding, make sure it is actually starting to generate power into the auxiliary supply reservoir cap since it looks like the controller IC is only bootstrapped from the +370 DC rail, then is supposed to be powered from that auxiliary winding on the primary side.  The controller IC obviously needs to maintain voltage on Vcc to continue operating after trickle-charging up from the +370VDC.

You could check to see (I would even scope it, multimeters are slow) what happens with the voltage across that resistor, see if it looks like it is actually sensing high current and going into overload due to the current sense input "seeing" too much current, (which would point to some sort of potential overload or short on the output side,) although you do have LEDs come up, so it can't be a total short.

I still suspect it is probably a primary side power drop-out to the controller IC, but it still could certainly be lots of different possible things going awry, I suppose. 

Edit: If it were me, I think I would put any removed components back in, except pull the MUR615CT out, tack a 5W resistor of something like 470Ω or 1KΩ across the secondary winding so it doesn't try to generate crazy high voltage spikes on the secondary winding, then power it up and see if it runs.  If you measure the AC voltage across that temporary secondary resistor (yes, it will be high frequency but even most multimeters will show something resembling a measurement) you should see something, probably something like 25-30 or more volts AC because it should just run full-on in open loop.

If it runs, then your problem is probably something on the secondary side.

If it doesn't run open loop like that, then there must be a primary side problem that needs to be addressed first before it will ever work driving the real laptop output circuitry on the secondary.  If it doesn't run you'll have to check things like the power to the optocoupler collector, the controller IC, etc. that are on the primary side that might be dropping out.

[drussell]

That picture you shared of the SLT/286 PSU, was created by me a while back, before I replaced the caps in it.

Well, then, I guess there was no need to post that since you apparently already knew.   

Those IT parts sites always show stock for things they don't have. They're mostly just dropshippers, who will look for your part when you ask for it. They always turn up nothing for things you actually need.

Indeed.  That's why I specifically said supposedly. 

[schenkzoola]

Hello,

So, I have a similar power supply where I was troubleshooting this exact issue - Blinking green light, pulsing voltage on both the primary and secondary side.

I finally removed the battery from the laptop and plugged in the supply, and voila the light turned solid green and the computer powered. 

There must be some logic in the laptop that allows the power supply to stay running.

I have a power supply for a SLT/286 and the light stays solid green regardless of whether the computer is plugged in or not.  The power supply for the SLT/386 does not apparently.

[badgerthing]

Does it do the flashing when powered on but unplugged?