Safety Question: Harvesting parts from a PC power supply

[Pasky]

So my PC power supply recently took a dump on me, it over voltage'd motherboard and completely fried my PC.  However there's lots of great inductors I'd like to scrap from it before I throw it in the trash.  Now I know these suckers can hold a charge still and I have had it powered off for roughly a week now but I don't wanna be taking a trip to the morgue this week.

What's the safest way to discharge the PSU and safest way to know it is indeed discharged before I start tinkering with it?  Thanks.

[Pasky]

No, there's inductors I want from the board.  I'm aware the capacitors are the deadly portion  .  I just don't feel like jolting myself trying to remove an inductor.

[Ian.M]

Its good practice to include bleed resistors in the design to make the main reservoir capacitors discharge safely if not in use, but the typical 'Wun Hung Lo' cheap PSU probably hasn't got them and if the primary side startup circuit has blown (and it is a faulty PSU), I've known them to give a nasty shock hours later.  Why risk it?

Discharge all big caps on the primary side into a high power wirewound resistor.  27K 5W is good and will discharge a fully charged 100uF 400V cap in under 10 seconds.  If you connect a multimeter on DC volts across the resistor and wait till it reads under 12V you wont have any problems.   If you don't have any suitable resistors, a 40W 240V incandescent bulb is usable and will discharge the caps much quicker, but check it on mains before and after use.

[Sirius631]

Discharge all big caps on the primary side into a high power wirewound resistor.  27K 5W is good and will discharge a fully charged 100uF 400V cap in under 10 seconds.  If you connect a multimeter on DC volts across the resistor and wait till it reads under 12V you wont have any problems.   If you don't have any suitable resistors, a 40W 240V incandescent bulb is usable and will discharge the caps much quicker, but check it on mains before and after use.

I would add that after discharging the big caps, you should wait a few minutes and then do it again. Big caps have a noticeable charge recovery, which can be significant and may catch you out.

[retiredcaps]

Of the 50 or so consumer electronic devices that I have fixed over the last 5 years, 2 of them (4%) have had the SMPS power supply big filter capacitors hold a full charge (165V DC) over several days/weeks.  With PFC circuitry involved, those big capacitors could hold up to 400V.

Ian.M gives excellent suggestions on how to discharge and measure the big filter capacitor.

[f5r5e5d]

really have to discharge the caps to touch/remove any parts safely

I just don't feel like jolting myself trying to remove an inductor.

[Electro Fan]

Of the 50 or so consumer electronic devices that I have fixed over the last 5 years, 2 of them (4%) have had the SMPS power supply big filter capacitors hold a full charge (165V DC) over several days/weeks.  With PFC circuitry involved, those big capacitors could hold up to 400V.

Ian.M gives excellent suggestions on how to discharge and measure the big filter capacitor.

What's the longest (after being disconnected from the source) those capacitors would hold roughly 25% and 10% of the charge?  How long to 0%?  Thx

[Seekonk]

Taken lots apart and have never seen one that didn't have bleed resistors.

[grumpydoc]

Taken lots apart and have never seen one that didn't have bleed resistors.

But is the time taken to stick a DMM across the cap(s) really too much effort? - given that it might save your life (or at least prevent an unpleasant surprise).

[sleemanj]

Taken lots apart and have never seen one that didn't have bleed resistors.

Had a re-used printer power supply go tango uniform on me a few weeks back (my own fault, screwed it down to hard and flex'd the PCB).  Good thing I discharged the caps before taking a closer look, using a screwdriver being the butcher I am, a real good crack from them.

Don't trust an already faulty device to not also be faulty in a most unpleasant way.

[Ian.M]

Its quite common to have no explicit bleed resistor in 220v-240V only "China Export" grade equipment.  The topology is direct rectification to 340V DC often on a marginally rated 350V cap, with either push-pull CT transformer or a single ended one depending on the power level and associated chopper transistor(s) and driver chip. The chip gets its startup power from a high value dropper direct off the 340V bus, but when this is implemented as a single resistor it usually under excessive voltage stress and is prone to going O/C leaving a dead PSU with a main cap that charges fully with no bleeder.

110/220V switchable supplys usually have bleeders, because the two main caps that are part of the voltage doubler for 110V operation are in series for 220V and need balancing resistors.

[Pasky]

Perhaps a stupid question, but how exactly do you use a bleed resistor, or even recognize it?  I understand the concept that you take the resistor and short the cap with it but I'm unsure how they work when they are integrated into a PCB or what they look like.

[Evil Lurker]

Perhaps a stupid question, but how exactly do you use a bleed resistor, or even recognize it?  I understand the concept that you take the resistor and short the cap with it but I'm unsure how they work when they are integrated into a PCB or what they look like.

Most PSUs have two bleed resistors back to back. If you follow the line and neutral input traces to the positive and negative leads on the the primary capacitors you should find one or two resistors in a high value (in the 200-400K range) that connect the two together. Those will be your  bleed resistors.. they can be either SMD or through hole.

[Ian.M]

A bleed resistor is a high value resistor that must be rated for the working voltage of the cap, usually permanently connected directly across the cap's terminals.  In the case of motor start/run caps it may even be integrated into a plastic housing round the terminals.  It continuously wastes a little current so that the capacitor will discharge to a safe voltage within a minute or two.   If the equipment case has enough screws, it will take longer to open it than that.

In the case of high power equipment, it may have a much lower value bleed resistor that is connected by an interlock switch that is activated by a long front panel screw.  By the time you've got the screw all the way out the equipment is safe to open.

[retiredcaps]

Of the 50 or so consumer electronic devices that I have fixed over the last 5 years, 2 of them (4%) have had the SMPS power supply big filter capacitors hold a full charge (165V DC) over several days/weeks.  With PFC circuitry involved, those big capacitors could hold up to 400V.

Ian.M gives excellent suggestions on how to discharge and measure the big filter capacitor.

What's the longest (after being disconnected from the source) those capacitors would hold roughly 25% and 10% of the charge?  How long to 0%?  Thx

If you are asking me about the bolded parts above, in the two pieces of equipment (a Motorola PVR and LG 2234 monitor), both held their full charge of 165V DC.  I wrote about the Motorola PVR here.

http://www.badcaps.net/forum/showthread.php?t=41319

The Motorola PVR only discharged after I used a power resistor to manually discharge it. The time it takes to discharge depends on the value of your resistor.  Typically, after a couple of minutes it should be below 5V DC.

If I didn't answer your question correctly, rephrase it please.

[Red Squirrel]

If I'm not sure I just check capacitors with a multimeter before I touch them.

[Electro Fan]

Of the 50 or so consumer electronic devices that I have fixed over the last 5 years, 2 of them (4%) have had the SMPS power supply big filter capacitors hold a full charge (165V DC) over several days/weeks.  With PFC circuitry involved, those big capacitors could hold up to 400V.

Ian.M gives excellent suggestions on how to discharge and measure the big filter capacitor.

What's the longest (after being disconnected from the source) those capacitors would hold roughly 25% and 10% of the charge?  How long to 0%?  Thx

If you are asking me about the bolded parts above, in the two pieces of equipment (a Motorola PVR and LG 2234 monitor), both held their full charge of 165V DC.  I wrote about the Motorola PVR here.

http://www.badcaps.net/forum/showthread.php?t=41319

The Motorola PVR only discharged after I used a power resistor to manually discharge it. The time it takes to discharge depends on the value of your resistor.  Typically, after a couple of minutes it should be below 5V DC.

If I didn't answer your question correctly, rephrase it please.

I was just curious - for example,  in your project where the caps held over 160 Volts for 4.5 weeks, how long could such a charge be maintained - several months, a year, longer?  Or is the message that no matter how old something appears or how long it might have been out of service it's prudent to check capacitors with a DMM before proceeding?

[Ian.M]

Always treat medium sized and large used high voltage caps as potentially dangerous unless they have been stored with a short across the terminals.
http://en.wikipedia.org/wiki/Dielectric_absorption

I wouldn't go so far as always using a voltmeter, but our workshop ESR meter had a piece of brass strip foam taped to its case next to the terminals, and *EVERY* cap being tested was always shorted on the strip before testing - a policy I introduced after having to replace its fusible input protection resistor twice in the same month.

[Pasky]

Thanks for all the information everyone.  I'll get to it this weekend with all this mind.

[retiredcaps]

I was just curious - for example,  in your project where the caps held over 160 Volts for 4.5 weeks, how long could such a charge be maintained - several months, a year, longer?

Hmm, I don't know for sure, but I'm guessing since it already lasted one month, that several months or even a year is not out of the question.

The unit is now in use so I can't take it out and just leave it unpowered for a year to find out. 

[Cherry]

I agree with In the case of high power equipment, it may have a much lower value bleed resistor that is connected by an interlock switch that is activated by a long front panel screw.  By the time you've got the screw all the way out the equipment is safe to open.

[Zero999]

So even big brands like lite-on don't put bleeding resistors properly.

I'm feeling lucky being still alive.

Bleeding resistors are unnecessary for capacitors in switched mode power supplies. The circuitry connected to the rectifier forms a very good bleeding resistor and adding an extra bleeding resistor would just waste power.

If in doubt, measure the voltage across the capacitor and if it's too high, connect a 10k resistor in parallel and leave it for a minute before touching it.

[Cherry]

I agree with Ian.M.

[timb]

A large insulated handle screwdriver has always worked for me.


Sent from my Smartphone