Generic question re. Safety in working around high-voltage Capacitors (in PSUs)

[bsodmike]

Hi all,

Over the years I've mainly worked at a bench with voltages well below 20Vdc (in low-current applications) and just recently started working on a Constant current DC load project (which has been educational!)

I've got a couple faulty ATX PSUs that I want to open up but I want to take as much precaution before doing so.  The large caps are 240V rated and here are my first questions:

1) how must I guard myself when opening up the ATX PSU? I've opened these up in the past and just took the precaution of not touching the caps.

2) I plan to attach a 60W filament bulb across the terminals of the cap (on the rear PCB); someone once told me that they connected a bulb with multimeter leads as a way of probing the cap.  Is this a good idea? I was thinking of adding a couple 1MOhm resistors in series as well? (I'd calculate power-dissipation first and use high-wattage resistors as needed.

3) I would not power this on with the case opened for sure.

Any other bits of advice?

Thanks!!

[wn1fju]

First, it is not a good idea to touch ANYTHING with your fingers unless you are completely familiar with the circuit.  I would take a multimeter and test the voltage across the capacitor terminals (and from the capacitor terminals to the chassis ground also).  If it is more than a fraction of a volt, you can short the capacitor with a resistor (something like 50K).  Again, test it with a multimeter after you are done.  Also, many supplies already have resistors in the circuit to discharge the caps and you might find that they are already at zero voltage.

Once you read up on how a switching power supply works, it should not scare you as much as you seem to be.  However, for safety reasons, when debugging an opened up switching power supply, you should always plug it into an isolation transformer, not just simply into the AC power outlet.  And if you are working anywhere around the primary side of the supply, don't hook up oscilloscope ground clips unless you really understand the circuit (although this is mitigated with the isolation transformer).

[bsodmike]

First, it is not a good idea to touch ANYTHING with your fingers unless you are completely familiar with the circuit.  I would take a multimeter and test the voltage across the capacitor terminals (and from the capacitor terminals to the chassis ground also).  If it is more than a fraction of a volt, you can short the capacitor with a resistor (something like 50K).  Again, test it with a multimeter after you are done.  Also, many supplies already have resistors in the circuit to discharge the caps and you might find that they are already at zero voltage.

Sure, this makes sense and it is what I planned to do.  I guess my "fear" is due to all the warnings regarding dealing with high-voltage supplies.  Could you clarify one thing though - I assume these warnings are in the context of opening the powering up the device right?  Should one be as guarded if the PSU is totally unplugged and you're taking it apart (and verified the caps are near 0Vdc)?

No 2.  How weary/dangerous is a charged cap - say 240Vdc; I would assume this is to be treated as hazardous right?  Of course, as you said, most designs should have resistors in parallel to drain them - but that's an assumption again!

Once you read up on how a switching power supply works, it should not scare you as much as you seem to be.  However, for safety reasons, when debugging an opened up switching power supply, you should always plug it into an isolation transformer, not just simply into the AC power outlet.

This is to "float" the PSU and to prevent any return to mains-earth reference?

And if you are working anywhere around the primary side of the supply, don't hook up oscilloscope ground clips unless you really understand the circuit (although this is mitigated with the isolation transformer).

Most definitely; I'm careful when it comes to using scopes due to the mains-earth reference on 'scope probes, and I do not have any differential probes (yet!).  Even in that scenario my oscilloscope manual states:

Risk of electrical shock - maximum input voltages
The maximum input voltage on channel inputs must not exceed:
● 400 V (peak) and 300 V (RMS) at 1 MΩ input impedance
● 30 V (peak) and 5 V (RMS) at 50 Ω input impedance
For the external trigger input, the maximum input voltage is 400 V (peak) and 300 V (RMS) at 1 MΩ input impedance.
Transient over voltages must not exceed 400 V (peak). For further specifications, refer to the data sheet.
Voltages higher than 30 V (RMS) or 42 V (peak) or 60 V DC are regarded as hazardous contact voltages. When working with hazardous contact voltages, use appropriate protective measures to preclude direct contact with the measurement setup:
● Use only insulated voltage probes, test leads and adapters.
● Do not touch voltages higher than 30 V (RMS) or 42 V (peak) or 60 V DC.

That last bit is quite handy, and at my workbench I have several Wiha 1000Vdc insulated tools.

[bsodmike]

Once you read up on how a switching power supply works, it should not scare you as much as you seem to be.  However, for safety reasons, when debugging an opened up switching power supply, you should always plug it into an isolation transformer, not just simply into the AC power outlet.

Oh, I'm not planning to power it up at all - I just want to inspect the PCBs for any burn marks as on one of the units the cap let out its magic smoke and a bang a while back.

So I guess the main takeaway is be cautious around large capacitors that may have a high charge/voltage on them.  The slightly concerning bit is having to take the PCB out of the chassis before I can get to the capacitor terminals.

[wn1fju]

Well, if you are just taking the thing apart for inspection, there should be minimum risk.  I would wait several minutes if the thing had been on to give the caps a chance to discharge.  Many test equipment manuals also suggest a few minute waiting period.  Even so, it has always been my policy to check the voltages on the caps before I start fiddling around any further (just in case the supply doesn't have a built-in discharge path).

The caps on the primary side of a switching supply are not usually very high microfarads, although the voltages can be several hundred volts because they are operating on the rectified AC line voltage (sometimes doubled).  However, I wouldn't want to touch one that is fully charged.  And I'm not about to try touching one to answer whether or not it is lethal.  I recently took a 15 kV hit from a CRT and that was enough excitement for one day.

Yes, the isolation transformer is to remove the earth reference so when you ultimately do stick your hands where they shouldn't go, you won't be hurt!

As far as scope probes go, you should be using a 10:1 probe and the kind of voltages inside a switching power supply should not harm your scope.  At least, it has never hurt any of mine.  Normally, one uses a scope to verify the switching transistor is switching.  After that, you can usually move over to the secondary of the power supply.  At that point, it sort of looks like a regular linear supply with low voltages that won't hurt you.

[TheMG]

Most of the time, the capacitor will discharge within a couple minutes of unplugging the power supply. In some circumstances, usually due to a fault, the capacitor(s) may take much longer to discharge.

I just confirm the voltage with a multimeter to make sure it's discharged, if it needs additional "help" to discharge to a safe level, I just put a 100ohm resistor across the cap.

One point to keep in mind, is that the capacitor in a non-PFC power supply will be charged to roughly the peak voltage of the incoming AC - about 170VDC for a 120VAC input, or 340VDC for a 240VAC input. In power supplies with a PFC pre-regulator, the cap will be charged to about 400VDC regardless of AC input voltage.

The caps on the primary side of a switching supply are not usually very high microfarads, although the voltages can be several hundred volts because they are operating on the rectified AC line voltage (sometimes doubled).  However, I wouldn't want to touch one that is fully charged.  And I'm not about to try touching one to answer whether or not it is lethal.  I recently took a 15 kV hit from a CRT and that was enough excitement for one day.

Plenty enough to be lethal if it goes through the right path - through your heart. One sweaty hand on the negative and the other sweaty hand on the positive of the capacitor would be one of the worst case scenarios.

However, if say one or two fingers on the same hand come across the charged capacitor, it will definitely get your attention but is unlikely to have any serious effects.

No need to be afraid of it though, you just have to be aware the potential hazard exists, and take precautions, like not touching any part of the primary circuit of the power supply until you've verified the capacitors are in fact discharged to a safe voltage. If you must poke around in there while the power supply is energized, or capacitors potentially still charged, keep one hand in your pocket. When doing energized testing/troubleshooting, use a proper isolation transformer.

Generally speaking, you'd probably have to be VERY careless to receive a lethal shock from the stored energy in the capacitors of a powered off switching power supply.

[Cerebus]

Very old, but still good, safety film on working with high voltage from the UK Atomic Energy Authority:

[Miti]

No 2.  How weary/dangerous is a charged cap - say 240Vdc; I would assume this is to be treated as hazardous right?  Of course, as you said, most designs should have resistors in parallel to drain them - but that's an assumption again!

It is deadly for those who don’t know what they’ve doing. However, after repairing countless TVs, monitors, PC power supplies, VCRs, DVD players, three phase VFCs, you name it, in the 1990s and 2000s, I’m still here to write this reply. My worst zap was while repairing a CRT monitor. I repositioned the board vertically to do some measurements on the bottom side and I forgot there’s a power button on that side of the board... and that I didn’t unplug the cord. My left finger was right across the capacitor. Europe, 230V.
The two white craters and the pain lasted quite a long time.
But since my motto is “ living involves a high risk of dying”, I did not let that event to discourage me.
Know what you’re doing and focus on what you’re doing and you should be fine.

[m3vuv]

maybe worth mentioning caps have a habit of recharging back up after beeing discharged,thats caught me out a few times!

[tggzzz]

Do not presume that an isolation transformer will make it safe to work on equipment; they don't solve some dangers and can introduce others. That has been well addressed elsewhere on this forum, so I'm not going to repeat that.

If you are going to use a scope anywhere near mains, make sure that you use the right type of probe. FFI see the praxis and safety references at https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/ There are many poor techniques that you can "get away with", just as you can get away with running across roads without looking - usually.

[bsodmike]

Potentially a silly thought, but what about wearing rubber neoprene gloves as a "just in case"? I recall Fluke selling insulating gloves for technicians working with mains A/C etc...

Thanks to all for the advice, I'm glad that my assumptions were mostly sensible so I think I should be OK!

Yet, the left hand in my pocket is a rule I picked up from my days in the University lab.  Unfortunately, we were never exposed to hazardous work, mostly TTL level stuff with analogue ICs and PICs etc (circa 2002).

[Cerebus]

Gloves as personal protective equipment is one of those things that you need to consider a lot of factors before deciding whether they will increase safety or decrease it for a particular application. It's the difference between saying "Thou shalt wear gloves" anywhere where they might apparently be a help and making a proper safety assessment.

Things to consider (for glove applications in general, not just the case in handl):

• Does the loss of dexterity make things more dangerous?
• Does the loss of sensation make it more rather than less likely that you will come into direct contact with a hazard?
• Do gloves increase the risk of trapping a hand or reduce your ability to quickly withdraw a hand?
• Is the combination of a chemical hazard and the glove material more dangerous than the hazard plus skin? (e.g. Nitric Acid)
• Will gloves give a false sense of security and make the user more careless or more likely to bypass safety measures? ("I don't need to isolate it, I'm wearing insulating gloves.")

If you think, I'm sure you can add more to the list, these are just off the top of my head.

On the subject of the "one hand in your pocket" rule, I prefer to put one hand behind my back, palm outwards, gripping my belt or the waistband of my trousers. It pulls your elbow into your side making you more manoeuvrable and reducing the risk that you'll knock or catch something with your elbow. I feel a bit more controlled with that arm doing something active (gripping) than I do with it just passively hanging there, hand in pocket. The active nature also makes it less likely that you'll unconsciously bring your spare hand back into use without active thinking; the extra half second it takes to pull it around your body is time enough to think "Hold on idiot, that hand's back there for a reason".

[Gyro]

I recently picked up a Martindale voltage indicator / tester ( https://martindale-electric.co.uk/product/vt12-voltage-indicator/ ), together with a proving unit, really cheap on ebay, as I was planning on doing some, um, electrical work. I've taken to using it as my default device for general safety checks on medium-high voltage circuits too. These things are designed to be fool-proof and bullet proof - Cat IV rating at all sensible voltages, still indicate above ELV even if the battery is flat, nice captive probes and no possibility of accidentally selecting the wrong DMM socket or range. Obviously there are other manufacturers too.

Strictly you should use them in combination a proving unit too, but for low energy circuits the risk is minimal, you can check them against a known-live voltage source periodically.


P.S. Only use a 2-probe one - never trust a single-contact tester.

[aqibi2000]

Wow.

This isn’t this overly politically correct


Short the storage caps with a 10w resistor or a light bulb and call it a day

[bsodmike]

P.S. Only use a 2-probe one - never trust a single-contact tester.

Ha, I was about to ask about the Martindale setup when I saw it comes with a proving unit.  Looks brilliant!

I've used a simple Fluke (non-contact) tester in the past for simple home-electrical work and you can very very easily fall for a "false negative", if you're not careful or not aware of the pitfalls in using one.  My rule of thumb now is to "verify, and verify once more" and always take a DMM reading at the end.

[aqibi2000]

You can’t test DC charge stored in a Cap with a non contact tester even if it was a testable tool. It’s only for AC magnetic fields

[bsodmike]

You can’t test DC charge stored in a Cap with a non contact tester even if it was a testable tool. It’s only for AC magnetic fields

Indeed; Sorry, I was not meaning to claim that a non-contact tester could determine DC charge stored on a cap.  My comment was just in general regarding non-contact testers and their tendency to provide false negatives in dangerous situations.

[bsodmike]

Very old, but still good, safety film on working with high voltage from the UK Atomic Energy Authority:

Fantastic, just what I was looking for.

[garrettm]

You can’t test DC charge stored in a Cap with a non contact tester even if it was a testable tool. It’s only for AC magnetic fields

That’s actually not true. My Fluke AC VoltAlert "non-contact voltage tester" detects the changing electric field on a nearby conductor (or insulator!). The changing field need not alternate in polarity either. For example, I charged a Teflon rod by rubbing it with a piece of fabric and then waved it around. This was enough to trigger the light on the VoltAlert. So you can detect DC electric fields if they are in motion relative to the probe. And also when there is a significant ripple voltage on a DC rail (would need to be some unusually large ripple though).

Otherwise you are correct about in circuit testing, they won’t help for checking if a capacitor is charged or not.

To conclude, you can easily see false flashes due to electrostatic effects of rubbing your feet on carpet, sliding cloth against plastic, plastic on plastic, or even a shirt across skin on a very dry day. And that goes for static discharges as well. So long as the tester detects a changing electric field of a large enough magnitude, the probe will trigger.

[Gyro]

Very old, but still good, safety film on working with high voltage from the UK Atomic Energy Authority:

What a great video! The suspense near the end is edge of the seat stuff (I thought he was going to fry) - I can't see the most complacent of audiences dozing off during that one.


P.S. Those opening credits would do justice to any sci-fi film of the day!

[Cerebus]

What a great video! The suspense near the end is edge of the seat stuff (I thought he was going to fry) - I can't see the most complacent of audiences dozing off during that one.


P.S. Those opening credits would do justice to any sci-fi film of the day!

Back in the day when the BBC was ramping up colour television BBC2 was an evening-only service. During the day BBC2 transmitted 'test transmissions' which consisted of training films about colour television so that the nation's TV repairmen could get up to speed and other training or documentary films produced by government institutions. This was from 1967 onwards until at least the early 70s. Quite a lot of government institutions had their own film units producing documentary and training films for institutional and public use. As well as UKAEA, there was a Post Office film unit, the Central Electricity Generating Board had one and they all contributed films to the BBC2 test transmissions. Doubtless there were others that i have forgotten.

The above video was one of the films in the latter category. How I came to know it existed was becuase I used to watch these when I came home for my lunch from junior school. I was probably the only 8 year old kid in the country who knew how to degauss and align a colour telly, and how not to get fried working on plasma physics experiments (the work featured in the film is early UK fusion research). Because of the age, and the time-slot it was transmitted in, it's a safe bet that nobody was getting fried on air, but I agree you don't know that until the very last minute.

[tkamiya]

I'll just leave you with one story.

My own "near death" experience happened with unit unplugged and switched off.  I was working on a tube amplifier.  Internal voltage was about 500 volts, so similar to your switching power supply.  Those electrolytic capacitors hold enough charge to terribly shock you and hurt you for long time. 

I had power supply wired for 1000V at first.  I had few filter capacitors with bleeder registers across it.  Bleeders burned and were open circuit.  I remembered to crowbar it with a screw driver.  I rewired it to 500V for testing.  I dropped my guard.  One time I forgot to discharge with screw driver, it bit me. 

I lost use of my arm for a whole day.  I had a small but deep burn where my finger touched the voltage source. 

Approach any unknown circuit as a live one, especially if you are troubleshooting.  All it takes is once.  In right circumstances, 400V can kill you.