A more idiotproof PSU, what is this feature called so I can shop for it?

[Myself]

Alright, I let the smoke out of my bench PSU this morning. Siglent SPD3303X-E. I was using it to charge a 55AH LiFePO4 cell, and after disconnecting it to shuffle some things around on the desk, I hooked it back up but got the polarity backwards. Turns out the current limit limits what the supply will source, but not what it will allow to flow if you drive it in the wrong quadrant. Sounded like popcorn... Nothing actually burst into flame though, so the fire extinguisher merely got grabbed, not discharged.

Anyway, I'm looking for replacement recommendations. I've actually been super happy with the supply prior to this, and I'd be fine replacing it with another of the same, and I'll just try to be more careful. But I'm curious, since I've evidently become the better idiot we hear so much about, are there nicer units with more idiotproofing?

Where would one even look in the spec sheet to find out? I think that's the crux of the question -- is there such a thing as a PSU that's not vulnerable to this mistake, and what is that feature called?

In lieu of robustness by design, I think I might just add fuses to whatever I get. The fault current here was probably on the order of 40-50A based on how hot the test leads got, and for a unit that can't source more than 6.4A in parallel mode, I feel like some 10¢ 10A fuses might've saved me.

[nctnico]

You aren't the first one to make this mistake. And some PSUs have crowbars which will short the battery when something goes wrong. The only way to safely charge a battery from a bench PSU is to connected a (big) diode in series. A fuse could be an option but likely it doesn't prevent damage.

[Wrenches of Death]

Why not just use a rectifier in series with the output of the power supply? I've been doing this for decades when charging batteries. Especially automotive batteries. I've got one rigged with a male plug and a female banana jack.

You'll have to adjust the power supply output setting to compensate for the voltage drop across the rectifier.

WoD

[2N3055]

I think that's the crux of the question -- is there such a thing as a PSU that's not vulnerable to this mistake, and what is that feature called?

It is called battery charger.....

Laboratory PSU is not it...

[bdunham7]

The feature might be called robustness with reverse polarity protection.  You are more likely to find it in older design linear PSUs with much less "bang for buck".  The BK Precision 1652 I have is one example of a supply that is protected against this sort of thing--but it is only a 24V 0.5A per channel (1A total) output.  I have some others that I suspect might be OK, but I'm not willing to experiment. 

A fuse might help in some cases depending how the failure occurs--if it is just thermal damage due to high currents, a fuse is your savior.  In other cases it won't help. I personally would use a 7.5A ATO fuse right at the battery in your application no matter what PSU you have.  You can always have other issues, like accidentally shorted leads, that would be a pretty serious issue with a battery that size.

If you don't have a bullet-proof PSU available, the best way to do accurate charging is to use a PSU that has a separate sensing terminal.  Run the charging current through a large rectifier and then connect the sensing terminal directly to  your battery via a resistor--there will usually be a spec regarding how large that resistor can be, but 1k should be enough.  That way you avoid the issue of the voltage drop across the rectifier.

Your burnt PSU may be repairable--crack it open and lets all have a look.  You may find that some parts near the front end have sacrificed themselves and the bulk of it may be OK.  And you can modify and hack it now that the warranty has....expired?   

[bdunham7]

It is called battery charger.....

Laboratory PSU is not it...

To be fair, there are plenty of products sold as battery chargers that would go up in flames just as fast in the same situation, or if not, are only saved by a fuse or breaker.  Or possibly worse, would simply apply maximal current to the reversed battery and charge it backwards until bad things happen.

[HKJ]

is there such a thing as a PSU that's not vulnerable to this mistake, and what is that feature called?

There may be some, but generally no. Most power supplies has a diode across the output to handle reverse polarity, this diode is not a super high power diode, but most likely in the 1A to 3A range.
A SMU may handle reverse polarity without any trouble (Mostly it do not have reverse polarity), but it is rather expensive.
I uses power supplies as chargers very often, but I usually have a relay on the output to prevent any mishaps.

[TomWinTejas]

You aren't the first one to make this mistake. And some PSUs have crowbars which will short the battery when something goes wrong. The only way to safely charge a battery from a bench PSU is to connected a (big) diode in series. A fuse could be an option but likely it doesn't prevent damage.

Newbie question.  Will a bench power supply with sense capabilities be able to adjust for the voltage drop across the diode automatically, or will this be too large of a gap versus the expected drop due to the wire's resistance? 

[tautech]

Newbie question.  Will a bench power supply with sense capabilities be able to adjust for the voltage drop across the diode automatically, or will this be too large of a gap versus the expected drop due to the wire's resistance?

Should be in the datasheet like this for SPD1000X models:

In the 4-wire SENSE compensation output mode: By using a separate measurement circuit, the supply can more accurately compensate for any voltage drops due to high resistance connections or long cables.
Maximum compensation voltage is 1V.

[bdunham7]

Maximum compensation voltage is 1V.

Good point.  Most such setups limit the sensing differential so that there isn't a runaway if the remote leads come loose.  So if the voltage is critical, a large power Schottky would be in order.  Or a MOSFET ideal diode solution.

[2N3055]

It is called battery charger.....

Laboratory PSU is not it...

To be fair, there are plenty of products sold as battery chargers that would go up in flames just as fast in the same situation, or if not, are only saved by a fuse or breaker.  Or possibly worse, would simply apply maximal current to the reversed battery and charge it backwards until bad things happen.

Sure, there are many bad battery chargers...
But a good one will be fit for purpose..


Also a word of caution: if you put output diode, and then use sensing to ensure good regulation, sense circuit might not be able to survive reversed battery....
Only safe way is to put diode in series with battery and live with the voltage drop, or compensate for it approximately by raising the voltage of PSU....

Unless PSU is specifically made to stand this type of use, there are no guarantees..

[mawyatt]

Maybe adding a series resistor in-line with each sense lead and a reversed biased diode at the sense input terminals should protect the sense terminals from a reversed battery condition, and allow regulation with a series protection diode to the battery.

Best,

[bdunham7]

Also a word of caution: if you put output diode, and then use sensing to ensure good regulation, sense circuit might not be able to survive reversed battery....
Only safe way is to put diode in series with battery and live with the voltage drop, or compensate for it approximately by raising the voltage of PSU....

The designs I've seen have diodes across the sense and output terminals that limit the voltage differential and short out any reverse polarity, that's why I said to use a 1k resistor to limit the reverse polarity current.  But you're probably right that just a diode is a more comprehensively safe system and a MOSFET ideal diode would pretty much solve all of the issues with the caveat that you'd need to design it carefully to use it at low voltages. 

[nctnico]

Maximum compensation voltage is 1V.

Good point.  Most such setups limit the sensing differential so that there isn't a runaway if the remote leads come loose.  So if the voltage is critical, a large power Schottky would be in order.  Or a MOSFET ideal diode solution.

A Schottky diode can easely have a drop of over 1V at serious currents. It is easier to just set the voltage a little bit higher.

[bdunham7]

A Schottky diode can easely have a drop of over 1V at serious currents. It is easier to just set the voltage a little bit higher.

Sure they can, but you can keep it well under 0.5V quite easily by selecting the correct part.  Or, use the ideal solution. 

The voltage drop actually is problematic for some situations.  If you are charging a single Li-ion cell at 4.1V and set the PSU to 5.1 volts, as the current tapers off the V_f will as well and then you overcharge your cell.  I actually charge cells and batteries like this at times (repairing something where there is an issue with the charging setup) and it is fairly handy to be able to just dial in the charge voltage and bulk current, then simply watch for the current to taper off.

I have a PSU that doesn't have any issues doing this, but if I had to do it with an explodable PSU I would try using a Schottky large enough that my V_f at the cutoff current was as low as possible.  For example, given a nominal 1A charge rate, the 5A rated SL54AFL gets the job done with a 250mV drop @1A and a 180mV drop @ 0.1A.  So I can set the current to 1A and voltage to ~4.25V and have little to no risk of overcharge, presuming I'm aiming for 4.1V and 0.1A cutoff and I turn if off reasonably soon after the current drops to the cutoff.

[Someone]

Maximum compensation voltage is 1V.

Good point.  Most such setups limit the sensing differential so that there isn't a runaway if the remote leads come loose.  So if the voltage is critical, a large power Schottky would be in order.  Or a MOSFET ideal diode solution.

A Schottky diode can easely have a drop of over 1V at serious currents. It is easier to just set the voltage a little bit higher.

More true:
A Schottky diode can easily have a drop of under 0.4V at 10A.

Depends which Schottky diode, and what you call "serious" current.

[BrokenYugo]

For the occasional lithium cell couldn't one just put a cheap protection board after the diode to provide overcharge protection?

[Someone]

Maybe adding a series resistor in-line with each sense lead and a reversed biased diode at the sense input terminals should protect the sense terminals from a reversed battery condition, and allow regulation with a series protection diode to the battery.

Robust supplies have that built in. Also noting that specifications for sense voltage/current can be: functional/operational, or absolute/damage.

[Myself]

A SMU may handle reverse polarity without any trouble (Mostly it do not have reverse polarity), but it is rather expensive.
I uses power supplies as chargers very often, but I usually have a relay on the output to prevent any mishaps.

So, the SMU is kind of where my mind is going. I have a few BOSS/BEP units that I picked up at auction. They're far too bulky to live on the bench for everyday use as PSUs, but perhaps I should change my thinking and make room! I believe a 4-quadrant power supply should be able to survive exactly this condition, because it's a normal part of the operating envelope.

The whole rest of the thread is talking about fuses and diodes, and since these are so simple and evidently common for a few folks to use, it makes me wonder why they aren't a normal part of the PSU output structure. Of course a power supply could be simple and cheap and omit all these things, but we pay three or sometimes four figures for them because they're robust and versatile and include all sorts of protections and limits. And of course I don't make a habit of hooking things up backwards, but I don't want to flush three or sometimes four figures down the toilet every time I brainfart.

I had hoped for someone to say "well yeah you idiot, you cheaped out and bought a Siglent instead of a Keysight, of course those have reverse polarity protection", but such advice has not been forthcoming.

Dang.

[nctnico]

A Schottky diode can easely have a drop of over 1V at serious currents. It is easier to just set the voltage a little bit higher.

Sure they can, but you can keep it well under 0.5V quite easily by selecting the correct part.  Or, use the ideal solution. 

The voltage drop actually is problematic for some situations.  If you are charging a single Li-ion cell at 4.1V and set the PSU to 5.1 volts, as the current tapers off the V_f will as well and then you overcharge your cell.  I actually charge cells and batteries like this at times (repairing something where there is an issue with the charging setup) and it is fairly handy to be able to just dial in the charge voltage and bulk current, then simply watch for the current to taper off.

Why would you want to charge a battery to the max voltage for a repair using a PSU? I never do that; only to a level where it is enough to do testing or get the device going from it's intended charging circuitry. With a Li-ion cell you'll be bypassing other protection circuits (like temperature monitoring and cell balance/voltage reversal) as well. In case of a functional Li-ion battery pack (not a cell), it should switch off once the pack is at the maximum voltage anyway.

[2N3055]

A SMU may handle reverse polarity without any trouble (Mostly it do not have reverse polarity), but it is rather expensive.
I uses power supplies as chargers very often, but I usually have a relay on the output to prevent any mishaps.

So, the SMU is kind of where my mind is going. I have a few BOSS/BEP units that I picked up at auction. They're far too bulky to live on the bench for everyday use as PSUs, but perhaps I should change my thinking and make room! I believe a 4-quadrant power supply should be able to survive exactly this condition, because it's a normal part of the operating envelope.

The whole rest of the thread is talking about fuses and diodes, and since these are so simple and evidently common for a few folks to use, it makes me wonder why they aren't a normal part of the PSU output structure. Of course a power supply could be simple and cheap and omit all these things, but we pay three or sometimes four figures for them because they're robust and versatile and include all sorts of protections and limits. And of course I don't make a habit of hooking things up backwards, but I don't want to flush three or sometimes four figures down the toilet every time I brainfart.

I had hoped for someone to say "well yeah you idiot, you cheaped out and bought a Siglent instead of a Keysight, of course those have reverse polarity protection", but such advice has not been forthcoming.

Dang.

well here it comes than. You asked for it ...

"well yeah you idiot, you complicate things and instead of buying charger you want to use things that are not meant to be used for this purpose...""
Sorry.. I couldn't resist..

And again, no, Keysight PSU costing 5-6 times more would happily blow up all the same. They are not more robust than that Siglent in that regard, same as most of lab PSU out there. They all say it in the manual too. No backdriving current please...

Unless it is 4 quadrant PSU or a device specifically made to charge batteries it won't survive pushing twice the rated voltage into it's outputs and reversed power source
 connection... Not to mention battery can give serious current without any limit...

[tautech]

Maximum compensation voltage is 1V.

Good point.  Most such setups limit the sensing differential so that there isn't a runaway if the remote leads come loose.  So if the voltage is critical, a large power Schottky would be in order.  Or a MOSFET ideal diode solution.

Point, no actually it's a datasheet spec !

The only point in this thread is the careless mistake the OP made with his PSU. 
Tough lessons they are too but I have a mainboard of an X-E that I can take pics of and maybe partly RE the output stage to help with a repair.

[bc888]

Alright, I let the smoke out of my bench PSU this morning. Siglent SPD3303X-E. I was using it to charge a 55AH LiFePO4 cell, and after disconnecting it to shuffle some things around on the desk, I hooked it back up but got the polarity backwards. Turns out the current limit limits what the supply will source, but not what it will allow to flow if you drive it in the wrong quadrant. Sounded like popcorn... Nothing actually burst into flame though, so the fire extinguisher merely got grabbed, not discharged.

Anyway, I'm looking for replacement recommendations. I've actually been super happy with the supply prior to this, and I'd be fine replacing it with another of the same, and I'll just try to be more careful. But I'm curious, since I've evidently become the better idiot we hear so much about, are there nicer units with more idiotproofing?

Thanks for the story. I'm rethinking my doing this as well. Not a LiFePO4 for me but a 12V. Been charging it with an old Tektronix CPS250 that I got cheapo when a college retired a bunch of them from their lab. I looked one up recently and much like a fine vintage wine, the price has increased, I wouldn't replace it at today's prices. Dave has a good video I watched some time back on charging Lithium Ion Cells with a PSU. I wouldn't attempt doing it with my old hand crank model given the vagaries, charging termination and complexities Dave discussed. Good vid though.

Curious what you are thinking of replacing it with so I'll stay tuned, sorry but I lack advice in that area. 

[Myself]

The feature might be called robustness with reverse polarity protection.  You are more likely to find it in older design linear PSUs with much less "bang for buck".  The BK Precision 1652 I have is one example of a supply that is protected against this sort of thing--but it is only a 24V 0.5A per channel (1A total) output.  I have some others that I suspect might be OK, but I'm not willing to experiment. 

Whoah!

Somehow I missed this gem in my first skim through the thread. Yeah, wow, a lot of cheaper linear units do indeed have a reverse polarity spec! It seems entirely absent in all the "nicer" newer units I've looked at, including from all the high-end brands. How utterly bizarre.

Curious what you are thinking of replacing it with so I'll stay tuned, sorry but I lack advice in that area. 

It turns out that I've never actually used the remote-control features of the fancypants digital unit anyway. If I have to give those up to get the common-sense and obviously-important-in-my-use reverse-polarity protection, I think I'll be doing just that.

It looks like I can add whole shelf of Triplett PS303/PS305 or similar, for considerably less than another Siglent. They look like a downgrade in every way, except the whole not-exploding thing.

I am really gonna miss the ability to see/adjust the current limit setting with the load connected. It looks like that feature is mutually exclusive with reverse polarity protection. Every single unit I've found specifies the "unhook the load, short the terminals, twiddle the current knob, remove the short, hook the load back up" dance. Oy. I could probably make an interposer board with a big ol' SPDT switch, just gotta be ultra sure it's break-before-make...

A fuse might help in some cases depending how the failure occurs--if it is just thermal damage due to high currents, a fuse is your savior.  In other cases it won't help. I personally would use a 7.5A ATO fuse right at the battery in your application no matter what PSU you have.  You can always have other issues, like accidentally shorted leads, that would be a pretty serious issue with a battery that size.

This is a really good point, and upon further consideration, there's no reason that any test lead should ever be unfused, except I figure that would make them mechanically snag on things. Figure someone could make a banana plug with an integrated fuseholder that would probably solve that, but has anyone ever done so? Doesn't look that way.

Your burnt PSU may be repairable--crack it open and lets all have a look.  You may find that some parts near the front end have sacrificed themselves and the bulk of it may be OK.  And you can modify and hack it now that the warranty has....expired?   

It's outside now airing out; once the stink is down to a manageable level I'll be doing just that.

Tough lessons they are too but I have a mainboard of an X-E that I can take pics of and maybe partly RE the output stage to help with a repair.

The front panel of the SPD3303X(-E) always did bug me, the jacks not being 19.05mm on-center like all my paired Pomonas, and I'd been considering a replacement jack-field anyway just to fix that. Also I'd like to add switches that tie all the negatives to a common bus, which is a frequent operation on my bench (and I wish there was a button and internal relays for that just like ser/para operation). And I might as well add some fuses, ideal diodes, perhaps it can be modded for remote sense.... why am I not just buying a whole new one? Oh right, because none of the new ones have those things either.

Unless it is 4 quadrant PSU or a device specifically made to charge batteries it won't survive pushing twice the rated voltage into it's outputs and reversed power source connection... Not to mention battery can give serious current without any limit...

Just for clarity's sake, it wasn't twice the rated voltage. It was 3.65 volts, from a single 55AH cell. The PSU is rated to 32 volts, just only the other way around...

[2N3055]

It will depend on topology of PSU. In the olden days (and many cheap units are simply copying some old designs), manufacturers did try out many different topologies and variations. They all had some pros/cons. Nowadays it seems that most use similar topologies that can have digital control and have large output capacitors to make it unconditionally stable.They also have low drop main regulator with some kind of preregulator.
There might be switching LAB supplies that could be more robust... And would be inexpensive and good enough for battery charging.

Also, as you said, you connected only one cell. But it was enough to forward bias protection diodes...

To be honest, simplest is to just put in a diode. Measure voltage drop on a diode when current is LOW, some 10s of mA. And compensate (raise) output voltage for that condition. When battery is empty, you need to limit current and voltage is not important, and then as the current drops with the state of charge voltage will slowly converge to desired max voltage.. There is some tempco on a diode but nothing really important for this case. If in doubt, set target charging voltage to few percent less than max for the battery (which is good for battery lifetime anyways).
I did that many times in a pinch for an odd job here or there.

For batteries I charge on regular base, I have propper chargers..