Rail-splitting a wall wart

I also thought he didn't deal with musicians and didn't even like music!

Here's the power supply section of a Korg MS2000 synthesizer, I worked on one one time that somebody had plugged an AC adapter of unknown type into, as I recall the list of f***ed components was long, some bits like the SMT input filter and diode were rattling around in the enclosure, F1 was done, IC15 was done, FU1 was open, fortunately no damage to the CPU or DSP though.

Oh, I swap power supplies freely, but there's the small matter of reverse-engineering the connections beforehand. There's some amusing moments: lots of 13V zeners burn up fuses protecting CD/HD external drives. The +12 for those drives tolerates lots of overvoltage, the zeners and fuses were overkill, and axed perfectly functional (but loose specification) power.

My Samsung television from the thrift store works fine with the right Dell power brick. There's likely never to be a Samsung power brick next to such an item...

** Long time ago, I was sent a Korg analog synth for repair that was the *victim* of a main power accident. A faulty ( home made ) extension lead had exploded a 12 inch length of ground track on the PCB along with a cup full of ICs, tantalum caps, diodes etc. Took quite some time, but I eventually fixed it. I remember the young owner was grumpy about the fee.

...... Phil

A P-channel FET or a series Schottky diode fixes that. I'd never omit polarity protection from any design--even my hand-wired protos have shunt 1N5819s or 1N5823s.

I don't use shunt diodes in products or customer designs--in the low-noise instruments world, it's far from unknown for people to power

24V stuff from two boat batteries in series.

Interesting point, though--it might be a good idea to put an NFET in the negative lead as well as a PFET in the positive one, to protect against shorts from V- to ground. It could be a depletion device to guarantee startup behavior. I'll think about that.

Cheers

Phil Hobbs

On 2023-02-06 10:25, Phil Hobbs wrote:

Okay, here's one possibility. It protects against reverse polarity warts as well as shorts to ground from either side of the wart.

The short-circuit protection needs 5 extra parts--four FETs and one quad pack resistor.

(Reposting with the .asc file inline for those whose news servers don't support text attachments.)

Cheers

Phil Hobbs

Version 4 SHEET 1 880 680 WIRE -1584 112 -1616 112 WIRE -1504 112 -1584 112 WIRE -1312 112 -1408 112 WIRE -1136 112 -1312 112 WIRE -992 112 -1136 112 WIRE -896 112 -992 112 WIRE -768 112 -800 112 WIRE -416 112 -432 112 WIRE -1312 128 -1312 112 WIRE -1136 128 -1136 112 WIRE -992 128 -992 112 WIRE -1616 144 -1616 112 WIRE -1184 144 -1216 144 WIRE -432 144 -432 112 WIRE -608 192 -624 192 WIRE -608 208 -608 192 WIRE -192 224 -192 208 WIRE -992 240 -992 208 WIRE -992 240 -1184 240 WIRE -880 240 -880 160 WIRE -880 240 -992 240 WIRE -384 256 -480 256 WIRE -1616 272 -1616 224 WIRE -1568 272 -1616 272 WIRE -1424 272 -1424 160 WIRE -1424 272 -1488 272 WIRE -1360 272 -1424 272 WIRE -1312 272 -1312 208 WIRE -1312 272 -1360 272 WIRE -1264 272 -1312 272 WIRE -1216 272 -1216 144 WIRE -1216 272 -1264 272 WIRE -432 272 -432 224 WIRE -1360 288 -1360 272 WIRE -480 288 -480 256 WIRE -464 288 -480 288 WIRE -1312 304 -1312 272 WIRE -1136 304 -1136 224 WIRE -992 304 -1136 304 WIRE -880 304 -992 304 WIRE -384 304 -384 256 WIRE -384 304 -400 304 WIRE -368 304 -384 304 WIRE -1184 320 -1184 240 WIRE -608 320 -608 288 WIRE -544 320 -608 320 WIRE -464 320 -544 320 WIRE -368 320 -368 304 WIRE -192 320 -192 304 WIRE -992 336 -992 304 WIRE -608 336 -608 320 WIRE -432 352 -432 336 WIRE -1424 384 -1424 272 WIRE -880 384 -880 304 WIRE -1264 400 -1264 272 WIRE -1232 400 -1264 400 WIRE -1616 432 -1616 272 WIRE -1552 432 -1616 432 WIRE -1504 432 -1552 432 WIRE -1312 432 -1312 384 WIRE -1312 432 -1408 432 WIRE -1184 432 -1184 416 WIRE -1184 432 -1312 432 WIRE -992 432 -992 416 WIRE -992 432 -1184 432 WIRE -896 432 -992 432 WIRE -768 432 -800 432 WIRE -608 432 -608 416 WIRE -592 432 -608 432 WIRE -432 448 -432 432 WIRE -416 448 -432 448 FLAG -1584 112 +wart FLAG -1552 432 wartRET FLAG -368 320 0 FLAG -592 432 loadRet FLAG -624 192 +Load FLAG -1360 288 0 FLAG -192 208 +Load FLAG -192 320 0 FLAG -768 112 +Load FLAG -768 432 LoadRet FLAG -544 320 div FLAG -416 112 +Load FLAG -416 448 LoadRet SYMBOL voltage -1616 128 M0 SYMATTR InstName V2 SYMATTR Value 24 SYMBOL res -624 192 R0 SYMATTR InstName R1 SYMATTR Value 10k SYMBOL res -624 320 R0 SYMATTR InstName R2 SYMATTR Value 6k SYMBOL Opamps\\UniversalOpamp2 -432 304 R0 WINDOW 0 74 -39 Left 2 WINDOW 39 28 51 Left 2 SYMATTR InstName U1 SYMATTR SpiceLine Ilimit=1.5A SYMATTR Value2 Avol=1Meg GBW=1.5Meg Slew=1Meg SYMATTR SpiceLine2 Rail=0 Vos=0 En=0 Enk=0 In=0 Ink=0 Rin=500Meg SYMBOL res -176 208 M0 SYMATTR InstName R10 SYMATTR Value 20 SYMBOL pmos -800 160 M270 WINDOW 0 16 -37 VRight 2 WINDOW 3 -28 -28 VRight 2 SYMATTR InstName q1 SYMATTR Value AO6407 SYMBOL nmos -800 384 R90 WINDOW 0 -10 133 VLeft 2 WINDOW 3 -23 57 VLeft 2 SYMATTR InstName Q4 SYMATTR Value AP9465GEM SYMBOL res -1472 256 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R7 SYMATTR Value {Rshort} SYMBOL res -416 128 M0 SYMATTR InstName R5 SYMATTR Value 1m SYMBOL res -416 336 M0 SYMATTR InstName R6 SYMATTR Value 1m SYMBOL res -1008 112 R0 SYMATTR InstName R4.3 SYMATTR Value 10k SYMBOL res -976 320 M0 SYMATTR InstName R4.4 SYMATTR Value 10k SYMBOL nmos -1232 320 R0 SYMATTR InstName M1 SYMATTR Value 2N7002 SYMBOL pmos -1184 224 M180 SYMATTR InstName M2 SYMATTR Value BSS84 SYMBOL res -1328 112 R0 SYMATTR InstName R4.1 SYMATTR Value 10k SYMBOL res -1296 288 M0 WINDOW 3 30 109 Left 2 SYMATTR Value 10k SYMATTR InstName R4.2 SYMBOL pmos -1504 160 R270 WINDOW 0 16 -37 VRight 2 WINDOW 3 -28 -28 VRight 2 SYMATTR InstName q2 SYMATTR Value AO6407 SYMBOL nmos -1504 384 M90 WINDOW 0 -17 54 VLeft 2 WINDOW 3 70 124 VLeft 2 SYMATTR InstName Q3 SYMATTR Value AP9465GEM TEXT -136 376 Left 2 !.tran 1m TEXT -416 208 Left 2 ;Behavioral\nTCA0372 TEXT -136 400 Left 2 !.param Rshort 1m\n.step param Rshort list 1meg 100m TEXT -56 88 Left 2 ;Wall Wart Polarity and\nShort Circuit Protection\n \nP. Hobbs, 2023-02-07

On 2023-02-07 04:41, Phil Hobbs wrote: <snip>

I should add that there's nothing special about the choice of FETs--they're just what came up in the LTspice library dialog. Obviously things like ESD protection capacitors are missing as well--probably 1 uF ceramic from each side of the wart to ground.

Cheers

Phil Hobbs

I generally use a polyfuse and a big unipolar TVS first thing. That handles reverse voltage and some (not all) over-voltage cases. But I'd rather have the TVS fail shorted, than blow up everything downstream.

TVS shorted can be blamed on the customer!

Bulletproof startup can be challenging.

We hard-ground everything in our boxes and assume that a customer might want to power us from their, usually grounded, roughly 24 volt DC buss.

That's kinda complex for my taste. My zeners do reverse polarity protection, with a polyfuse or such to limit current. Most warts current limit pretty well, cw or burp mode.

I'd love a fast 2-terminal current limiter device with thermal shutdown. We've tried some ic e-fuse parts which liked to blow up.

TCA0372 is supposed to thermal limit but it doesn't always.

<snip>

My threat model includes:

1. Reverse polarity wart
2. Either side of the wart shorted to ground, normal otherwise
3. Either side shorted to ground, reversed.

I have customers who like to use huge boat/RV batteries to power low-noise test setups, and I have occasionally been guilty of siamesing one wall wart into two boxes, so I assume others are too. The NFET/PFET thing turns off both sides, so no damage or weird behavior should occur.

An extra five tiny parts to do all that isn't bad, I don't think. (All the FETs can be SC-70s, probably.)

Well, with way over an amp worth of output and a small thermal mass, I can easily believe you might melt an output device before the thermal limiter knew about it. Unlikely to be an issue in this case because of the series-terminated outputs.

Cheers

Phil Hobbs

We often do that too, and used to do it more when we could still get PolyZens, which were a wonderfully complete and compact solution (*sniff* *sniff*).

A big TVS can desolder itself and fall off the board if the PCB is vertical.

Most of our stuff has switchers and lots of filtering, and so runs fine off a common power buss. The rail splitter thing is for light-duty use.

Cheers

Phil Hobbs

I don't usually repair things professionally (probably a good idea), so this one I ended up keeping for myself once it was running again. I bought it off a friend's buddy for $50 in unknown condition other than no power-up.

Sounds like you had to rebuild most of the internal power supply-bits, in the case of this later all-digital Korg product in the interest of expedience I replaced what I could get easily enough from Mouser and then just carefully bypassed the blown 3.3 volt buck, and bodged in one of these:

with the shutdown pin connected appropriately and it works OK.

I expect some fashion of hypothetical client out there would be unhappy with that as it would then not be "original" (maybe the same type that would be unhappy if I'd charged what it cost to ensure it was.)

But I didn't really feel like messing with 3rd party vendors selling discontinued buck controller ICs of unknown provenance and hope for the best...

Sounds like an application for an SPLD with analog comparators and state machine, e.g.

They're about 50 cents in quantity programmed. 8 states seems like enough to build a little fault tree that could quickly determine the most common disallowed states (over or under-voltage, reverse polarity, AC plugged in instead of DC) and respond appropriately

bitrex wrote: ==========

** IIRC the damage was mostly in the audio output and frequency generation sections. The Korg synth was correctly earthed, the tube amplifier it fed was using the bad AC lead.

High 50Hz current flowed to mains safety ground via the Korg's audio output jack and earthed PCB traces.

.... Phil

Where would it get its power from?

Take a look at IXYS CPC1540GS.

It's a high-voltage bidirectional SSR with current and thermal limiting.

I've used it as just a current limiter. This and a TVS might be a nice combination. Drive the LED side with a resistor or a depletion fet.

Some clever LED-side driver might do under/overvoltage lockout too.

I was thinking something like a 1N00x bridge followed by resistor + zener and float it, connected ahead of a normally-open DPDT relay that's in line with the rest of the circuit.

That plus a dual optocoupler and few other parts seems like it should be enough to detect reverse polarity and AC, and then close the relay once that checks out and over/under voltage comparisons also good

That could work if all the details were right and it wasn't too complex. The uP power need not be floating if reverse polarity were handled right, like a series diode or equivalent... pfet maybe.

It could be a product, a universal over/under/reverse/overcurrent protector. It may as well do the powerup reset function, which is hard to do right.

I'll do the circuit design if you'll program it. You could sell them. One could sell the cpu alone or a little pc board with everything. I might buy some.

That could be a new market, power inlet boards that solve all the problems. One version could include USB-C connector and negotiation. A version might make +3.3 or +-10 or something like that.

It's shocking that nothing like this seems to exist.

Nice part, but its maximum Ron of 25 ohms is on the high side.

Thanks

Phil Hobbs