low noise supply

I'm not sure about common mode noise. But I used an LC and then cap multiplier. What sorts of voltage/ current do you need?

George H.

That's more or less the old-time modem trick--use a 24VAC wall wart, and do everything else inside.

I've had very good luck with just using fancyish capacitance multipliers with CM chokes, beads, and bypass caps on the input connector.

What are you seeing that's so awful?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

If you're trying to use those unfiltered(!) charger warts, there can be very little recourse.

CM noise needs to be shunted across the source. This is why Y rated caps exist.

Without access to the noise source (i.e., mains and output sides of the charger), all you can do is use a massive CM choke and pray. You'll likely still end up with low frequency problems (under a few MHz, say?), because the CMC has impedance, and with nothing to work against (just the ever-increasing reactance of self-capacitance), it simply doesn't matter. You don't have a low impedance to filter against, and even if you do (say your device is connected to other, grounded equipment), you're still left with the antenna that is the charger's cord itself, and the current loop between all three.

I think there are only two ways such chargers are released: 1. regardless ("Can't Enforce" mark); 2. by testing with no load, or with a fully charged device (same thing).

The fact that the cord is short and typically has a floating load will help, but not enough to pass >10MHz.

It's worth noting that, if you use a standard adapter to power a device, you are allowed to use any device, including a bench supply, for testing. At least for the relevant IEC 61000-4 parts. If you specify a particular choice, you can opt to use that, too.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

Something like this:

The Signal split-bobbin are cheap with about the best CM isolation you can get without going specialty.

Dunno if this is the best deal in town:

In one case, we are transmitting a clock over a fiber link and doing a phase noise spectrum at the receive end. The noise floor is about -120 dBc/Hz (which we plan to improve) but it has a couple of nasty noise spikes in the low hundreds of KHz range. One is probably from the wart and one from the on-board charge pump in the receiver that makes our negative voltage.

A bigger box would help, so that we could add shield cans around the power supply and the photodiode/TIA. But the wart will also jam ground loop noise into our output coax and the rest of the world.

So, back to 60 Hz supplies!

I guess we could just buy separate 120 volt and 240 volt transformer type warts and ship the appropriate one. Half-wave rectify to get + and -, or do one of those charge-pump-from-the-bridge horrible kluges to get some negative voltage, or add a small transformer to make a little negative power.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I've had pretty good luck with the 150-kHz Simple Switchers driving toroids. With a cap multiplier, they really don't talk to the rest of the world much.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

On Tue, 28 Jun 2016 12:03:01 -0700, John Larkin Gave us:

If you want low noise, you figure out what your consumption level will be and you put a huge cap bank on the output of the dongle. The supply needs to be powerful enough to keep it filled, and that is where the problem usually lies.

An alternative is to use supercaps or small battery packs and refill those at specified intervals.

18650s are mil spec units.

If you need it to be always on continuous and no batteries in the solution, then you should go back to 60Hz fed linear as you previously stated switchers can present too much noise for sensitive HF equipment.

IOW what the world needs is a good 3-wire wall wart with a Y cap. ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Hmm well probably a silly idea, but a higher voltage wall wart, with a linear rail splitter for ground. Then the wall wart goes over there, and the sensitive electronics here and a warning to users to keep the two apart. (I've "rediscovered" the interference spikes from my 'scope ~135 kHz, and my wall wart ~150ish kHz a few times in testing... about 3' separation works for me. It's getting so that I know the "spectral finger print" of all my test gear... (and room lights))

George H.

24V sounds like a lot, my adsl modem has 9V AC supply, The modem probably has a doubler to get +/-12V

-Lasse

With 24VAC, you can double to circa 70V and run all those '48V' telco DC/DC converters. Or fullwave bridge, with a choke, for 24V and use mil/aviation '24V' components. Power sources at low V are a nuisance, IMHO; let a point-of-load converter do the job.

For noise critical situations, I revert to battery operation. The wallwart, used to charge the battery, is disconnected or turned off during noise-critical operation.

Failing that, then a wall wart that is known-quiet, 60Hz or otherwise.

Is it practical to produce a low-noise critical widget that is susceptible to every form of external interference that you've mentioned?

RL

I do a lot of low noise stuff too, and I find that an inverter made from a LM2594 buck and a toroid, followed by a two-stage cap multiplier, does a nice job. Following up your 50/60-Hz AC rectifier with one of those will also get rid of most of the ripple, which is a plus.

My new board is a 1-uA full scale photoreceiver that's shot noise limited above about 10 nA in a 1-MHz bandwidth, and it's built like that, except with a DC wall wart input.

At low frequencies and low currents, it's not as good as the series-connected-photodiode feedback method, but it should be better at high frequency due to a greatly improved bootstrap. However, we'll see when I have one to test. It should be stuffed today, so we'll see--I'll post some output spectra when I have them.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Then you get a 2.2nF capacitor to the mains, which is not necessarily a nice thing to have, depending on what crap is on the mains. Note that in many countries, neutral and live are interchangeable as the power plugs go in both ways.

It would be nice if someone made a wall wart with three internal screens between the primary and secondary windings, with the primary screen connected to the rectified mains, the secondary screen connected to the DC output negative, and the screen between those ones connected to mains earth, (and no Y-capacitor from the DC output to the mains).

The output cable should have three conductors: the isolated DC output positive and negative terminals, and a screen conductor connected to the mains earth pin, and thick enough to provide a good earth connection if the load needs it, (able to withstand 25 Amps for 10 seconds as applied by portable appliance testers operated by minimum-wage employees who tend to blow up test equipment in the sort of institutions that bother with that stuff). Note, very importantly, the negative pole of the DC output would NOT be earthed inside the power supply, (because otherwise by Ohm's law, if current is flowing then this would mean that the negative terminal could not be at earth at the load end of the cable).

For test equipment use, I think people would pay a good price for such a thing.

On Wed, 29 Jun 2016 10:16:07 -0400, Phil Hobbs wrote:

Maybe I'll make a really slow charge pump, with an opamp as the driver and some pretty big caps in the doubler part. A few KHz and really lame slew rates.

Or I could do this, except it needs some gigantic caps.

Version 4 SHEET 1 1492 680 WIRE -64 -32 -128 -32 WIRE -32 -32 -64 -32 WIRE -128 32 -128 -32 WIRE 16 32 -128 32 WIRE 80 32 16 32 WIRE 240 32 144 32 WIRE 400 32 240 32 WIRE 480 32 400 32 WIRE 544 32 480 32 WIRE 832 64 800 64 WIRE 880 64 832 64 WIRE 1024 64 944 64 WIRE 1104 64 1024 64 WIRE 1264 64 1168 64 WIRE 1344 64 1264 64 WIRE 1408 64 1344 64 WIRE 400 80 400 32 WIRE 544 80 544 32 WIRE 1264 112 1264 64 WIRE -128 128 -128 32 WIRE 1024 128 1024 64 WIRE 80 160 -32 160 WIRE 240 160 240 32 WIRE 240 160 144 160 WIRE 400 208 400 144 WIRE 544 208 544 160 WIRE 1024 240 1024 192 WIRE 1264 240 1264 176 WIRE -128 304 -128 208 WIRE -32 304 -32 160 WIRE -32 304 -128 304 WIRE 16 304 16 32 WIRE 80 304 16 304 WIRE 240 304 144 304 WIRE 304 304 240 304 WIRE 304 352 304 304 WIRE 832 384 800 384 WIRE 880 384 832 384 WIRE 1024 384 944 384 WIRE 1104 384 1024 384 WIRE 1408 384 1408 64 WIRE 1408 384 1168 384 WIRE -128 432 -128 304 WIRE 80 432 -128 432 WIRE 240 432 240 304 WIRE 240 432 144 432 WIRE 1024 448 1024 384 WIRE 1408 448 1408 384 WIRE -128 496 -128 432 WIRE -64 496 -128 496 WIRE -32 496 -64 496 WIRE 1024 560 1024 512 WIRE 1408 560 1408 528 FLAG 544 208 0 FLAG 304 352 0 FLAG 400 208 0 FLAG -64 -32 A FLAG -64 496 B FLAG 1024 240 0 FLAG 832 64 A FLAG 1264 240 0 FLAG 1024 560 0 FLAG 832 384 B FLAG 1408 560 0 FLAG 480 32 V+ FLAG 1344 64 V- SYMBOL diode 80 48 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName D4 SYMATTR Value 1SR154-400 SYMBOL res 528 64 R0 WINDOW 0 59 38 Left 2 WINDOW 3 56 74 Left 2 SYMATTR InstName R1 SYMATTR Value 100 SYMBOL voltage -128 112 R0 WINDOW 0 -131 33 Left 2 WINDOW 3 -196 77 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value SINE(0 10 60) SYMBOL diode 80 176 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName D1 SYMATTR Value 1SR154-400 SYMBOL diode 144 288 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName D2 SYMATTR Value 1SR154-400 SYMBOL diode 144 416 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName D3 SYMATTR Value 1SR154-400 SYMBOL cap 384 80 R0 WINDOW 0 57 20 Left 2 WINDOW 3 49 55 Left 2 SYMATTR InstName C1 SYMATTR Value 10m SYMBOL cap 944 48 R90 WINDOW 0 77 31 VBottom 2 WINDOW 3 85 28 VTop 2 SYMATTR InstName C2 SYMATTR Value 3.3m SYMBOL diode 1008 128 R0 WINDOW 0 59 26 Left 2 SYMATTR InstName D5 SYMATTR Value 1SR154-400 SYMBOL diode 1168 48 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName D6 SYMATTR Value 1SR154-400 SYMBOL cap 1248 112 R0 WINDOW 0 60 18 Left 2 WINDOW 3 51 56 Left 2 SYMATTR InstName C3 SYMATTR Value 3.3m SYMBOL cap 944 368 R90 WINDOW 0 75 30 VBottom 2 WINDOW 3 86 26 VTop 2 SYMATTR InstName C4 SYMATTR Value 3.3m SYMBOL diode 1008 448 R0 WINDOW 0 47 28 Left 2 SYMATTR InstName D7 SYMATTR Value 1SR154-400 SYMBOL diode 1168 368 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName D8 SYMATTR Value 1SR154-400 SYMBOL res 1392 432 R0 WINDOW 0 58 33 Left 2 WINDOW 3 54 64 Left 2 SYMATTR InstName R2 SYMATTR Value 300 TEXT 472 464 Left 2 !.tran 200m TEXT -312 232 Left 2 ;AC WART TEXT 160 232 Left 2 ;BRIDGE TEXT 1096 280 Left 2 ;BRIDGE TEXT 400 344 Left 2 ;Quiet Bipolar Power Supply TEXT 416 400 Left 2 ;J Larkin June 29, 2016

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

LEDs and solar cells!

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Magnetic fields reduce more rapidly with distance. Their effect can also be reduced by re-orientation of the victim's receiving loop.

Capacitive radiation is pernicious, in that the earth forms a return path for common mode currents, which will mode-convert at the slightest physical imbalance.

Apparently simple 'widgets' that perform their intended function flawlessly seldom reveal all the issues that determined their final configuration.

RL

It gets worse. if you want 120/240 you probably want 50Hz magnetics.

--
This email has not been checked by half-arsed antivirus software

It's like those capacitive drop supplies, what happens if you get a surge pulse? Kaboooom