Noisy USB car charger

[rexxar]

I went and picked up one of those little 12V switchmode USB chargers today. To my surprise, when I plugged my phone in, I lost all but the local radio stations. 

I stuck it on the scope to see if it was putting some sort of HF crap back up the 12v line, and sure enough, there's a 1.344MHz signal coming out. When I turn the load up to 1A (what it's rated for) the noise shoots up to 500mV p-p! Opening it up, there's not even so much as a bypass cap on the input

What gets me is, this was an Energizer brand charger. I thought I'd actually get my money's worth from a big brand, but the internals of this look exactly like the $5 charger that melted on my bench.

Anyway, I can't take it back to the store now, but I thought it'd be a good opportunity to learn about filtering. Putting 100uf or so on the input cuts the noise in half, but what else can I do?

Surely this violates some sort of FCC regulation? I'm sure a device wiping out the FM band would make them a bit unhappy...

[han]

Are you sure the spelling correct?
not chinglish Energiser?

[pa2ees]

I think a 100uF cap is probably too big to filter the 1MHz well.  I would put a smaller cap on it (1 or 2 uF), or maybe even a few of them in parallel.  Also combine that with a ferrite bead on the charger side of the caps, and you'll have yourself an LC filter.  Make the knee of the filter at a low frequency, maybe a few KHz (or whatever your size constraint of the ferrite permits) and you should see that 1MHz noise go away.

My reasoning for the smaller caps: they tend to have lower impedance at higher frequencies, thus they do a better job filtering the frequencies you're looking at.  putting several of them in parallel will lower the impedance even more.

Hope that helps.

 - Erik

[VK5RC]

Agree with above,  consider a ferrite on both input and output,  often both a 1uF amd 100pF on worst offender (input or output)

[planet12]

Here's my bet.

The charger is based on a buck converter, something like a mc34063 or maybe something a bit better given the 1A rating.

The >1Mhz coming out of it is probably not the fundamental switching frequency, but instead the ringing from the inductor and the parasitic capacitance of the bucking diode emanating from the main switching node.

As such, although you'll get improvements from adding filters, the most effective option may well be to add an appropriate RC snubber between the switching node and ground, to critically damp the ringing.

This http://www.maximintegrated.com/app-notes/index.mvp/id/3835 is a good intro to figuring out what values to use (aimed at a push-pull converter, but the general ideas are the same).

[planet12]

How about you reverse engineer the schematic, and take photos of the PCB... you know, for our entertainment 

[Marco]

The >1Mhz coming out of it is probably not the fundamental switching frequency, but instead the ringing from the inductor and the parasitic capacitance of the bucking diode emanating from the main switching node.

A standard MC34063 buck converter copied from the datasheet shouldn't ring like as badly as he says though should it?

Maybe they didn't use a Schottky diode?

[planet12]

The >1Mhz coming out of it is probably not the fundamental switching frequency, but instead the ringing from the inductor and the parasitic capacitance of the bucking diode emanating from the main switching node.

A standard MC34063 buck converter copied from the datasheet shouldn't ring like as badly as he says though should it?

Maybe they didn't use a Schottky diode?

I wouldn't expect it to be that bad either, but without seeing the circuit and PCB who knows what's been done and what corners have been cut - maybe there's really bad layout, wrong components, etc. It's just the frequency is much higher than I'd expect the switching frequency to be, and it's in the range ringing is often found with the L+Cparasitic seen in these sorts of dongles.

[rexxar]

Okay, I've attempted to draw a schematic from this thing.

It is indeed a switching supply, and the switching frequency does appear to be that 1.3MHz. Or, at least I've not found any different frequencies probing around the board.

The main IC is only marked FRW, and that hasn't turned up any search results.

The OEM of this thing appears to be a company called PDP. I sent them an email, and they said that all their chargers are tested for FCC compliance, so mine must be defective... They want me to mail it back to them. 

[SeanB]

Means the one they sent in was the one with the filter caps and ferrites in it. this was then optimised out by the OEM to reduce the BOM cost after the first batch. Now it may not even be close to the original aside from the case.

[amyk]

It's probably an ADP2301, SP1076, AMS1076, MP2259, or any one of many other integrated DC-DC bucks with the same pinout and around the same frequency. Probably a dozen or more Chinese manufacturers make the same thing.