EEVblog #765 - LED Panel Lighting Switching Noise

[EEVblog]

A follow-up to the previous video checking if there are any noise issues with the new lab LED panel lights.
And a noise comparison with the old LED lights.
LED Lighting Install Video:

Common Mode Noise Video:

[jcreedon]

Was the noisy plug-pack identical to the other non-noisy ones?

Maybe do a tear down of a good one and the bad one and compare the two and do a post-mortem?

[bktemp]

+1 for the teadown and further investigation. Please show us the indside of this device.

[Corporate666]

+1 for the teadown and further investigation. Please show us the indside of this device.

Yeah I was hoping for that too.

Being a Chinese POS, I wonder if it's just a fault of assembly (part not soldered correctly) or something more sinister, like parts entirely left our.  I'm very curious to see.

[SeanB]

More likely the caps inside were the low end of the curve on lifetime and simply failed faster than the rest. You probably would find all are leaking and popped, while the others are also losing the fight and just held on for longer.

[EEVblog]

You get what you pay for. No need for that kind of language though.

I'll use whatever language I like.

[nixfu]

+1 for the teadown and further investigation. Please show us the indside of this device.

+2.  Have Dave2 fish it out of the trash and lets find out why its different than the others.

[woox2k]

+1 for the teadown and further investigation. Please show us the indside of this device.

+2.  Have Dave2 fish it out of the trash and lets find out why its different than the others.

+3 Please make a follow-up video showing us the insides of that crappy power supply. This should be very educational of how not to build a power supply.

[darrylp]

Yep I'd say, let's see what is needed to fix the noise. I'd guess its poor caps, but would be educational I'd think.

Sent from my Nexus 7

[apis]

Mhm, a post mortem would be very interesting.

[mikerj]

You get what you pay for. No need for that kind of language though.

It was hardly terrible language, are you American by any chance?

BTW. is your sig supposed to be ironic or did you just get it wrong?

[David_AVD]

My guess would be the output capacitor on the PSU has gone low capacitance and/or high ESR.

[Someone]

You get what you pay for. No need for that kind of language though.

It was hardly terrible language, are you American by any chance?

BTW. is your sig supposed to be ironic or did you just get it wrong?

mojo-chan seems to live in the UK but has spent a lot of time in Japan, but then uses US terms and politics a fair bit. Educated in the US perhaps?

[ez24]

Since the PS ended up in a wok, will there be a cook off ?  I just love cook offs, but I suggest do it outside because of the fumes.  Wasn't it an Australian who ate a bicycle (I wonder if he is still alive)?  If so a PS should be no problem for an Aussie especially if it is cooked in a wok.  Go ahead Dave, eat it ha ha ha.

[pickle9000]

PS remedies

- bury it
- hit it with a big hammer
- drive over it with a compactor
- find it a nice place to rest (land fill)

To be honest I don't really find the supply all that interesting.

[EEVblog]

My guess would be the output capacitor on the PSU has gone low capacitance and/or high ESR.

Yup, 99% chance. That's why it's only a problem under load.

[steves]

You get what you pay for. No need for that kind of language though.

I'll use whatever language I like.

And I'll defend to the death your right to use it, even if you are a ******* ****.

[Fungus]

BTW. is your sig supposed to be ironic or did you just get it wrong?

No.

Yes you did.

You can argue all day over the definition of "kilo" (and nobody will change their mind).

"kibi", OTOH, is always 1024. The "bi" part of "kibi" was invented for that purpose.

[Monkeh]

BTW. is your sig supposed to be ironic or did you just get it wrong?

No.

Yes you did.

You can argue all day over the definition of "kilo" (and nobody will change their mind).

"kibi", OTOH, is always 1024. The "bi" part of "kibi" was invented for that purpose.

Please don't feed mojo-troll.

[JonnyBoats]

Dave, why did you use an oscilloscope rather than a spectrum analyzer to look for interference from the switching power supplies? When looking for unintentional radiation isn't it quite possible that it could be limited to specific frequencies at relatively high levels while still having relatively low levels of broadband radiation?

[f4eru]

+1 for the spectrm analyzer, or, at least, the FFT mode of the scope...

If it were easier a vast amount of stuff could be saved from landfill.

If it were easier, a lot of people would get shocked by primary caps, and would repair dodgily the broken secondary wires (and some people really solder badly, which would endanger the class II isolation...

Basically hit the side with a hammer, shatter the glue.

it's not glue. it's laser or ultrasonically welded. I prefer to "gently" pop it open by deforming one half with a bench vise. Does not look good after that, but it'S far less messy...

[Monkeh]

+1 for the spectrm analyzer, or, at least, the FFT mode of the scope...

If it were easier a vast amount of stuff could be saved from landfill.

If it were easier, a lot of people would get shocked by primary caps, and would repair dodgily the broken secondary wires (and some people really solder badly, which would endanger the class II isolation...

Basically hit the side with a hammer, shatter the glue.

it's not glue. it's laser or ultrasonically welded. I prefer to "gently" pop it open by deforming one half with a bench vise. Does not look good after that, but it'S far less messy...

Padding with leather or thin pieces of wood helps with the pretty problem. If you can avoid actually marring the surface, the shape is rarely affected.

[bitwelder]

Dave, why did you use an oscilloscope rather than a spectrum analyzer to look for interference from the switching power supplies?

It was the first instrument available that could still show the issue during a quick video shooting?

[electronic_eel]

+1 for a teardown of the bad power supply.

Must not be the ESR of the caps. Another possibility is that they forgot to install the Y-cap connecting primary and secondary.

[Monkeh]

Please don't feed mojo-troll.

Please don't troll/stalk me.

I am doing neither, but it's nice you think you're special.

[DutchGert]

+1 for the teadown and further investigation. Please show us the indside of this device.

+2.  Have Dave2 fish it out of the trash and lets find out why its different than the others.

+3 Please make a follow-up video showing us the insides of that crappy power supply. This should be very educational of how not to build a power supply.

Yes pleasw !

[LordNobady]

+1 for the teadown and further investigation. Please show us the indside of this device.

+2.  Have Dave2 fish it out of the trash and lets find out why its different than the others.

+3 Please make a follow-up video showing us the insides of that crappy power supply. This should be very educational of how not to build a power supply.

Yes pleasw !

an other cap replacement video? no thanks.
Or make it something special.

[Fungus]

+1 for the teadown and further investigation. Please show us the indside of this device.

+2.  Have Dave2 fish it out of the trash and lets find out why its different than the others.

+3 Please make a follow-up video showing us the insides of that crappy power supply. This should be very educational of how not to build a power supply.

Yes pleasw !

The internal design might actually be OK. It could just be the bad soldering, cheap-ass caps and general penny-pinching manufacturing that's letting it down.

That would only be a 2-minute video though.

And if it's potted...? Forget it. Simply not worth the effort. There's a thousand examples of that sort of manufacturing out there.

[station240]

At best that duff PSU would be a 2 minute tear down, 10 seconds of smashing it with a hammer, 1:50 of talking about all the things wrong with the PCB extracted.

[EEVblog]

PS. If it is potted then maybe it isn't worth it.

Almost certainly not potted. Just needs a messy dremel to open the welded seems. maybe can be cracked open by force, but I've never had much luck with that.

[EEVblog]

Dave, why did you use an oscilloscope rather than a spectrum analyzer to look for interference from the switching power supplies?

It was the first instrument available that could still show the issue during a quick video shooting?

a) Everyone has a scope. Not everyone has a spectrum analyser. They can't play along at home.
b) It's most common to see noise like this when measuring stuff on a scope, so it's much more relevant and meaningful to people's everyday measurements, and I think easier for people to relate and understand.
c) I've done a previous video explaining this that used a scope, so it ties in.
d) Noise measurements (particularly PSU rail noise) are done with scopes with a 20MHz bandwidth limit.

I shot some extra spectrum analyser stuff today with the remotes.

[EEVblog]

Have you ever tried to open one?

Yes, many times.

It's pretty difficult, but there are videos of people doing it without completely wrecking the enclosure in the process. Basically hit the side with a hammer, shatter the glue.

IME they are usually ultrasonically welded, not glued.

[Corporate666]

an other cap replacement video? no thanks.
Or make it something special.

I was thinking a 2 minute teardown.  It's not an especially interesting circuit or device, but what is interesting is why one is so noisy and the others aren't.  IIRC, wasn't Dave having noise issues from day one with those lights? 

I am mostly curious to see whether the unit had a failure and they just shipped it anyway (no testing), or whether the right parts were even installed in the first place.  There is a video out there of a guy tearing down Li-Ion cells and finding some of them filled with salt/baking powder.  It's very interesting to see the depths some of these no-name Chinese manufacturers will sink to.  There could be a goldmine of hilarity inside that power supply.

[Monkeh]

PS. If it is potted then maybe it isn't worth it.

Almost certainly not potted. Just needs a messy dremel to open the welded seems. maybe can be cracked open by force, but I've never had much luck with that.

Vice works a treat on traditionally shaped ones.

[f4eru]

A dumb idea : you can test ther ESR of the output caps by putting the ESR meter on the output wires, without opening the darn thing
Could be a standard test for PSUs

[David_AVD]

Vice works a treat on traditionally shaped ones.

+1 on that.  You can hear the crack as the two halves let go.  Sometimes you need to reposition the case in the vice a couple of times.

If you don't want to reassemble the unit you can be less careful and loose some plastic in the process.

[Kjelt]

I always use an iron saw on all sides, if it doesn't matter that you saw a bit through some elco's since you're gonna throw it away anyhow it is much easier