Hi This is a new Raspberry Pi Model 3 B+ and it's got the new power over Ethernet capability. That's what this little pin header is here. so you can get the new Tada Raspberry Pi official Raspberry Pi Power over Ethernet hat which plugs into the top of that and plugs into that jumper down there and that takes the power over ethernet. They can plug into the main board, which is quite smart.

It actually taps the primary side power pins off, takes them to those pins up there so that you can put it through a DC to DC step-down converter and generate the 5 volts in here. So 48 volts comes in here DC to DC converter 5 volts out and then it goes out and powers your Raspberry Pi Now quite a few people have emailed me about this and said hey, Dave there's a problem with this new official Raspberry Pi Power over Ethernet hat. It powers the Raspberry Pi just fine, but when you actually plug anything into the USB ports, he can't provide any power and it shuts down the USB ports. So I Thought we'd have a little investigation see what's going on now.

A few people have complained about the surface mount pair. Oh, the ethernet connector here. And when they remove the board because it's only got the four little pads there and they are absolutely tiny, they've had this connector rip off the board. Yeah, just be careful when you remove this thing.

They probably there's no reason to use surface mount connector like that with the holes in back. There's no reason they should have just used the through-hole but even this one, when you plug it on, it's got the holes in the top, but you can't really. Then go plug another hat on top of that because there's really hardly any pin space left. and it comes with these standoffs.

And anyway, yeah, just be careful when you're removing this thing. Now the designer. This looks ok. a little transformer in the cutout there to get a lower profile that's in a pretty standard fare.

They've got an isolation slot down here, but curiously okay. They've you know, enough gap down there. but look at the gap up there. the clearance gap under this surface-mount cap.

here. it's just enough. Also, somebody forgot to peel that back there, but we've just got a 1206 cap here between the grounds on either side of the isolation transformer there. That's for a noise reduction, but apart from that, you know it looks pretty good.

The transform looks good enough. We've got the controller here. we'll have to have a look at that. We've got an eighty Tiny micro here, so I'm not entirely sure at the eighty Tiny Micro is doing anyone.

I Don't know I Really haven't looked into the details of this, but it's got a little last soon on brand fan here. There must be a temperature sensor somewhere and it just comes on. Maybe that's all the Eighty Tiny is doing is doing the fan controller I Don't know. So likewise with the clearance or lack thereof up here, they've done okay there.

But up here. look, they got that really close as well. This is the secondary ground over here and this is the primary side. So what they need to peel that back as well? You know? maybe they should have put a little slot in there? perhaps.

So yeah, it's not the world's best isolated design, that's for sure. And of course, the isolation would have to do with around here and how the traces are actually routed. They've pulled back the flood fill around there, but all the internal traces. You know, who knows what the spacing is because it's got to get all the way from here, all the way over to a primary side of the transformer over here to there.

So how they're routing that out here, somewhere with clearance and all that sort of stuff I Don't know. and this is the controller. It's a monolithic power Mp8 double-oh-seven and it's a primary side switch. It's just got a quite a very nice actually ultra low drop diode here, and these are the output capacitors.

They're out. They look like they're in series, but they're actually in parallel there. and it's the 5 volts directly out. So there is actually no feedback on this thing.

To do. secondary side regulation. it's actually done on the primary side. It requires a careful design of the transformer and diode and load selection and stuff like that to ensure that the output is regulated.

So for feedback, it actually uses a secondary primary side coil here to actually sense the voltage on the core. So you know you've got to have a careful designer your transformer here. Comparing it from a 48 volt, 0.35 amp power over Ethernet adapter here and it works just fine. Trust me.

I've had a look at we're getting 5.07 volts on the 5 volt rail there. We can just confirm that I have confirmed that this site USB adapter is hunky-dory There we go now. My particular board works just fine with the keyboard and mouse and this which draws basically you know very little, but we can plug in an electronic load here and it's drawing. You know, 15 milliamps of its own accord.

Set this down here. 40 milliamps like that I can switch our load on and it will draw that just fine. There it is. plus plus there.

Well there's a little bit of wiggle room there, but if we go up on that, boom it just. it switches off. So anything over about roughly about you know 40 to 50 milliamps something like that. Sometimes you can just leave it there for a bit and it will actually switch off.

and if of course if I you know said up to like 190 millions for example and then just press the button to switch it it switches off the USB ports and of course they they recover just fine. You haven't damaged anything, so we want to actually check for switching noise on this because this is a switching step-down a regulator. So I'm gonna probe properly I'm going to get rid of the crap you know, the antenna earth lead on that. I'm going to use the proper low inductive probe in down here and let's have a squiz.

We've got our switching frequency on there. It's about twenty four point six kilohertz or something like that. It's actually quite high 320 milli volts peak-to-peak so we don't have a huge load on there. just the just a Raspberry Pi itself.

Although that is a fairly large low, but it's not doing anything. it's just like booted up and sitting there doing nothing. But you can see you'd expect to see some switching frequency there, of course. I don't see any higher frequency stuff in here.

There's no ringing. it's fairly clean, if somewhat relatively higher 320 millivolts, but that's kind of what you expect really. I Mean it doesn't have much output, capacitance, and filtering on this thing. Okay, so what I'm going to do now is apply the load 200 milliamps here.

I'm gonna set my trigger point just below now. Ripple there. and here we go. We'll switch it on and see if our 5ol trail dips.

Nope, not a sausage so it's got nothing. Do off the Ray or dip in. Okay, let's just try that again with some AC coupling 200 millivolts, we're getting a bit more accurate now. So there's our switching frequency like that.

It's fair betta jitter on there, but that's what you'd expect and let's just try and do that again. She'll just take it down a bit just in case it does something, and I'll apply the load again, see if there's any dynamic change in that. Yeah, there was. Look at that.

It's doing something strange now. Is it like some pulse skipping mode or something? Oh yeah. look, there's some higher spikes up there that we didn't see last time. That's interesting.

So if we single-shot capture there, we go look at that. Oh yeah, look, let's change frequency from there to there. Some sort of mode change in the controller chip which causes that to change Now of course, if we unplug that and replug it and I won't change anything. Oh no.

There we go. No, it's the occasional. We heard the occasional glitch there. note there it is.

Nope, it's still doing it. so that's got nothing to do with it being shut down. It just changes frequency a bit there. If you're curious to know.

there you go: Twenty-three point five, and thirty two point eight five. Anyway, the thing that really would be a problem was is any dips in there. And trust me, if I set that single shot on there I cannot get it to do anything. I Tried it, tried mucking around that quite a bit and I just cannot get it to dip or anything like that.

so it's it's not a problem there, there's something else maybe the USB controller doesn't like the amount of ripple. that's just me disconnecting. You know, all the some other little you know spiky noise aspect to that switching converter. Okay, I'm running the latest raspbian stretch.

What we're looking for here is to see if we can pick up any of these boot messages these USB over current change on the various ports. so people are reporting this. There we go and now we can have a look through here and I've actually had a look through and I cannot find any of these USB over current messages. Yes, I've got my mouse and my keyboard hooked up.

Okay, will actually connect the power, then we'll plug in the USB Afterwards, we're in like Flynn our USB devices and now let's go in and have a look right at the end. Generic. Nothing about over currents or anything like that. It's just detected the device.

But what are we going to do now? and I'm going to hook up power meter just that's up trips a couple hundred milliamps. you can see it's 5.07 volts there and then I'm going to switch it on and it should disconnected. Anything over like a few tens of millions will disconnect it. Bingo! All right and reconnect over current.

There it is. we got it. So just a Microsoft moose and a Microsoft keyboard. They're obviously not enough to make it over range in my particular case, but your mileage may vary.

We've got multiple ones there and why it like it shut down like it's saying port to port three? It's say multiple ports and then 248. That's the when we plugged our keyboard and mouse back in and Bob's your uncle. So there you go, we are getting something is causing and logging the overcurrent message there. Okay, so what we're going to do now is actually measure the supply on its own and see what happens.

feeding in 48 volts here from my bench power supply and we've got an electronic load on the output here. so I've got it set to 1 watt load at the moment. Sure enough, we get in enough 5.08 volts there and we're feeding in. Just split 24 volt rails there so we're drawing a load of 1 watt.

Here's but look at the rail here, we're talking 1.4 what's total? So we're you know, pissing away about 0.4 Watts in this converter. but when you're powering it from power over ethernet me doesn't matter. Now here's the issue: I talked about before with the antenna earth lead in this case, an inductive loop, a big ground lead like this going suspiciously near the transformer just happened to be the way I wired it and wow, look at the output, there looks horrible. Look at all that switching component on there and you'll notice that that's actually a high frequency switching we can trigger on that.

There we go it. We can zoom right in on that. and there's all the switching crap. That's just absolute garbage.

but you'll notice that that is just pick up from the lead. That's just bad probing technique. So if actually moves that further away from there, it should get lower and lower in amplitude. There you go.

That's just bad probing. So I'll just move the probe over to this side and we don't have to worry about probing that Now it's all hunky-dory Look at that. So we're just getting what we saw before. no workers.

9 kilohertz. Let's just have a look see if that changes with our loads. So let's go up to a 2 watt load for example. - What's there we go.

Yeah, doubled there you go. 19 kilohertz now. So the frequency varies with and I'm sure if you read the datasheet, this is exactly what it's supposed to do, but the frequency varies with the load. But this baby is supposed to be able to do 5 volts at 2.5 amps, so that's 12.5 watts.

So well, let's go all the way with LBJ Yep, it's still our putting 5 volts, no worries, But our frequency? Well, it's gone way up to 122 kilohertz there. but a ripple voltage is still pretty much stayed the same. so that's not too shabby. And look at those extra switching components.

Hmm. now I've actually got a high-res mode turned on there, so they can be a trap for young players, so we'll take that off. And there we go. That's our that's our switching component down in there where it's a hundred millivolts per division.

So you know 10 20, 30, almost 40 sorry 400 millivolts our peak to peak there. that's on our 5 volt rail at the full output power. It's not terrific, is it? And if you want to know what happens I doesn't regulate properly at lower loads as well, well check it out. We're at 0.1 what's there and well, you know 5.1 it's creeping up if we go down.

look at that. 5.8 Yeah, it needs a minimum of like null point 1 watt. but of course that's no problem whatsoever because it's always getting that load due to the Raspberry Pi So P here, nothing to worry about there, no workers and it's supposed to operate down to our 37 volts. So I've changed it down to 37 volts and 12 and a half Watts output and it's working just fine.

So there's essentially nothing wrong with this. Harare Pi Pero, the Ethernet hat pretty much it. You know it's doing the business except perhaps in the ripple. Department it may just have too much ripple which is passing Pasal.

I've done a video on this how ripple can easily pass through our voltage regulators. The linear the 3.3 volt linear, voltage regulator. So any ripple on the 5 volt rail is going to translate through. Mostly it's going to, you know, especially these sort of frequency since pretty much mostly going to pass through to the 3.3 volt.

And if you got as we're seeing there like hundreds of millivolts, ripple like that can cause all sorts of issues to digital USB chips and stuff like that. So it's the issue. Has to be there. The chip is glitching doing something I suspect it's that USB chip that's a glitch in YouTube just noise and crap on the rails.

Perhaps because like it's certainly not dropping out. which was my first thing that I suspected and it's definitely not doing that and the suppliers do in the business. Okay, so what I'm doing now is 37 volts again. Inputs I sort of like it.

you know. worst case voltage at the raided are twelve and a half watts output power five volts at two-and-a-half amps. Let's just get the thermal camera on here and it's pretty horrendous. Yes, it's calibrated.

There you go, you know, near enough. I'm using Emma City emissivity of 0.95 We've got the Diode here. That's the that's the output diet. On the secondary side, we're talking over a hundred degrees on that diode.

Hundred and ten. The alignments a bit off in terms of like the image camera in this thing to the heat map. but yeah, that diode and then the other diode on the input over here like a hundred and thirty degrees. This is ridiculous that the chip went to our sorry the chip is like we took like a hundred and twenty.

It says nuts his thing's getting ridiculously hot. it's right next to that electrolytic cap to bugger I think I killed it unfortunately I've killed it I I was going to show you that you know if you don't trust the thermal camera I was getting in there with my thermocouple. I was going right on the output side of the diode down in here. so I was going right on the output side and unfortunately I shorted the output of the diode to the filter cap.

So what? what? what? wha? don't? So there you go. I'm gonna do I'm sorry, but I'm gonna call it quits now because I'm not going to go and troubleshoot repair this stupid power supply. So if you liked the video, give it a thumbs up and by all means have a good laugh. Down in the comments at my the stupidity in my alignment of the temperature probe because it's metal and it shorts out and I got in there and it just slipped off and boom.

it got between one of the caps and the diode and something just went and it doesn't work anymore. Magic smoke escaped. Damn. So I'm not going to be able to readily troubleshoot this thing right now.

The USB chip. Yeah, we'll have to leave that for a part two perhaps, but that's where this is looking. So I hope you like that investigation. catch you next time.