Hi I Thought would take a quick look at the power supply sequencing, which is an important concept in many areas of product design, many areas of electronics. I happen to be doing a repair here of this edge alone five, four, six, two, two d oscilloscope and it's got our four different voltage rails and sure enough, spoiler alert the I Well, one of the problems with it is that the power supply is blowing. but what I wanted to do is actually power up the oscilloscope. Obviously I don't have a replacement power supply, you know, right to hand.

So I'm pairing it from my bench power supplies. In this case, I've got three different power supplies here generating four different rails I'm generating 3.3 volt rail down here at Four Odd Herbs I'm generating when I switch it on I'll be generating fifteen point seven volts on channel one here and minus five point two volts on channel two and this one up here which I haven't turned on yet because it doesn't have an output load switch that's going to be the five volt rail or actually five point two volts. So I need to power up this oscilloscope using these convoluted array of three bench power supplies. But I can't just go turn on, turn on, turn on, turn on in whatever sequence I Like or it might be able to.

But this is why we need to talk about power supply sequencing because these four different voltage rails. Ideally, when you turn on this supply, they all power up and give their required output voltages from the output regulators here at roughly the same time you know, give or take a millisecond or two you know something like that, or they're actually sequenced to power up in step by step. I Because depending on your system electronics that you've designed your product like an FPGA for example that has many different power supplies. If you actually go in and take over the datasheet for these, it might actually have a recommended or even a strict powerup rail sequence.

so you might have to bring up the 1.2 volt core voltage first and then the 3.3 volt. IO. You know, ten milliseconds later, you may need delays in there, and there's actually a whole range of chips on the market just dedicated to this particular function. Power Supply Sequencing: Because not all products need to just bring all their rails up at the same time, sometimes you might have to delay them tens of milliseconds hundreds of milliseconds in a particular sequence in order to prevent all sorts of issues with your logic be it designed into the silicon issue or about whether there's like a complex system stuff.

Because if you've got a complex you know board like this, right? that's a real complex digital system that you're going to pair up many different rails and if you pair up things in the wrong sequence in like the best-case scenario, it just may not work for example, but in the worst case scenario, it can cause SCR latch up which I've done a video in which I might have to link down below and cause the magic smoke to escape from the chips. You can actually blow up chips and other circuitry by incorrectly sequencing the power supplies on boot-up ie. turning one on before another. There might be you know, diode protection and stuff in there that causes over current and latch it.

all sorts of issues you just don't know in a complex system like this. So certainly when you're doing a a repair like this, you don't know unless you've got a good one and you can actually put this or scope on see which power rails power up first. In this case, you go yeah, we just want them to sort of power up all at the same time. That should be reasonably safe and that's probably what this power supply here is actually doing.

I Don't think this should have any sequency that we could have some crude RC sequencing, but I don't see any active logic power supply sequencing stuff in there, so this one, it's not that critical I Think you know just to bring them all up at the same time. but hey, I've got to operate. you know, like I've got to go shift output on I've got to turn these to one and I gotta flick the switch up here all at once I can easily be half a second off or and I was second off if I goof at pairing at one of these rails and I could actually destroy the product. But in this case I think we're fairly safe.

But hey, ideally we want to build a little circuit that allows us to pair up all these rails at the in time. So let's take a look at it, right? So what I'm going to do now is just do it by hand of course. I've got no circuitry in there I'm just going to try and use my third hand here. Pair up all these at once.

I've set up my oscilloscope all four channels so we'll be able to capture all four power rails. Okay, so I've got to hold my tongue at the right angle. Let's put in single shot to capture mode ready to capture. I've got to go shift like this.

I'm going to have two fingers on here. one thumb up here. turn at the right angle. Here we go.

Well, I missed one. they missed one. But anyway, our oscilloscope is booted so no worries. Okay, but we've got our power supply sequencing over here.

Let's take a look. Actually, just the kicks. I think I might redo this with the road and Schwarz touchscreen scope with the annotation. Yeah, hold on a thing.

Okay, this is what I got after I powered it up. and do note that there are different volts per division settings here for the different waveforms just so we can fit them on screen and see a bit more detail in here. Now the first one I powered up here is Channel 4 at 1 volts per division. so 1 2 3.

that's our 3.3 volt rail. We see how it powered up nice and cleanly. That's the one. I Obviously pressed first and we're on 500 milliseconds per division, so you can see there's quite a, you know, like 800 milliseconds before the next one before.

I Press the next button I Deliberately press them a bit slow so that we can see some detail here. and perhaps a few odd things happen in here, which could be a real issue. So let's take a look at it. 3.3 volt rail powered up first and the next one to power up here is not the green one.

I'll talk about this in a second. Next one to power up is the yellow one. so it's 5 volts per division. so that's our 15 0.7 volt rail there and that one's smelly, you know, powered up nice and cleanly.

If you actually go in there and have a look, there's a bit of a bit of a little wiggle in there, but that's you know, just the power supply starting up. Not a not a big deal or didn't overshoot. None of these power supplies are doing any hideous overshooting like that, which can really are damage components a good power supply should not. Good loud power supply should not overshoot.

Well, any power supply shouldn't overshoot. really. And then we've got and then the next one about six, seven, and 700 milliseconds later after that one is the - 5.2 volt rail because I've got them all set to exactly the same Center reference right in the center of the screen. Zero volt reference so that one obviously went negative - five point two and then once again, another 700 milliseconds later I didn't though I timed that nicely, didn't I Then we've got our three points.

Three Volt rail. Odd, sorry our five volt rail - Four five volt rail there. But look at this. you see how this five volt rail here is actually going up at 2 volts per division is go up to like one and a half volts immediately after we powered on that our 3.3 volt rail there.

So we powered on a 3.3 volt rail first. But I swear I did not turn that power supply on until last. So it's forced through the 3.3 volt rail through probably back diode protection or other things happening. Is it's forced the five volt rail to actually go to about, you know, one and a half volts or thereabout.

So that could cause a serious issue depending on your system electronics, so you can see it's actually forced that to happen. So this is not the power supply starting up and then doing something you know, and then doing something weird for a second, a half or two seconds going. Oh, I don't know what to do I'm going to set my output to one that both know. it's the 3.3 volt rail, dragging the five volt rail to that level until we switched on the five volt rail.

And then it's a big low impedance output and goes right. I'm going to provide my real 5 volts. Boom up She goes. So there you go.

That right there could be a problem. It could cause your circuitry not to start up properly - based on all sorts of complex system stuff that you probably can't even think of. And there's lots of traps for young players anyway, that that could cause all sorts of issues. Now you notice that I'm pretty much the fifteen point seven volt rail.

Nothing's dragging that bit, right? so there's no issue there. It's probably separate. It's not. it's doing its own thing.

stays at zero volts until we switch that on. Second, everything's hunky-dory there. But you also notice that look at the minus five point to the minus five point to here. You'll notice if we actually go to that select that channel.

Okay, you'll notice. Look right up there. Once again, we power on our 3.3 volt rail. It's actually dragged the negative 5.2 volt rail high by about what are we 500 millivolts and say, like 200 millivolts are straight at high and then it's And then only once we switch it on is it low impedance enough to go right? A bugger that I'm not going to be forced to do anything I'm going to go to where I'm supposed to go at Five Point Two.

So there you go and you can see like a little. Also a little dip in the green one there, which is the five volt rail as well. When the negative five Point Two volt rail switched on, we're going to put the little you know, little blinky so this is a complex behavior going on. There a lot of power supply forcing, which is quite common, which is another reason why a you might have to sequence your rails like this one is a classic sequence.

okay, like this one and then 700 milliseconds and the next one delay. boom boom boom boom. In this case, we've got a very dumb power supply inside this oscilloscope. Most likely we want them all just to switch on at the same time and we don't want any of this force in rubbish.

But I obviously can't do this with just my fingers I'm not quick enough I've experimented with it and I can get it reasonably close all within maybe half a second or something, but half a second. it's a long time in electronics, let me tell you. and at one point I actually I think powered them up in an incorrect sequence or something I actually got an error message on the scope. I'll put that clip up right now just to show you this.

I got this error message when I accidentally switched it on, then off, then back on like half a second later. Oops. So either one, our power supply to power up in sequence like this 1, 2, 3 and then fall like that after you know a period of time, all we want them to start up. Unless you know what you're doing, you're usually wouldn't sequence power supplies you just which among all at once is generally the safest option unless you have data to prove otherwise.

So what I'm going to do now is just build a very simple MOSFET switching circuit that'll allow us to switch on all four of these rails at the same time. Please excuse the crudity of this: I've already attached the probes I'm going to take them back off to show you up close, but I've got three yard power MOSFETs in there I'm going to switch our three of the rails I'll show you the schematic in a minute, but I'm basically going to well switch the 15.2 seven volt, the three point three, and the five volt or at the same time. Okay, so what I've got now is all three of my supplies all switched on at the same time five point two, or fifteen point seven and three point three. they're all switched on and I've just got a lead here which then I can plug in to ground which will then switch those on and the fourth rail the minus five point two I'll just switch that on separately.

So here we go. Give it a bill. There we go. I've captured something.

Have a look. Please excuse that I Accidentally swapped a couple of the channels here, so the colors are the same, but you can see that all three rails are fifteen point seven volts. Three point Three and five all start up at the same time. and yeah, it's a little bit off of 102 and three on a 420 milliseconds off.

Starting up the negative one, you can see that way they all start up or practically identical. There's a little bit of funny business going on in there. Now here's actually a good example of what's going on here. I recaptured it with a little bit of funny business.

Look at all this garbage in here. Let's just zoom into the start here. But look at all this, right? But they all start up reasonably. You know there's a little loop thing happening that I could easily be the power supply or whatever you know, like probing issues.

whatever. But look at all this, we've got digit all this sort of stuff happening in here and if we go further out to right out here like this, it's really horrible business. Look at all that garbage. What I was doing there as I was actually our because I'm switching it with just a connector like this.

or if you're using a push-button switch or whatever that is contact bounce. So when I'm actually plugging that into the banana jack, it's not just making one solid contact. I was kind of like I In fact I did it like a couple of seconds like a second or second half afterwards. I'll say like one second afterwards I was just going to happen and that is contact BIOS for you.

So you've got to be careful that of course this wouldn't be an issue if you're driving it with like a logic signal because you get a nice clean signal. but if you've got like a physical switch, just be aware that the MOSFET will switch off and on very very quickly for that hard. Now in this case, our cell scope works just fine. It could handle that.

no problems whatsoever, but more sensitive systems that switch bounce may be a real issue. but of course if you use one of the proper our power supply sequencing chips or if you just use a digital logic or a pulse generator or whatever a single shot or you simply add an RC filter to the to the contact the gate of the MOSFET then you know it can solve the problem. I Just wanted to demo that. Here's a quick take a drawing of what we've got: I've got three power MOSFETs these are sub 53 po6 or - 24 those playing along at home.

Kind of specifically designed for logic load level switch and I won't get into selecting the right MOSFET It's got to have like a low Vgs voltage and stuff like that. And of course it's got to have a low RDS on a low on resistance for the particular current so it doesn't drop a lot of voltage across the X If you're feeding in fifteen point seven volts and you know a couple of amps, you want to get your fifteen point seven volts out. You don't want your MOSFET switch to be dropping a couple of volts or half a volt even. You know a few hundred millivolts could ruin your day.

And of course the MOSFETs have to have a particular are Vgs voltage and a maximum voltage and a you know, maximum drain source and everything. There's a lot of stuff that goes into picking the right MOSFET if you really want to go into the details anyway, this is just like a jellybean, a heavy current load switching. it's a P-channel MOSFET I'll explain why in a second and we've got a pull-up resistor. If you just ignore this, it's just ignore the other ones.

We basically got a pull-up resistor that connects the gate to the source here and that keeps the MOSFET switched off because it's a P-channel one. and of course our load goes down here to ground. But if we ground our gate with our switch down here, you notice I've tied all three gates together so they're basically or working identically and I tried it to the higher voltage up here. So if you get a multiple rail like this, that's where you'd be trying to watch out for maximum voltages on your MOSFETs as I said.

but you turn on the one switch, the gates all go to ground. You get your desired Vgs voltage here. In this case it'll be the negative because the gate and Saucer back to front and Bob's your uncle. You must fetch, switch on and supply power out of your rails.

Easy. Now of course that's using a P-channel MOSFET and you typically for this like high side application like this should use a P-channel MOSFET Just for the simplicity of being able to ground this pin here, it's just a P-channel is a simpler control operation on the gate. but hey, you can do it with an N-channel mosfet here. it looks very similar except our drain and source I swapped here and it's an end channel and it's basically doing a similar thing.

Are you grounded to turn it on? But the difference is is that the you require a higher VG voltage than your V in here. so the V the gate voltage here has to be greater than or equal to the output voltage V switched I've call it a load voltage plus V our threshold voltage of the the Vgs, our threshold voltage you get from the MOSFET So and it's just. it's a little can be a little bit more trickier to switch with N channel. so P channel is a bit easier just for the simplistic application we've got here and I didn't bother to switch the negative rail I didn't want to confuse things, didn't have them part of there.

Let's not go into the whole thing, but that's how you switch the three rails. but I could have done it also switching the negative rail as well at the same time, but your control requirements get a bit more complicated sometimes so there you go. I Hope you enjoyed that. Look at MOSFET switching and power supply sequencing over rails like this and if you did, please give it a big thumbs up.

And if you want us cuss it, links down below to the forum and YouTube and blog comments and all that sort of jazz. Yeah, hopefully they'll be repair video, but this thing seems to have quite a few issues I shot half of it but I didn't want to release it until I had I Don't really have a happy ending on this puppy yet. but yeah, coz there are a couple of things wrong with it anyway. Um, I found a little issue with this - I might do another second channel video for that as well.

but anyway, hope you liked it. Catch you next time.