Hi Welcome to Fundamentals Friday Today, we're going to take a look at Ripple and noise measurement and specifications. You're familiar with it. You've seen it on your power supply, your bench power supply that you've got lying around. No doubt you see that Ripple and noise measurement.

They might give a typical value for a PSU like 1 molt RMS Sl5 molts Peak to Peak Ripple and noise. What exactly does that mean? What's Ripple and what's noise and how do you measure it? What are the traps for young players? Well, I'm glad you asked. So what does Ripple mean? Let's take a look at that first and you're almost certainly familiar with this term. You've seen it used in terms of linear power supply, for example, and we'll get into that.

now. it can be correctly described as the charge discharge cycle of the storage element in whatever power supply you're actually using, be it linear or switch mode. There's a bit of confusion there, people think Ripple Only Uh is 50 60 Hertz main H that sort of stuff out of your traditional linear Supply that you're used to here halfway half wve Bridge rectifier for example, and your capacitor. then? well, you'll get that 50 hz/ 60 HZ depends on where you are Ripple on the output.

a fullwave bridge rectifier. You'll get double that frequency and you've seen that it's a basic Uh building block thing. You've no doubt mucked around it with your scope if you're a beginner, but that Ripple uh. The term Ripple also applies to a switch mode power supply a DC Todc converter.

Let's take for example, this uh Buck Uh converter here, which you know converts a higher voltage down to a lower voltage. The storage element in this case is the inductor here, and the charge discharge cycle of the inductor in the switch mode converter and that will give you. you know it doesn't look quite as smooth usually as your Uh Main's frequency, which which uh is derived from the mains, which is of course, a sine wave. You don't generally get a sine wave out.

You might get something sort of funny looking like that, but it's still going to be periodic and relatively low frequency. In terms of a switch mode power supply could be you know, tens of kilohertz, up to a couple hundred, Kilz, maybe even a megahertz or two or something like that. But it's generally defined as that base frequency of the discharge and charging of your storage element. be it your rectifier here, or your inductor, your DC to DC converter.

So that sort of base frequency and what? What is noise? Easy noise is everything else. Um, pretty much mainly due to in terms of a switch mode power supply. for example, you generally won't get noise in just a linear power supply like this unless it's being coupled in Via something else. But in a DC to DC converter, for example, you can get parasitic inductances or all over the place and they can cause uh, some high frequency noise or ringing when you've got large didt technical term, it just means large changes in current over time, which you get charging and discharging your storage element.

These uh, parasitic inductances generally much lower inductive values. So therefore, they're going to ring and generate noise at a higher frequency. So you'll find that the noise typically will have sort of, you know, noise superimposed on there like that I can't draw it in there, but you'll see that'll have much more higher frequency content and that's generally what noise is. And in terms of power supply specifications, Well, they lump them all together and say Ripple and noise.

So they combine the two and they give you two figures. They give you a peak-to peak value of course, which is your value from there to there, your absolute maximum Peak to your absolute minimum. and they also give you a value in RMS as well at least your good suppliers. Do you know your cheap ass supplies? they might just give you the RMS value big as well.

Marketing wank, right? The RMS value is always going to be lower than the peak-to peak value now. I've done videos on noise before, and you should probably know from those that a noise figure is generally pretty useless unless you specify it over a particular bandwidth. and well, what is it in the case of power supplies? Well, a lot of Manufacturers will will not tell you, so there's actually no real standard for it As such, pretty much manufacturers will just throw a number out their 1 molt. RMS They won't even give you the bandwidth.

What does it mean? In fact, they won't even tell you what current it is at because the Ripple and noise is going to change with your output current. The Uh noise, for example, the parasitics in the inductors, the value of the change in Uh current with time that I talked about there. Well, that's going to vary with your output current. So the Uh voltage and noise figures.

Well, unless the manufacturer actually specifies it, you've actually got no clue it's It's kind of almost meaningless, but there is a semi defacto standard for it and that's 20 MHz bandwidth. So generally, if it's not mentioned, that's what the manufacturer is really pretty much telling you that it should be over a 20 mahz bandwidth both Ripple and noise. hence why your oscilloscope has that bandwidth button on it and it's 20 MHz or vice versa. The Uh bandwidth figure was taken from the fact that Scopes actually had 20 megaherz bandwidth limiting on and your analog Uh scopes for a long time had sort of like a base level 20 MHz bandwidth.

So really I the number was just sort of picked out of the air pretty much, but most scope should have that 20 MHz bandwidth limit on it. and if they don't, well, if you're using the Yde bandwidth, the scope you're going to get the wrong result. You're not measuring it properly, so if your scope doesn't have that, hey, you might have to build up an external filter to put inside and that's all there is to the theory. Pretty much in fact, I've probably spoken longer than I Should Let's go to the bench.

So let's take a look at two typical power supplies here. We've got this: Powertech Mp390. It's actually a Manson uh 9400. I've done a tear down of this before.

it's just rebranded and it's just a high current switch mode uh, power supply. not really a bench, uh power supply as such. And then you've got your higher quality Ryol Dp832 up here. Let's take a look at their data sheets.

So here's the Manson 9400 and well, look, it's pretty basic Ripple and Noise 10 molts RMS As I said, they're just too scared to put in the peak-to peak figure in there. and they don't even specify a bandwidth or anything like that. And of course, they don't specify what output current it's over. but almost no manufacturer actually specifies what output current.

It's actually at all different values for different output currents, so it's generally taken the Ripple and noise figure generally taken to mean at the maximum output current or maximum output uh PowerPoint And here's the Ry Gold Dp832 and look at this much better. Here we go: Ried Noise and they specify the bandwidth 20 HZ to 20 MHz There's the de facto industry standard there at 20 MHz but hey, it may not always be. So there you go. normal voltage mode.

Here it is. Once again, they've lumped them together. Uh, and we've got less than 350 microvolts, Rms2 molts Peak to Peak. So that's a pretty low noise power supply.

So which one actually means more to you? The RMS value or the peak-to peak value? Well, actually, that's up to you and your requirements for the circuit you're actually powering. But generally speaking, the peak-to peak value is really. you know that's the one that's going to uh, be a pain in the ass because you will get those peaky spikes out of it as we'll see on the scope. So you might think it's pretty easy to measure the Ripple and noise of your power supply.

Just hook your oscilloscope probe up to the output like that and and measure it with or without a load on there. But hey, that's rule number One is that. Uh, generally the rippl and noise is going to be higher at higher load, so you generally want to test it at either the maximum output current or your intended output current for your circuit under test, for example. So we've got a 5vt output here and I've got it connected up to my BK precision.

uh, constant current load up here and I've set it for 2 amps. So there it is. it's drawing 2 amps. Let's go over to the scope and see how we set it up.

and because power supply measurements are typically going to be low amplitude values like in the terms of Mill volts or even submol. Really, you want the best scope you can get with the lowest noise. Front end with uh if possible, you know a good 1 molt per division range. or in this case the Ry Gold 2000 series scope has a 500 microvolt per division range.

Fantastic ideal for testing uh, power supply, uh, stuff like this or pretty much as we'll get into anyway, the way you want to set it up is, well here's channel one: uh, feeding our signaling. you always want AC coupling. You've got to remove that uh DC content of course. Bandwidth limit very important because we have to measure over the bandwidth we can disable.

Well, we can go into say 100 MHz or turn it off and look at that. That is the difference between having your full bandwidth or your what. the actual specification is over a 20 MHz bandwidth. So if your scope doesn't have that hey, you'll have to add a series filter in there.

and I've set up normal mode manual triggering on this so I can adjust the trigger level and well look I can get it to not trigger at all barely. Oh, we got a big spike there, but barely because I'm adjusting the trigger level like that. it can be tricky to actually trigger on uh, noise like this. So generally you want your uh trigger threshold maybe on a noise Peak like that for example.

So anyway, um, generally you don't want to use your auto triggering mode. Sometimes it's not going to work very good anyway. Um, what we've got is our RMS value. The scope can tell us the RMS value before Scopes could calculate this sort of thing.

You would typically use a wide bandwidth uh, multimeter specifically for the task with a true RMS uh value mode to give you the RMS value. but these days, your oscilloscopes can do it. and we've got a current value of not sure if you can read that, but it's like 2.6 M volts or something like that. and these are the statistics average and then the peak to Peak.

So there's our two figures. We' got 20. you know, almost up to 20 MTS Peak to Peak there. And of course, we can, uh, freeze that and actually take a look at the waveform what we're typically getting and it's pretty ugly.

Pretty noisy, but hey, we've got some basic figures there, but you might notice something here. I've used a Times a fixed x 10 probe for this thing. Well, that's not so great when you're doing noise lowlevel measurements like this. you don't want to be using that divider probe really, so you want to stick in a Times One probe.

So I've changed that to a Times One probe. and uh There It Is I've set it up as Times One. We're basically getting similar values to what we got before similar waveform, but you're going to get uh, better signal Fidelity out of your Times One probe because you're not dividing it down. But another trap is that with a Times One probe as you've seen in previous video I've done is that a times One, the bandwidth of a Times One probe can actually be pretty low in the order of like 10 or 20 MHz.

So just check the data sheet for your particular. uh. if you're going to use a scope probe like this, check your particular probe and what bandwidth cuz you may not be measuring over a 20 MHz bandwidth anymore. You may actually be limited by the bandwidth of your scope probe.

There you go. I'll link in the video for that down below. If you haven't seen it, you got an older scope with limited memory depth. uh, during regular sampling, then you might need to use Peak detect mode.

In fact, you probably should. you know, as a general rule, be using Peak detect mode so that it can actually detect the absolute Peaks and you're not missing it based on your time base and your memory depth and stuff like that. So if you want your true peak-to Peak reading, it should be in Peak to Tech mode. That's what it's there for.

And just to show you how that rippl and noise changes with load, Well, that's with my two amp load. If I turn off my 2 Amp Load Bingo Look at that Big difference. So yeah, make sure you know and specify what low current you're testing it at, But you guessed it. I've deliberately added a trap for young players here.

The probing method that I've just showing you before is actually wrong. You shouldn't be doing it like this and I'll show you why it won't be a huge example. I could probably set up a better example, but you'll at least see the difference at the moment. I've got my lead uh, Studio lights above me and they're pulse width modulated and uh, those things generate you know, a whole bunch of noise which gets coupled into our test system here and our test leads and everything else.

Absolutely horrible stuff. So what happens if I turn off my lights here? Watch the waveform. You won't see a huge change, but you should see a difference ready. There we go.

and if I switch them back on there we go. You actually get a bit more noise and it can actually be a lot worse than that depending on the scenario and how it's actually being picked up. In fact, I'll show you a much better example of that. Let's hook it up to my Ryo 832 power.

SL Exactly the same as before. 5 volts out. we're drawing 2 amps into our load over here and I've got my standard uh, oscilloscope probe times one with our Earth lead on there. Let's check this one out and as you can see totally different waveform, totally dominated by the you know high frequency noise content.

because this is a linear power supply as opposed to a switching power supply that we saw before and we're down to 1 molt per division here. Let's that's with my lights on. let's turn it off ready Tada Look at that huge difference. let's switch it back on wa look at all that that is common mode noise being picked up by our piss poor test connection.

We didn't do it right. So the next rule of power supply Ripple and noise testing: don't use your big antenna Earth Lead like that. It's a huge inductor just picking up all sorts of crap. So instead what I'm going to get is a BNC adapter like that.

and I've got a banana plug to BNC like that and I'm just going to plug that into our power supply much better so we don't I mean I can still leave this lead dangling off here. it's not doing anything anymore, but generally you take that out and then we can plug like it straight in nice low impedance, low inductance path through to our power supply connection right at the test. Connection By the way, you always want to measure it right on the output and not way over here. You don't want to measure it over here because, well, that is just going to pick up all sorts of crap.

Forget it. So there we have it. Beautiful low inductance path directly in our load, connected directly across there, probing via the BNC fully shielded, no big inductive path That's as good as we can possibly get for measuring the output of a bench power supply like this. And what does that give us? Look at that and that's with my lights on.

Look I'll switch them off and Tada Look, it adds very little high frequency noise to that where 1 molt per division, we can actually go down to 500 microvolts per division cuz this scope is really, really good. and look, we really can't see those lights. Oh, switch them on. Yeah, there we go.

We've added a little bit more, but it's nothing like before. It's like, you know, half an order of magnitude less than what we'll get in before because we've got a proper loow inductance shielded test connection. But now the problem is with that what's called single-ended connection that we're testing with at the moment, that that's good. and you can do power supply testing like this.

but it's not absolutely ideal cuz we still don't know where our noise sources are coming from. Look, we've got some spikes here I could probably try and Trigger off those, but you can see it drifting across like that. Are they being generated into by the supply or is it coming from something external? Uh, something? you know. and we're getting common mode coupling onto El cable? Well, I don't know.

and for those curious to see what it looks like on a real old-fashioned analog scope, which is still the best choice for something like this. Well, here's my Tetronics Triple 25. once again, it's also very rare on the market that's got a 500 microvolts per division vertical range and see. You know, we can see a bit more detail in the high frequency content in there, but we could also probably see that on our digital if we actually stopped and zoomed in and stuff like that.

But there you go, we're seeing. we're also seeing some of that noise which I'm not sure if that's still common mode noise. uh, common mode pickup from something or what. but there you go.

generally. basically exactly the same thing we were seeing before on our analog scope and there you go. I've got that a bit better on the uh digital scope over here. I've uh, triggered manually now so I'm in there and I've got AC coupling I've got some noise reject on as well I don't think that matters a huge amount, but I've got it set to normal and I'm just holding my tongue at the right angle and uh, tweaking that trigger level and you know, pretty much uh, we can capture that and of course then zoom in on any of the Uh detail.

We were sort of seeing that a bit clearer on the analog uh type stuff, but that is your high frequency noise and the rest of it is that low frequency content like that is your Ripple And of course we can trigger on that uh Ripple because we can go into our source for our triggering and we can trigger off the AC line. There There we go. That's the 50 HZ So there you go. It doesn't drift anymore so you know that is your Ripple.

But of course that sort of line triggering of course only works for a linear power supply where you're going to get that 50 60 or 100 120 Hertz Ripple on the thing. You're not going to be able to do that on a switch mode power supply which has a free runin frequency for its switching converter. Now this is our best possible single-ended test connection we can get for a bench power supply like this. Well, what happens if you want to measure your own Uh design or measure one of these little brick converters or something like that? Let's take a quick look at that.

So to measure your own Supply or a a brick converter like this for example, or something on any piece CCB switching be it's switch converter or linear for example, you always take measure the output directly on the output filter capacitor like that I'm not sure exactly which one and here is here I'm presuming it's the this big ceramic capacitor here. so you'll put your scope probe directly across that capacitor with as low and inductance uh probing technique as possible. So you might use one of your little uh, low inductance, um, ground uh, spring clip adapters that you should have got with any uh, decent set of Scopes and you would probe it directly across there like that or as I've shown in previous videos, you can actually uh, solder a bit of uh, you know, dedicated wire on there like a little hook and loop. So you can basically uh, make one of these out of a wire solder directly on the board and then stick your probe right in like that you want the lowest inductance path possible.

Forget about using this garbage, it's an antenna now. I Said this was the best single-ended method possible. and well, by single-ended your scope probe is a single-ended probe. I.E It's got uh, your input and a ground.

Basically that is a single-ended test connection. Well, that is not ideal cuz we still aren't 100% sure of the of the no noise on our scope. Is it common mode noise or is it actually coming out of the power supply? The only way to be absolute sure and the best possible and recommended way to measure Ripple and noise of any power supply is not to use a single-ended scope probe like this, but to use a differential probe. Now you might be familiar with a high voltage differential probe like this Lroy AP 031 and these are fantastic to have.

and the tool for measuring high voltage stuff because they have differential input like this. Yes, there a positive and negative, but it's a differential input, not single ended and it can tolerate High common mode voltages on the input. But and and it gives you a single-ended output. so it converts differential to single-ended Output That goes into your scope like this, but this is actually useless for our task here because this is designed for high voltages, it's not low noise and it only has Uh 110 or 1/ 100th attenuation, so no good at all.

What you need is a proper differential probe Andor uh differential probe with a pre-amplifier on the input. Now the Ducks Guts is of these. It's a Lroy has very high bandwidth, higher than the 20 mahz required, but hey, it costs thousands and thousands of dollars. So pretty much there's not much on the market in terms of proper differential probing for doing power supply measurements like this.

Now, there's a poor man's way to do this, but it actually works kind of reasonably well and gives you a good ballpark indication of whether or not uh, you've got common mode noise or not. and that's to use a the old technique of having using the Dual channnel of your scope and getting a differential measurement. That way, you'll notice I've got the two scope probes here, but there is no ground connection at all. It's just the center connection on both the positive and negative of our power supply, so the grounds are not connected at all.

Our oscilloscope, of course, is Main's Earth reference. So we're going to get all sorts of crap coming from each Channel But when you subtract one channel from the other Bingo it should get rid of all that crap and give you a true differential measurement across there. Now, the way to do this on an oldfashioned analog scope uh, and a digital as well. But I'll show you analog first.

Uh, we've got both. Our inputs here must both must be AC coupled. Both must be set to the exact same vertical uh attenuation range. In this case, I've Got 5 MTS per division I've pulled out my time 5 uh, time 10 Magnifier! So we're 500 microvolts per division Channel 1 and channel two we're displaying.

We've got both channels active and we're inverting Channel 2. That's important because an analog oscilloscope doesn't have a subtract function. It's only got an add function. But if you have channel one plus the inverse of channel two, that gives you subtraction.

So there we go. we're on ad mode. we're Channel 2 invert and as I said, it's important that these two are, uh, exactly the same range. Otherwise, if you got that cow, make sure your cow's adjusted correctly.

Otherwise, it's going going to be completely out of the shop. And of course, this is a big reason why this isn't a very good technique. You don't get good common mode rejection ratio. uh, using these common mode rejection using this technique.

but it's good enough. But look, what what the hell is this? What is this? It's hopeless. It doesn't work at all. Well, the reason is we've got no ground connection between the scope and our system under test.

So it's picking up a whole bunch of common mod garbage on both these channels and it can't deal with it. So we have to really knock that, uh, common mode uh stuff on the head by adding a couple of 50 ohm Terminators on the input. If your scope has 50 ohm termination, turn it on. So let's plug both our channels back in with I Got a series 50 ohm Terminator on each one and Bingo! Look at that.

We're now in subtract mode as I said. if you muck around with uh, any of the vertical uh settings, if they're not completely mat Ed like that, you're just. it's just not going to happen. And of course, if you don't invert Channel 2 H you're screwed.

You're just looking at that. and bingo. We're getting bugger all there and you expect it to be bugger. All because.

Well, the Uh Ryo is a very good power supply. Let's go over to a better example: much higher noise. You know we can't go any further on a vertical. Let's go back to that horrible Manson switch in power supply.

So there you go. That's the test connection. Back on our Manson Supply there. Exactly the same load we had before and being go look at that.

There we go. there's our Manson power supply output. There's some of the high frequency stuff in there we can actually turn our alt zoom on and we can actually see that zoomed value of that zoomed part of that uh noise in there. Look at that.

So there's our Ripple and there's our noise using our differential measurement on our analog scope Like that and we should get exactly the same on the digital. Let's go back and try that and we're back on our Ryo here. here's our Channel One Channel 2 input and I've got the math operator on a minus B there and as you can see, it's a bit slow updating on the screen there, but we're basically getting exactly the same waveform we get before with some high frequency content in there we weren't seeing on our analog scope, so what we'll do is we'll just expand that out a bit. We'll go to our acquire menu here and we're in normal acquisition at the moment.

We'll change that to our highres mode with our Box Car average in and Bingo! Look at that There we go. We're getting exactly the same waveform will get in on that analog scope there on our digital scope. but the waveform updating? eh. little bit slow.

and that's one of the problems with the Rial scope and a lot of other scopes on the market. They will do all that math function in you know, processing in software so it takes actually time when you turn those math functions on. That's why they slow updating if we go to our adelent 3000 Series scope here. it does all the math uh stuff and in direct Hardware on the As6 so it is much better.

Much quicker updating. but the problem is this scope only goes down to 1 molt per division, but it really only has a true 2 molt per division. The 1 Molt is actually just a software tweak, so we don't get the greatest Fidelity out of our waveform here. So in that respect, the Rol scope with its true low noise 500 microvolt front end much better for this purpose.

And here's an interesting thing to note. What I'll do is I'll adjust the gain here on the well. the scale of our math function where current C ly at 1 molt per division there. So I'll tweak that down here and you'll notice.

Look at that. You start. if I go up one to 500 microvolts per division, you start seeing the individual bits in there of the math calculation. So this is uh, you know, one of the disadvantages of a digital scope and one of the advantages of the analog.

Once you get down to with large differences and dynamic range, you've only got that 8bit converter in there to play with. So really, you know? look, you can start to see the individual individual bits there. Just crazy, look at that. But hey, at least we can see our waveform.

So that's pretty good. And that value is going to change with our scale here. if we turn our vertical, uh up. Of course we get increased Fidelity and resolution in that calculated math value because it's only got those eight bits to work with or more if it's the high res mode.

But let's take it down to say 5 MTS per division on both channels. Look at that. Totally blocky because it the signal. The amplitude is down in the noise so it can only calculate look.

you've only got a couple of bits down in there. Ah, it's bugger all. So when you're using a digital scope, just make sure you maximize the uh use of your dynamic range of your uh front end by using the lowest uh, vertical scale possible. And here's another trap for digital Scopes as well.

there's our waveform. What happens if we move our one of our channels off outside the range of the ADC so it's clipping. Look, Look at that. Our calculator math value just goes to complete garbage.

That's one of the advantages of Anal Scopes Digital. Look at that. That's awful. Real trap if you don't know what you're looking for.

But if you were observant, you would have noticed that our amplitude here that we getting is much lower than what we were getting. Our differential, uh, waveform here is much lower in value in amplitude than we'll get in with that single-ended connection. Why we're still using Times One probes here. Nothing's really changed, just subtracting one signal from the other.

We should get the same value, but we're not. Remember, look at our scale. It's 500 microvolts per division here. and it's basically I Can uh, go in there into the math function and adjust the tweak that up.

It's like, you know, two divisions sort of peak to Peak there. What we're getting before we're getting. What about 10 m volts Peak to Peak About 10 times more. So what I've set up here in parallel is our single ended connection as well.

Yeah, you can do this. Ordinarily, you wouldn't but we're going to get away with it here. So we got a single-ended and our differential probe in it as well. So I've got the yeah, the proper high frequency connection there.

it's going that third. Channel Now single-ended is going to our agilant scope because our Ral scope is only two channels. So let's have a look. So there's the single Ed measurement on our agilant scope.

look. 2 m Vols per division. You know, 2 4. You know it's almost six.

sort of. You know, something like that? Hey yeah, let's not dick around with the triggering. And if we go up here and have a look at our differential measurement, then as we said, we're 500 microvolts per division. so we're barely even one mill volt there Peak to Peak.

So there's about a six odd times difference. Where is that coming from? Well, if you're a remember a previous video I've done which I'll uh Link in in that, um, the how these uh oscilloscope probes work. The coax cable isn't just a direct connection straight through, it actually uses a lossy coax. Which means it has a resistance in it and we can actually measure that.

Look, let's get our multimeter here. Here we go and measure our Center conductor, which you'd expect to be a dead short. It's not. It's about 300 30 ohms.

Aha, we've got a 50 ohm Terminator on our scope to get rid of all that uh, crap. and bingo, we've got a voltage divider. So if you work out how much the signal is being divided Time by, Well, it's roughly 7 and 1 12 times with that 330 ohms and the 50 in there. So that's why our amplitude on our differential measurement is so low.

So to get the True Value that you're actually measuring, you have to multiply that by the measured prob. But in this case, well, you wouldn't be using scope Pros for this measurement. So I've actually. uh, another trap for young players is when you're doing this sort of stuff, you wouldn't be using a scope probe like this.

It's good enough to get an indication like this to see if there's any uh, you know, get rid of any common mode noise or something like that. But hey, in this case it is not the correct method. You should be using direct coax 50 ohm loaded. So what we get down to is the ultimate correct method to do differential probing like this.

If you don't have a proper you know, really expensive. uh High impedance differential probe. This is how you would do it You' have your coax from the scope of course 50 ohm terminated on the end here and you would have a 50 ohm uh Source resistance in here as well. And just to get rid of any DC out of there you would have AC coupling in both lines and that is that becomes your differential prob but with the 250 ohms in there You' still got an attenuator so you fin value that you're measuring.

hey, you still got to multiply it by two. So there you go. But that is how you would do proper differential measurement with a scope or with a preamplifier. Usually you would, uh, you know, use the differential measurement into a preamplifier.

especially for doing something like the Riol scope here which has noise we can't really measure. You know, with even with our 500 microvolts per division here, it's just not really doable. So you would, typically, you know, whack in a x 10 preamplifier or something like that in there? Yes, you could do this with two single-ended uh, preamplifiers if you wanted to. with your scope and stuff like that and you know that is still work.

But a proper differential amplifier like that Lacroy one? That's the one you want. so there's no absolute requirement for actually having the 50 ohm termination here. If you've got a proper differential, uh, preamplifier. they're usually uh, High impedance.

You know that'll be 1 Mega or 100 Meg Or you know, really high impedance and you don't have to do that. But if you are going to turn them, M turn, terminate them like this like you do for a scope, then well, you're basically looking at, you know, transmission line stuff and you're supposed to match source and load impedances so you don't get Reflections and things like that. So if you are doing that, yeah, proper 50 ohm Source impedance and 50 ohm load over. There is the way to do it.

And if you were doing this with the home brew probe approach, of course you would keep absolute minimum. Uh, these paths in here you wouldn't have anything exposed so you'd have your coaxes something like that. You'd have your 50 Ohm in series, it'd be nicely heat shrunk, and you'd have your tiny capacitor and you connect it directly across your bypass cap of your power supply to be measured. But you know, really, that's just to do-it-yourself sort of a custom hack if you need to.

As I said, by far the best way to do it is to use a proper differential probe, preferably with a preamp for measuring. uh, low noise power supplies like the Roo one we just did. And here's the money shot. You know how I told you that the reason we're doing this differential measurement is so that we can see if this noise that we're getting single-ended probing on back to my Ryol Supply here.

So we get that really horrible noise on there. those spikes. We wanted to know if those spikes were actually coming out of our power supply. Well, if we go up here and take a look at our differential measurement, look, we're probing the exact same point.

No. look, they're gone with the differential measurement. They're not there at all. So that noise is not coming from the power supply.

But if you're using that single-ended probing technique, then hey, you could be easily fooled into thinking that power supply was a lot noisier than what it actually is. And if you're curious to know the uh frequency of that noise, well, it's around about 142 odd Hertz between those with two spikes there and there. So that's being picked up somewhere from something in this room. Aha I Found the culprit.

It's part of the test set up. Check it out. That's the waveform we get in now and but look what I've done. I've disconnected my electronic load and I've got connected.

just a resistive, uh, dummy load here. Similar amount of current. we're taking 2 and 1/2 amps instead of 2 amps, but it's vanished. Bingo It's only when I use connect up my electronic load.

it's adding that switch in there at 142. HZ It's coming from this damn load. So you got to be careful of your test setup like this, and where your noise is coming from. And if we didn't use our differential probing here to actually confirm that, we would have thought for sure that it was coming from this Ryol power supply.

but it wasn't, This thing's clean as a whistle and this thing is a culprate add in, you know, normally not an issue, but when it's coupled in like this, well, it causes all sorts of stuff. common mode, so that's to do with the you know, the design of it. Who knows where it's picking it up internally, but it's definitely coming out of here and interfering with our test setup. TDA So there you go.

That's the basics of Ripple and noise measurement for a power supply, or maybe even one of these brick converters or your circuit or whatever like that. And we've looked at single-ended probing. We've looked at common mode noise. We've looked at differential probing.

We've looked at uh, you know, secret attenuation in your Crow probe Crow cathar osilloscope probe. that's what we call them here in Australia Anyway, Um, yeah, your scope probe. and well, there's lots of traps for young players. There's a lot of art which goes into actually getting, getting real, proper measurements on these things and knowing exactly what you're doing and not being fooled, especially by common mode noise.

Just because you see it on your scope doesn't mean it's actually coming from your device under test. So anyway, I probably there's some stuff I haven't covered in there as well. And yeah, there, there's a lot of around to try and do this right. but hope you learned something there.

And if you like the video, please give it a big thumbs up. Beauty And if you want to discuss it, jump on over to the EV blog Forum that's the place to do it. Link is down below. Catch you next time.

.