Hi, Let's say you've got a DC to DC converter that you want to measure the performance of whether or not you've designed it yourself into a particular product, or you've bought maybe an off-the-shelf one like this which doesn't provide any I characteristic graphs, performance graphs, and efficiency grass for it. So how do you actually measure the performance of a DC to DC converter and get your typical efficiency curve which is the efficiency in percentage versus the output current draw. And you'll find this in practically every data sheet for every DC to DC D converter chip on the market and these are usually typical. but it but the actual efficiency of your particular DC The DC converter that you design is dependent upon a whole host of things: what type and what size inductor you've got output capacitance if you've got an external switching mosfet, what type that is, and all sorts of and the frequency you operate at all sorts of different parameters going to determine the efficiency of a DC to DC converter.

and sometimes this efficiency characteristic graph will also include power loss as well. Which is that quite typical because you want to know how much power is being dissipated in your little brick converter here. So how do you actually measure and graph your own characteristic curve like this? Well, let's take a look at it. I've actually kind of done this in several or much older previous videos, but not a dedicated video for it.

So let's have a look in the case of this little low digi Lindt 12-volt power Brick. Here it's a boost converter five volts into 12 volts out. Datasheet doesn't have the efficiency curves, let's measure it. The efficiency of a DC to DC converter is just the output power divided by the input power.

If the output power exactly matches the input power ie, you get one Watts out for one. what's it Then it's a 100% efficient converter, which is basically impossible. You can't get a 100% efficient DC The DC converter a typical really well designed good DC to DC converter will typically have an efficiency greater than 90% You know a real kick-ass one might be like 95% So you're going to have some loss in the converter here. So what we need to do is measure the input power going in and the output power.

So we need a power supply and we need an electronic load. I've done a whole video on making your own electronic load. Very popular do-it-yourself art project so I'll link that in down below. If you haven't seen it, you can make it for using Junk Bin parts for practically nothing.

but we need to get the input power and the output power so we can actually do that with these two instruments here. So we've got a modern, smarter lab bench power supply here. This is a Roy Gold EP 8, 3, 2 and it shows our input voltage and our input current 5 volts, 140 milliamps and it automatically calculates our input power for us so we don't have to calculate that with our calculator later. Beauty! So basically the input power here is not 0.7 watts and on our output, we've got our electronic load.

Modern one like this can easily accurately measure the output voltage and the output current as well here. and it also calculates your output power. So I'll set a constant current output load here of 50 milliamps. Naught point.

Oh, Five Amps. And our output power is not Point Five Nine Watts. So not Point Five Nine. What's divided by not Point Seven Watts input here gives us an efficiency around about 84% That's okay, but if you remember that efficiency curve that we want, its efficiency on the Y-axis versus output current on the X-axis So we have to sweep the output current here, set different loads, and get all the data points for the efficiency.

So that divided by that for different values of load current here. and we have to do that over an extremely wide range of output currents. It can typically for a you know, a universal type DC the DC converter. The data sheets as an example here will show typically are 10 micro amps up to an amp for example, and they'll do that on a logarithmic graph because otherwise you can't fit it all in.

But AHA There's a big trap for young players here, and I've mentioned this in many previous videos, but it's very important in this scenario so I'll go over it again. You'll notice how our power supply on our input here is showing nice, precise 5.00 zero volts. It's a real accurate power supply, so you can believe it. but that's five volts sensed right at the output terminals here.

We've actually got these wires going over to the breadboard here now. I've actually got another multimeter set up that's actually probing directly on the input pin there. Okay, so it's actually after the drop in or in the in the wires here. Okay, so the ground and the input and bingo.

look four Point Six Six volts at the input to the converter. So our five volts is way off there, right? So we've got a real large, very significant error there due to the drop in our wires. Oh lucky we actually measured it right at the input. And likewise, we're going to get a similar error on our output here because we've got these long wires going over there.

You know they're reasonably thin wires. They're long. There's going to be some drop on those at a significant current. We're only draw in 80 milliamps.

But hey, look at the error here. Okay, we've got our nice precision supply here. Look at this. Eleven Point seven, Seven Three Seven.

and this is a this is a really kick-ass electronic load point. Oh, five percent precision. Fantastic. But it's sensing it at the output terminals here.

So it's including the drop along these wires. We don't want that. We're measuring the efficiency of the converter, not the converter, plus the input and the output wires here. And once again, look at the discrepancy here.

I've got a meter on the directly connected to the output pin and the output ground there and you can do this because there's a 10 mega ohm input impedance on your multimeter. so it's drawing no current through any of these leads. So you are actually measuring the true voltage on the output. There, you're sensing that this is called a four wire sense measurement and look at the discrepancy.

It's this converters actually outputting twelve Point One, three volts, but the our load is only measuring Eleven Point Eight. So we've got a discrepancy here. Very significant and a very significant discrepancy on the airport. So if we just use these two instruments and didn't do for wire terminal measurement, but we can get very significant errors which would completely ruin our efficiency curves.

So real trap for young players. Beware, make sure you do four terminal measurement. So simple. Dave CAD Drawing showing you four terminal or four wire measurement sometimes called a four wire sense measurement whatever you want to call it.

And we've got our DC to DC converter brick Here, our input and our output. the ground just assume that the ground is the same one common pin whatever it happens to be, and you see that. We've got voltage sensing right at the input pin there and right at the output pin. Therefore, both the output for the positive in and positive out and also the grounds as well.

Because you're going to get losses in both your ground wiring and your a positive wiring as well, on both input and output and then your ammeter. your current meter goes after that, so then you've got your variable load over here. It can be a dummy resistor electronic load doesn't matter what it is and your input, a meter here, and your adjustable power supply here. So you don't want to read the value on your power supply here.

you want to read the value on your voltmeter here unless you're working at very low currents. In which case, you're not going to get any loss across your wires, but just assume that you're going to get losses and you need to measure using the four terminal technique. This is why it's called a four wire or four terminal because there's one, two, three, four wires for each measurement point. And of course, you don't need fancy gear like a a modern programmable our power supplied to display power or a really, you know, high-end precision electronic load like this.

All you need is four multimeters and I said it before. I'll say it again. This is a classic example why any well-equipped electronics lab should have four multimeters. It's not hoarding, It's not a multimeter fetish.

It's to measure input power and output power of a basic DC to DC converter. Power supply is very common in any electronics lab. to do this. if you haven't got four meters, you can do it.

but it's a pain in the butt. So all you need any lab power supply and you don't have to worry about the voltage and current readings? Doesn't matter, it doesn't have to be fancy pantsy. Just any supply will do. two meters: measure inputs our voltage and input current and output voltage and output current.

And the good thing about this? we will actually have to resort to this because if you have a look at the efficiency curve again, you'll notice that it went down to 10 micro amps. Okay, right up to an amp. And if we have a look at, you know, a really good lab electronic load like this. It's only got one milliamp our precision here, only two digits on the output power.

It's bugger-all so you know we can't We can use this for you know, large output currents and the voltmeter is very precise so that's no worries whatsoever. But the but setting our load current is no good. Okay, so we need something else to actually generate the very low loads, the 10 micro amps and stuff like that, we at least need to be able to measure it with our and with a separate output to our current meter that can measure those low currents. precisely.

This particular thing can't do it. So yeah, I can just use a resistor. Something like that. We know the output voltage roughly 12 volts.

We can just whack a resistor in there and get our 10 micro amps on the output tissues. Ohm's law: Very simple and really good-quality electronic loads. Know all about our 4 terminal measurement and remote sensing and it's got a remote sense option. There's are some sense terminals on the back, sometimes they might be on the front.

this time it's screw terminals on the back and we can just select our remote sense on. Fantastic! And if we go back out and bingo, this now matches our meter well. our resolutions? not there. But yeah, it basically matches because we're now doing four terminal measurement with our electronic load.

And another thing to be aware of. in this case, I'm just measuring on a breadboard. It's a little bit Dicky in here. and if you muck around with the wiring and stuff like that, things can start to change.

You know Dickie contacts on bread boards and wires and stuff like that. If I was doing this properly and professionally, I would actually sacrifice this thing and actually solder the wires directly on four wires on the input and four wires on the output to on the terminals. So then nothing can go Dickey with your measurements and along with your efficiency. If your water, you can measure other parameters well, like you might measure the switching frequency.

for example, you'll get that by typically probing the inductor in there depends on the converter you're using and you can have a look at the switching frequency. Because your switching frequency, it might be a converter type. that's the switching frequency changes depending on the output power and this or fairly typically happened with converters that want to get maximum efficiency across like at very low currents as well. So that's you know.

it's not uncommon a lot of DC DC convertors a fixed frequency, but a lot of them will actually change their frequency to make them a more efficient over a larger output current range. and you might want to measure temperature for example. So you might actually get in there and attach a little thermocouple probe to your converter or near your converter or whatever may be on. If it's using a heat sink, you might attach it to the heat sink or something like that and you can plot temperature versus your efficiency and output and load dissipation as well.

Just you know if you want to be thorough and that could be a big deal because hey, your converter might work. No worries, it's ambient temperature or whatever. Everything works just fine in the lab. It gives you the efficiency you want, gives you the output power you want, but if it's running at a hundred degrees Celsius you could be in trouble.

You could come a gut sigh. And it's not going to work in the field. It's not going to, you know, have a long lifespan, whatever. So you know you might want to measure something like temperature as well and plot that along with your efficiency.

but we won't do that today. Now you could actually automate all this. Of course, a lot of modern instrumentation is all Ethernet LXi Controlled. For example, this power supply is this BK Precision One can be remotely controlled as well, so it could actually script this to generate different output voltages and stuff like that.

But hey, we don't have four terminal measurement on here, but I could hook up some data log in well, some Ethernet LX I connected bench multimeters that I have. It could automate the whole thing, but you could spend like a whole day just setting this up. It's easier just to hook for multimeters on the input and output and increase your current and just note them down on a notepad and then just walk them into a spreadsheet so you know it's you'd only automate this if you really wanted to just for kicks all. you had a lot of converters to measure and you can use some more advanced instruments like say this: Keithley 2400 Sauce meter or Shmoo SMU for example, you might have like a multi-channel SMU system for a real complex measurement, but as I said, you don't need any of this.

You can get away with just a couple of multimeters and a dodgy power supply and do it yourself. Electronic load and other stuff like I output ripple voltage might be important for example, or verses output capacitance. As many other different parameters that you can do to measure a DC to DC converter, it's almost the sky's the limit. Now, when we actually go to measure this, if we take a look at our little diagram.

Again, the burdened voltage on our current meter on our air meter on the output here is basically not going to matter as long as we've dialed in the load to get out output current. So if it's an active load, it's going to sync that. But if it's a resistive load, then you're going to have to tweak it depending upon your burn voltage here. Likewise, on the input that you might think that the burden voltage er ma, that doesn't matter because you're measuring the input voltage here and that might be true, but generally you want to are your performance a curve for your DC to DC converter is at a noan input voltage, so it might be 5 volts DC input.

So you don't want it to vary based on the current because as you decrease your load, you draw more current on the output. You're going to be draw proportionately more current from the input. Here, you're going to get extra losses across your burden voltage er, um, either your wires, whatever you've got in there, And sure you're measuring the exact voltage. But that's no consolation.

If you actually wanted a performance curve with a fixed knowing input voltage which is generally what you'd want, you don't want it to change. So really, you need to tweak your supply so that it so that you're taking into account the burn voltage, your multimeter and you could use something like a micro current for sample, but yet you're still go is still going to change. You're gonna get some loss across there, so just watch out for that. You may have to tweak the power supply each time.

So this is where an automated setup helps. If you've got an automated power supply that has remote sensing like this, you can program it so to provide exactly 12 volts on the input here, as well as measuring the current, it can do all that. But because we're using just manual multimeters, manual instruments like this. Yeah, and we can have a look at the effect here of the burden voltage.

You've got the input current here. Okay, I'm using my 10 amp current shunt range here. so the burden voltage is really low. Okay, so it's me.

It's the power supplies output in five volts. We're actually measuring the input here at 4.8 but if we actually want more precision on our current here, we switch over to our milliamp range which has going to have a much higher burden voltage. Wow Look at it Now it the input voltage power supply is still output in five volts, but the input to our actual module which is what we care about is drop down to low four point, one volts. and yeah, the input current.

So yeah, we can get the input power and the output power. That's still fine, but our input voltage is varying and that's generally a variable. We do not want to vary varible. In that case, the variable is the wrong term.

We want a fixed input voltage a no and input voltage. That's what they said these. The converters are typically specified at. So as we start to record our values like this: I might start at the highest current.

It doesn't matter, the lowest highest doesn't matter what it is and we're gonna do it in decades. So we might go 100 milliamps in 10 milliamp steps down to one milliamps and then we'll go under. That will go nine hundred micro amps, 800 micrograms, etc. down in decades because we're going to get a decade graph.

So the two key parameters here are our output current which we that is the x-axis of our graph. but we also want V in to be fixed. So we're going to have to go in tweet knobs over here in tweak this with our tongue at the right angle to get input voltage at our fixed 5 volts every time. So yeah it's you know, got a tweak a few things.

Oh well, it's not easy being green and I'll show you that here. Even if we've got no output current meter, we're just relying on our drop on our wires here to get our 5 volts on our output here. I've had to tweak this up to five point three one at our naught point one amps 100 milli amp output current which is the first one that we want to measure and then if we change our output current to say you know 20 milliamps or something, you'll notice that. Bingo! Our input voltage has changed, so we need to get in here and hold our tongue at the right angle, tweak our knob down until we get that input.

So you've got to do that every time. but you know by the time you automate this thing, it's not that hard. I Mean it takes seconds to do this when you're sitting down. Go and bang bang bang bang bang, right? So I've gotten down to 10 milliamps and basically I've reached the limits of measurement precision on my electronic load here.

So I'm going to use my Keithley 2400 source meter my SMU to actually set the output the sink current cuz you can use this this con source and sink current. In this case I can set it to minus 1 milliamps which means sync 1 milli ampere current instead of source 1 milliamp if I actually put that 2 plus 1 milliamp and it actually output current from here if I set it to negative, it'll source it back in. but most people are not going to have an SMU and if you do have an SMU, you probably don't need this tutorial now. I Can measure with really decent precision I set my compliance voltage to 13 volts just above what I'm expecting on the output here, otherwise it'll load, It'll clamp the output and load it and my I'm now sinking one milliamp here I can put a current meter in series with that to verify that.

Beer generally just don't need that and I could put a current meter in series with that to measure that. I'm actually measuring 1 milliamp, but hey, it can do that for me. We can just go measure and it's actually this is what we've set and this is what it's actually measuring. So I can get really precise stuff.

This is real schmick bit of kit. So yeah, we can easily go down to micro amps and and measure with the utmost of precision. No worries. But I still got to tweak the input to get my input voltage.

Okay, I'm down at a hundred micro amps now 0.1 milli amps output current and we're drawing thirteen Point One Six milliamps input. and if I actually disconnect the output, you'll notice that it doesn't drop down by not much. We're basically down to the quiescent current of our DC to DC converter. so there's no point going another decade from 100 micro amps down.

This particular, our DC to DC converter is just not optimized for low currents. And if we plot our three and a half decades worth the data. Bingo, Look at what we've got here. Here's our characteristic curve for the digi Lent nine volt power brick the efficiency versus the output current.

Efficiency on the y-axis here from 0 to 100% and in the output current on a logarithmic our graph which is important I'll explain that in a second from not 0.1 milli amps or 100 micro amps right up to in this case, I went up to about 320 milliamps before the overcurrent protection actually kicked in. Now granted, this DC DC converter is only rated to 100 milliamps output, but I went beyond that because I wanted to show you how it actually tails off there, otherwise it wouldn't have been very exciting would it. So all our data is over here and it was easy to enter it in by hand. It takes bugger-all time.

Once you've got it written down, you know I've had thousands of points and to take you some time. but when you only got you know like 60 70 points or something like I've got here then it's bugger all really. And the VN has always fixed that. 5 volts you remember I always kept I always in tweak that knob until we had a fixed 5 volts input and the input current.

that's the one we actually measured and the output voltage pretty much remained our constant. There were a little few little changes there, and the output current of course was our fixed nominal output current rule, dialing in with our electronic load and then the power up. What? We just calculate that it's the voltage times the current / a thousand in this case to get milliamps instead of amps. And then our power output, of course is just the output voltage times the output current time and divided by a thousand Once again for milliamps.

And then our efficiency is just our output power minus our input power times a hundred to scale it to 100% And also we've got the that's not display that's power dissipation there in Watts. But this has given us now a great characteristic curve. Look at this. There's a nice little hump in there.

That's because of the curve fitting algorithm that it's used. I can change? That doesn't matter. So here's our data points here. and you'll notice that because it's a logarithmic scale like between 10 milliamps and 20 milliamps, here is a fair jump and we don't actually have any data points.

That's just go. Chose to do a logarithmic data plot instead of a linear or a data measurement instead of a linear data measurement. so you know it just so happens in this case due to bad luck. That and Murphy that the you know all the interesting.

the interesting drop here in this curve is between 10 and 20 milliamps where we didn't actually take any data points like that, but it's going to be a fairly linear fit. You're not something gonna suddenly see it when go up to 90 percent, you know, efficiency curves always pretty much look like this. They might have a few little you know ripples in there, but nothing's gonna suddenly. You know at say 1 Milliamps, it's not going to suddenly curve back up and go there unless it changes conduction mode unless it changes the way from pulse width modulation to pulse frequency modulation.

So as a converter here, you can see you know from basically 20 milliamps up to its normal rate at hundred milliamps output current, it's not too shabby at all. It's above 80% which is reasonable for a little brick converter like this. it's not the most efficient. It never gets over 90 percent at any point, so it's not spectacular.

But for a general-purpose little power brick like this, it's okay. And by the way, remember this thing actually gives out plus 9 volts and minus 9 volts as well I didn't load or didn't test the - Nyvold output so there's going to be some loss there. But under 20 milliamps, you can really see it drop off a brick wall here. And you know, even at 10 milliamps, it's sixty percent efficient.

That's not great. and down at one milliamp, you might you know if you thought I would just use this power brick to power my until? give me plus minus nine volts for my you know, a little Op amp that I need or something like that will realize it's only going to be fifteen percent efficient or one milliamp. It's just it. And it basically is like it: 100 micro amps.

It's like, oh my god, it's ridiculous. It's not optimized for low current operation, so in the tens of millions, that's where it's designed to operate. But anyway, there you go. We got that nice characteristic curve, but ah-ha we're not done yet.

You're saying Dave we haven't plotted this power dissipation well. Yes, I have I've made the graph a little bit nicer. And tonight here's our final graph that includes the power loss. So this is this red curve here.

and we I've inserted another y-axis on the right hand side here. so I've got power loss from zero to one point eight What's up here and on the end, the same efficiency over here. So I should have color-coded those if I was doing this properly I would have color-coded there right? Ax Y axes there orange and the other one over here blue and you know anyway, you'd fuss around with that if you're putting it in some report for management or something like that. Not that they never bloody read it anyway.

Ah goodness, don't get me started. Anyway, so this is the power dissipated in the actual power brick itself. and you can see up to the hundred milli amp rated current. You can see why they don't raid it for anything more than that, because after a hundred milliamps, it really, the power dissipation in this little tiny surface-mount brick really starts to rise.

So it's only like in the order of you know, not point to what's there. 200 milli watts at the nominal 100 milli amp output current. But as that efficiency drops down, the power loss must go up. You'll always see these things match.

It's just, you know, a basic math. you can't avoid it. And yeah, you don't want to be dissipating or what. For example, in this tiny little serviceman power freak, there's no heatsink on an olive.

What is a fair bit of power for that tiny little power brick? So it's yeah. it worked on my bench here, but it's gonna. it's not going to continue to work at those sort of power dissipation levels. it's going to be.

You know, dire temperatures are going to be up to 100-plus degrees and I'll soon fail. And sure enough, the overcurrent protection or over temperature or whatever protection they've got inside this thing actually kicked in at 320 milliamps. even though it's only rated for 100 milliamps there. So you know, so they're they're fairly safe.

There they are. They have rated that fairly safely at a hundred odd milliamps. B Maybe could go a little bit more, but I certainly wouldn't go. You know anything passed maybe 150 milliamps there.

But anyway, always stick to the specs. don't go over them unless you want to live dangerously and you'll notice that I've actually labeled this V in equals five volts because that's where you saw in some of those our datasheet. Once they'll have different characteristic curves for different input voltages than if we wanted to do that well, we'd have to go through and read, log all our data again for a different input voltage. That's where something like a more automated test jig would be very nice.

And if you had an automated jig yes, you could do much finer steps in there and measure much quickly and get you know much. you know more data field graph, but this data is more than enough to get our characteristic curves. so no problems there. But yeah, we can go in there and plot all sorts of parameters that vary on this.

You can have this vary with output load capacitance for example. So you could have V in equals five volts for and then have 10 different output capacitances or ten different. you know, whatever versus load and you can measure a multitude of different things. that depends what's important to you.

So anyway, there's our finished graph. It's beautiful. It's like a ball one. And by the way, if you're wondering how I got this our logarithmic graph, you can't ordinarily do this.

Let's go over to the Y-axis here and actually format the Y axes. and if we go into scale Y, of course, has a logarithmic scale, so you can choose that. but we don't want a logarithmic scale for our Y Okay, so that's all in there. It's It's no problems, you just take that.

but you can't do that for the X-axes if you've got. if you're using a standard line chart type, now, it does actually work. In this case, we can go in and format our X-axis and sure enough, it's got logarithmic scale. So we can just switch off the logarithmic scale and you could have it like that.

but that's not the traditional way to display these sorts of characteristic curves. They didn't traditionally use the decade based logarithmic scale, and because you can see why, you know all the interesting stuff is all just jammed, you know, right down here. Whereas if you choose the logarithmic scale, then it's yeah, it's It's much easier to see those sort of, you know, interesting art changes. so that's the point.

But you can only do this if you actually using a certain chart type. You notice that I'm using the XY scatter chart type. You have to choose XY scatter if you chose your regular line chart. And when in here like this, you'll find that there's actually no option in there.

Look, you can even reverse the direction of the data which flips it side to side, but you can't get that logarithmic scale the line chart by choosing a line chart type, which is what almost everyone chooses. For something like this, it assumes that you want a linear x-axis It doesn't give you the option anyway. that's in that LibreOffice I Believe they think it's the same in the OpenOffice and probably in Excel I Haven't used Excel for a long, long time now. I've been using LibreOffice but yeah, that's just a trap.

You have to actually choose that specific XY scatter mode. So there you go. little trap for young players. Anyway, that is it for logarithmic scale.

thank you very much. Look at that beautiful graph. so I Hope you enjoyed that. It's been like half an hour video, much longer than I intended.

but that is a step-by-step process with lots of traps for young players in there of how to get a characteristic curve of a DC to DC converter which you might have to do one day if you're designing your own or you've got one in this case without the aspects. and now hey, we have an efficiency curve for it. You're working welcome did you let. Feel free to use it.

So as always, if you liked this video, please give it a big thumbs up and comment and engage in all that sort of wonderful business and discuss it in the comments eevblog forums links down below. all that sort of Jess I hope you liked it Catch you next time you.