Hi I Was just doing some experiments on the new Eevblog 121 GW multimeter and hey, I Thought this might make an interesting video. It's to do with the true Rms converter chip used inside this thing. It's the Analog Devices A DAT 436 Now, it's a relatively new chip compared to the ancient Ad 5, 3, 6 it is that's been used in you know, every true Rms multimeter since, like the nineteen late 1970s I Think it is. That chip came out crazy.

Anyway, it's a very popular true Rms, our converter chip that's used inside all sorts of modern multimeters and we're using it inside this one to get a decent true Rms performance of this thing. But hey, there's some little limitations that I wanted to test out and we're actually going to be doing something a little bit naughty with this chip. We're actually going to be using it outside of its nominal operational voltage range specified in the datasheet, and normally you know you wouldn't do this in a bit of production kit like this new multimeter. but we have actually are concerned with analog devices that it will actually operate properly outside of this range and the company who's doing this multimeter with me.

they're doing some testing and I thought hey, I'd Just check it out as well. So it's all hunky-dory even though it's outside of its nominal voltage range. so you can actually do this. you know, as long as you've either characterized have fully characterized it yourself to be you know fairly sure that it's going to give you the results you want in production and/or you've I cleared it with the manufacturer because sometimes they're very conservative on the datasheet and it does actually operate outside of it.

So whatever, performance art spec it is just fine. But hey, they didn't want to put it in the datasheet for a little insert a reason here right? They didn't want to push the limits or whatever. So I thought we just do some more testing of that here and see what's happening. Now the AAT 436 I'll put the datasheet in here, operates or has a mirror operational voltage of plus minus two point four volt supply rail or in a single-ended configuration which we're basically are going to be doing inside here.

That's four point eight volts minimum. but hey, this multimeter actually uses our four double A batteries to power this thing. So four double A batteries gives us a nominal battery supply voltage of 6 volts. But of course when you're designing a battery power product, you want to actually maximize the usable capacity in that battery so you want to have the are cutoff voltage of your battery as low as possible.

Now in this case, of course if we dropped it out at four point eight volts that'd be wasting probably half of the kepada and capacity in the batteries I haven't checked but it would be a horribly high cutoff voltage of 1.2 volts per cell and that doesn't include the dropout voltage of the regulator. But hey this chip, you could actually run it directly off the batteries. But of course, if you've got a nice stable supply, you can guarantee the performance. It's not going to change over the supply range.

so 4.8 volts is like we don't want to piss away half the capacity in our batteries or whatever it is. Look up the characteristic curve for the batteries. I've done many videos on that, so we need to operate this thing down lower, preferably a nominal cutoff voltage might be. You know, a decent one might be say 1 volt per cell.

So we're looking at working down a 4 volts. So it's not point 8 volts lower than the datasheet value of this chip, but we actually want to operate it lower than that. We actually want to put in a three point six volt voltage regulator because we're going to three point six volt voltage regulator for other stuff and we want to operate it down that low and we're talk to analog devices and they said yep, you know it will actually operate down at three point six volts instead of the minimum four point eight volts in the datasheet. So what I'm going to do here today is: I've got a powered from an external our supply here I've actually taken the back cover off I won't show you too much but back cover off there.

And normally you need a spring, the battery, the batteries in the back, and the spring terminals. actually you know go down on that. So I've just sold an some wires in there that allow us to hook up an external power supply there and we can adjust that and operate our meter down to any voltage we like. so that allows us to adjust our power supply and check its performance over any particular range.

So I've got a siglent cig gin here that allows us to generate some A/c signals and I've got the keysight our three double four, Seven Zero A which is a beast of a 75-inch a seven, half-inch seven ARF digit multimeter. So we're going to use that to compare the readings. and basically you know if we adjust the different waveforms and our 121 GW varies from our reference keysight here, then you know we know to investigate further. But basically if we change the voltage range right down and we test various waveforms and it matches something like the keysight here.

hey, no worries. Now the AE 8436 inside here is actually powered from a low dropout voltage regulator in this case. So rather than like butcher the circuit and everything else and bypass that and put the existing 3.6 volt voltage regulator on there I thought I'd just drop the battery supply like this and actually let the voltage regulator drop out and then we can lower the voltage. But of course audios are famous for oscillating.

If you get them below that dropout voltage, it might be say, naught point one volts for say a five on voltage regulator. So I've been just going to hook the scope up here to the rail AC couple it there I'm only on that lantern. 200 millivolts per division should do it. Actually, let's go down to 100 something like that to see and I'll just drop it below its dropout voltage and just make sure it doesn't oscillate and then that's just an easy way to do it.

There it is, if sorry, jump up because it's when I change the rail like this. You should probably see it jump around a little bit. There we go. So I'm dropping out, regulators dropping out now and it's following.

It's basically tracking that voltage rail down. So up, four point, Three Point four. Okay, I've got it - three point, six volts. Now on the battery input here and the output.

So here's our input. What we got there. We go. We've got three point, six volts and then the output of the regulator.

three point five, seven. So there's just a little drop there. There's not much current, there's a bit of drop. It's basically track that down very nicely so we can get away with doing that.

The oscillations not going to affect anything, but that's worth checking and if you thought that was a problem, maybe in it like in real performance measurements, you wouldn't do that. you would go to the effort to bypass. I Don't want to butcher the meter. So yeah, that's going to be good enough for today.

So what I've got here is a 400 Hertz waveform generating. That's a typical figure and nominal life specification figure for a multimeter. We can use a Kilohertz or whatever or the frequency range of this I You know it's going to go up to many, many tens of kilohertz I Think the 80 8436 is like capable of a megahertz if your optimism, which if you're optimized, it's not going to be that hard and that high in this particular case is going to be sub 100k or something like that I believe. So I'm generating to one volt RMS signal and tada we have one volt RMS so it's actually working just fine down at that three point 6 volts.

No worries at that normal frequency. so you know as a first-order path that's working just fine. And of course, if I adjust it, it's going to take time to settle down. You know, when you heard just in the supply, you expect things to well jump around because the averaging caps and everything else have to settle.

You're changing the whole upsetting the applecart there. But there we go. That's basically nominal six volts, although that's voltage regulating there. But yeah, we can take that down to three point six volts, no worries.

And you know it's a little bit lower than what it was up there, but that's just a calibration thing basically. so that is well within specification, so it's already looking very promising. But hey, what we want to do is actually check it at our full scale here. and it's basically bang on to the key site because this is a few a 50,000 counter meter so we want to test it at its full range at full scale range.

No worries, I mean that's practically bang on. And then if we check it down at naught point, one volts will beginning right down there at the lower signal levels on the same range, then it's basically pretty much bang on as well. So beauty good. But of course we're using a pure sine wave here and that's a relatively low our crest factor waveform.

Now the crest factor is actually defined as a ratio, and it's a ratio of the peak value option to touch the ratio of the peak value Bloody Touch screens the peak value of the waveform to the peak value of a sine wave divided by the RMS value of the sine wave. So a sine wave actually doesn't have a crest factor of 1. It's actually got a crest factor of you Guessed at that 1.41 For that, you're a figure that you're actually used to, which is the peak ratio. So and in fact, if you put a square wave into this thing, then a square wave is actually a crest factor of 1.

Because it's peak value, it stays at that peak all the time. So the peak divided by the RMS is actually the value. It's a definition of a perfect crest factor. So you might think our square waves are bad.

For true RMS converters, they're actually not. They're real. You know, as good as you get in terms of that crest factor. and depending on the are true.

RMS Our converter chip used you might you know you'll typically have a crest factor limitation I might tell you and maximum crest factor is 4 or you might read the datasheet of a multimeter and it tells you a the RMS values are only in guaranteed for this spec up to a crest factor of 4 or 8 or something like that. The 8080 for 36 is specified. A specifications are given up to nominal specs are given up to a crest factor of anything up to 10. So you know that's some pretty horrific type our crest factor.

So we'll try something. we're just a little bit more Oh shall we? So we'll go into waveforms here we'll go into I Mean we compare Noise is a pretty bad one, but we'll go into Rd R and we'll actually do a sine X on X which you know it's mostly low for most of the period. Then it's got a big spike. So I like switch mode power supplies are a common example of this that will have bad crest factor waveforms and stuff like that.

And incidentally, the Ad 8436 actually has a specific pin to add an additional averaging capacitor like a higher frequency average capacitor for those shorter peaks, as well as having a larger average capacitor value. So this is a pretty neat chip. that's why it handles crest factors are quite well, and certainly the Eevblog meter has both of those caps built in to handle the higher crest factor waveform. So this is going to be a reasonably high crest factor waveform you might typically get.

You know, overshoots like that, and you know spikes and things like that. So hey, look right down already. Look at that. Look at that 22 point width.

basically. Being on. let's go back up to our amplitude here and go. One volt.

Oh, we can only do peak to peak here. Okay, we can't do the RMS anymore. But yeah, it's tracking that. No worries at all.

And if we go up to ten volts peak to peak, that it's look. It's pretty. You know it's not quite bang on, but it's pretty darn close as well within spec. so it's operating quite well with large crest factor signals down at a three point six volt ship's supply limit well under that datasheet value of four Point Eight.

So if this chip is very conservatively specified, and by the way, our three point six volts will be the the highest operational voltage part inside this meter. So this allows us to set our dropout voltage our low battery detected voltage not much above that three point six, so you might set it to have to look at the data sheet. So the regulator's the low dropout regulators used in this, but you know it might be typically 50 millivolts, hundred millivolts or not much above that. So three six volts nominal would give a low battery warning indicator of not point nine volts per cell.

And that's pretty darn good. You're using up, You know, vast majority of the energy inside those four double A batteries, so that's quite good design you don't want to be, you know, pissing away that battery capacity. And of course we can now select all sorts of other waveforms. Here was that exponential rise? There we go.

Doesn't matter what our type built here and this has got all sorts of building math functions and you know you can have engine so view like a cardiac pulse. There you go. We can choose a a cardiac pulse and bingo There we go. So it measures a cardiac pulse.

no problems with all at all. seem to be working very well. Are basically the only variable left. really.

that would be a concern would be temperature. Really? Because you know a load powered from batteries, low dropout voltage regulator, everything else you wouldn't worry about. You know, power supply rejection ratio and you know other system stuff like that's not really relevant here. so probably temperatures.

The performance over temp would probably be the only one left. So how does it perform over temperature? I'm glad you are. So we've got the older thermal chamber here, which we haven't seen in quite some time and I'll just like ramp it. you know, ten, Fifteen, or maybe twenty degrees if I can and basically just see if it matches so we'll give it a ball.

I'll come back. Just pretty boring to watch. So at the moment we've got a value of not point seven, seven seven. let's call it whatever, Seven Seven Seven seven and which is fairly close to the keysight there and of course the key sites out of the chamber.

so that's our control. It's not going to change and we'll see if there's any effect with increased temperature. But of course here we're not concerned whether it's an increase or decrease in temperature. We're just concerned with the temperature.

Delta ie. Delta just means change the change in temperature so you know it doesn't matter. We just want to get a measurement on the multimeter up front. at two different temperature points.

you know five degrees might tell you you know ten starts to be reasonable. You know a 10 degree C change might be a standard control or something like that. so we'll ramp it up at least 10 and see if there's a difference. If not, then I wouldn't be too concerned.

I Mean this is not some ultra full-on professional testing there to take weeks and weeks and weeks where we fully characterises over the entire operating temperature range and everything else. I Just want to see if it makes a difference because I Trust that the people at analog devices when they say yeah, it was the 3 point 6 volts. We just don't put that on the datasheet and you know I trust them that they're You know they know their part very well and they know it's going to perform at that. So one other thing we've got to test for is frequency range as well.

So I've taken up to 5 Kilohertz here. Where down it? Sorry, this is a bit complicated. We're down at three point six volts again and we're getting one point double O five there. So and we can take it up higher in frequency: 10 Kilohertz 20.

There we go. The by the way, the keysight ones got their 300 Kilohertz or something 20. We're still hanging in there, but now we're starting to get see. You know, a few a couple of percent out.

You know we're starting to get a a few percent. that's 36 Kilohertz 37 I Mean if you want to talk about the minus 3 DB point then it's going to be massive. You know? So we're up at 65 Kilohertz now. But anyway, that's not the point.

Okay, so what we're going to do is there we go. If you can, probably go down to the point 707 point at a hundred and twenty odd Kilohertz or something like that. but what we want what I'll do is I'll just tweak it to 1 volt precisely on there and just use that frequency point as a ballpark. What do you think? sounds good? There you go.

17 kilohertz Now time to ramp the temperature up and we're up to 39 degrees. And on the AC performance at our 39 degrees C we're looking at, you know, point 3 ish percent change or something like that. So yeah, not much. and if we change it back down to the 5 Kilohertz reference, it's almost spot-on isn't it basically ramped up to 38 degrees now, which is good.

15 degree differential on there and they end. By the way, the back is still off so the airflow is going directly over ship. it's going to. you know, there's little thermal mass inside there in terms of the chip, so it's going to change almost in real time with the chamber basically, and it hasn't drifted at all.

My sure. if you can see that yep and glare on there yet. No point in double seven, Almost double seven. it hasn't changed a smidgen in fifteen degrees.

So I'm going to call that a win. So there you go. I Hope you found that video interesting in using a chip in a commercial application outside of its nominal specification and ordinarily as I said, you wouldn't do this. but there's good reasons why we're doing it because we want to get that low dropout.

you know, voltage on the batteries and everything else, so you know it's pretty important and that's the chip that we want to use in that application. Yeah, there are other options, maybe, but you know that's the chip that we're going to be using here and it works. Just hunky-dory so I'm pretty darn happy with that. I'm going to do some more testing, but that's basically our confirmed that No problem whatsoever.

Yep, no worries. I mean, no problem in using that in this particular application. Now, that doesn't mean that the chip is fully characterized down to three point six volts, you know, at higher signal amplitudes which we may not be seeing in this and the closer you know it's full operational range which we're not using over and stuff like that it could. You know there could be traps in there for that, but you know for the particular application that we've got here at full-scale input voltage and at the low end.

So he measured full-scale and at the low signal level as well. No problems whatsoever, no problems over temperature, no problems with our crest factor. so I'll just do a some more experimentation. But yeah, that's that.

seems pretty solid. So analog devices were right when they say this thing works down at 3.7 3.6 instead of the nominal 4.8 in the datasheet. But yeah, you know, don't try this at home. Make sure you stick to the data sheets unless you absolutely have to go outside.

You have a very good reason to do it. What the traps of young players doing that? But anyway, hope you found it interesting. Catch you next time.