A/D for bench multimeter

[throbscottle]

Having calibrated my multimeter using this home-brewed reference, I realised I really have a crappy meter, and would like to have a go at building my own bench meter.

I should be able to calibrate a 5 1/2 digit meter with the reference (+/- 0.5mv at 5v or +/- 0.052mv at 175.963mv). Okay I don't actually need this level of precision, I'm just a bit obsessive!

Can anyone recommend an a/d converter intended for this type of use? Or at least something obtainable and affordable to mere mortals I can add whatever I need to get the desired result?

 

[Sceadwian]

Something like https://www.electro-tech-online.com/custompdfs/2012/04/MAX110-MAX111.pdf would put you in the ballpark, you're going to have problems with your accuracy goal without extremely high quality precision resistors in the scaling network.

 

[crutschow]

How about being a little less obsessive and go for a 4 1/2 digit meter?


Of course you will still have a problem with achieving the desired precision of the scaling resistors as Sceadwian mentioned.

 

[throbscottle]

crutschow, it was that Instructable that got me from wistful to serious in the first place! Then I started looking at commercial bench meters and thought, why can't I go the extra digit?

Vishay do 0.01% and 0.001% resistors - I've already had a couple of used 0.01% ones off eBay about £3 each - new they cost twice that, so that makes an expensive divider! Looking at the data sheets, I see whilst the MAX110 is specified as 14 bit, the MAX1498 is specified as 20,000 count. for 5 1/2 digits I think I'd need 200,000 count or 19 bits. OK so 4 1/4 digits is much easier...

Actually, thinking about it, since my intention is that this thing is only worth building if I can include a lot of extra functionality, and I want to make it be microcontroller controlled, is there a reason I can't do a low-count/high count and add them together? Or am just out with the fairies now?

 

[throbscottle]

Huh, I found this thread seems very helpful...
https://www.eevblog.com/forum/projects-designs-and-technical-stuff/community-bench-meter-gt-adc-and-associated/

 

[throbscottle]

Okay so now I'm looking at
https://www.linear.com/product/LTC2415
and
https://www.linear.com/product/LTC1968 to get true rms

Getting excited now...

 

[unclejed613]

depending on how obsessive you are, you might want one of these to calibrate it.... https://us.flukecal.com/products/el...ctrical-standards/792a-acdc-transfer-standard

 

[Sceadwian]

If you can shelf your accuracy requirements and you want functionality take a look for multichannel ADCs that communicate over serial lines. You won't get the accuracy you're looking for but if you get one that can do say 100-500ksps you can build yourself a general purpose low speed multichannel scope. The cost for each extra bit of resolution you need can get very expenesive and you can use input decimation to increase resolution by oversampling somewhat.

 

[throbscottle]

unclejed613 - I'm not that obsessive!

Sceadwian - interesting idea. Having done my sums now I realise 5.5 or 6 digits is only worth doing if I'm going to have 1 microvolt resolution, which whilst I relish the challenge, just isn't worth doing as a hobbyist. It never occurred to me I could go for a multichannel scope (I have an ancient, single channel "generatorscope" someone gave me) so that's worth looking at, though I still want to build a decent bench multimeter.

 

[Sceadwian]

Having a 1uvolt resolution would be useful in some situations though. On a standard 50mv full scale shunt rated for 20amps, you'd have 4ma current resolution. If you can capture fast enough you can detail even small scale transients, but again you pay for every single last bit.

Building your own bench meter is more about defining what it is you actually want to do with it than anything else. Once you can define the parameters of exactly what you want to do, building isn't so bad it's just a process of determining components values and sourcing the parts.

A simple and accurate bench voltage meter would be very cheap and easy to build if that's all you want, every time you add features or increase resolution/accuracy you add cost and complexity. So start with the basics and add as you go.

 

[throbscottle]

Hmmm, food for thought. For a moment there I mis-read you as saying it's about defining who I am! I think there might be some truth in that! At the moment I'm after the most bang for my buck I can possibly get, with very limited bucks (well, pounds anyway). I'm guessing if I can build a very good voltmeter, it shouldn't be hard to base other functions on that. Reading a little more on the subject, it appears handheld 3.5 and 4.5 digit dmm's are using dual slope adc's, bench types are using delta-sigma. I'd never heard of either of these a few days ago!

Cheers
throbscottle

 

[Sceadwian]

That's primarily because of the advanced in cmos manufacturing in recent years, it's become trivial to throw processor speed at an application.

 

[unclejed613]

if you want to measure ohms with it, you add a scalable constant current source. with it at 1mA, for instance, a 1k resistor would give a voltmeter reading of 1V. so you would have a constant current source attached on one side to a 3V rail, the other side to the + probe and the + voltmeter input. the - lead and the -voltmeter input would be tied to the internal analog ground assumind (as is the case of most DMMs) a full scale reading of 2V, anything above 2k will give an OL reading. for the 200 ohm range you would set the current source to 10mA. for a 2 meg scale, set the current source to 1uA, etc... since you are building this, it's totally up to you if you want to add a milliohm scale (i.e.set the CS to 1A), but of course that's not a possibility if this is battery powered.

 

[throbscottle]

Okay, I've found a reference circuit in Intersil's an028.pdf including an ohms converter - it uses an FET as the current source - don't know if it's worth looking for something more stable - suppose I should. Been having a look at input resistance ladders. Voltage ones seem easy enough - current I don't really know what is considered standard for these things - I'm thinking a 2mA to 2A range with 9Ω, 0.9Ω, 0.09Ω and 0.01Ω, and then a separate input for high currents + appropriate range switching - depends what kind of resistor I can get for that! Does this look reasonable? I've gone with reed relays as the range selection switches since that's what Intersil suggest and I can't think of anything better. Hoping for 100M input resistance but wondering if it might be too noisy. Haven't shown 2000v and 20A ranges since it's only to show what's in my head...

I'm not looking to use the Intersil adc, but at least it's a starting point. Still think I'm going to go with the LTC2415 adc, LTC1968 rms converter, and some kind of microcontroller - something I'm only just learning how to use! Cost wise so far the resistors are the expensive part!