A DIY 6.5-Digit Multimeter Is A Lesson In Clever Circuitry

A multimeter is an easy prospect, right? Back in the day you could make one fairly easily with a decent panel meter and a set of precision resistors, and now a digital one can be had for throwaway prices from China.

But what if instead of a cheap-and-cheerful bench instrument your needs extend to a high-precision device, a really good multimeter? It’s a path [jaromir.sukuba] has trodden with his 6.5 digit multimeter project, and along the way he’s offered us a fascinating window into their design that should be of interest to any electronic engineer even if they never intend to build a multimeter.

The range selection network of switches and resistors, microcontroller, and seven-segment displays are universal to a multimeter design, meaning that there is nothing too special about them in a high-precision instrument except that here he’s using an FPGA for timing.

Where the meat lies in this project is in the ADC and its associated voltage reference, and for that he takes a surprising turn. Instead of taking an off-the-shelf ADC part from one of the usual manufacturers, he’s created his ADC from scratch using op-amps, and to understand why that is the case he takes us on a journey into the world of dual-slope integrating ADCs. These circuits are very well explained in a 1989 HP journal article (PDF, page 8), and are a clever design that measures the time taken to charge and discharge a capacitor from the voltage to be measured and compares it to the same time from the reference voltage.

The beauty of it comes out in the HP article, that the mathematics of the charge/discharge cycle cancel out any effects of the analogue component values, allowing the much higher precision of the reference and the clock timing to dictate that of the reading. We look forward to seeing more of this project.

It’s surprising how few home-made multimeters we have on these pages, perhaps because of those cheap ones. Of the few we’ve had, perhaps this state-based Nixie one is most unusual.

29 thoughts on “A DIY 6.5-Digit Multimeter Is A Lesson In Clever Circuitry

  1. I don’t understand why people do build logs on hackaday.io, they are practically unreadable the way that hackaday.io is set up. If you click through long enough you can eventually get to the build log, but it is in reverse order(?!) and completely isolated from the rest of the project. Even instructables with all of its misgivings was better…

    Cool project though. Always fun to build things from junk laying around, it would be nice if you posted a followup to show how the performance is!

    1. If you think it is bad for the readers, you haven’t have to deal with the bug ridden editor. Sometime you loses paragraph just because you want to delete a picture. Other times when things were working, someone decided to change it again so your way of getting around the issues no longer works.

      1. Indeed. It’s about 30% less pleasant to use than hand coding html in Edlin.

        A couple of times, I’ve done a minor thing, a projectette that I thought others might find useful and thought “Shall I write it up on .io ?” then ended up digging out an ingrown toenail with a pocketknife instead.

      2. I learned long ago with it comes to submission forms. If your writing and article or anything much over a paragraph. Just write it in a word processor. Do all your proofing and editing in there. Then copy it to the form. It saves on the annoyance factor when shit bugs out. Which prevents the development of premature bald spots from hair pulling.

        1. “Which prevents the development of premature bald spots from hair pulling.”

          Right on! In my situations… speaking in profanic tongues more from the natural androgen enhancement I’m thinking.

          Going to add… make sure autosave is turned on also.

          Neat project and from my inexperienced studies… seems with the modern parts at hand or maybe needed with a few unobtainium parts or new old stocks or salvaged slowly dwindling… we can recreate equivalent and better than was done in the 60’s, 70’s and even 80’s. At least I have hope with some faith.

        2. If I use a word processor, I might as well generate a .pdf that preserve all the formatting and fonts. :P
          There are other hosting sites that are well maintained, autosave and not lose work.

          github.io is an alternative if you are already using github for source code.

      3. @tekkieneet said: “If you think it is bad for the readers, you haven’t have to deal with the bug ridden editor. Sometime you loses paragraph just because you want to delete a picture. Other times when things were working, someone decided to change it again so your way of getting around the issues no longer works.”

        That’s as bad as the HaD comment system here. It’s hosted at a separate address, bloated with layers of scripting, often drops replies to the bottom of the comment stack instead of inline with the proper thread (that’s why I always put “@Soandso said:…” in each of my reply comments now, so that if the reply gets orphaned, at least there will be a reference to who and what I am replying to). It seems sometimes my comment vanishes when I post it, but if I look for it the next day, there it is (if it went for moderation, there’s notification). Other times the comment will vanish when posted and never reappear. In all cases where the posts vanish, If I post it again I get a message that says, “you already said that” (or something similar). What a mess – but since it’s obvious by now this stuff is never going be fixed, I just put up with it.

        1. It was moderation. We’re human, and we’re slower on weekends when it’s only one or two of us behind the wheel.

          We still haven’t figured out the occasional orphaned replies. We’ve submitted a ticket, but because it seems to be entirely unpredictable, we can’t really help them debug. They probably think we’re all crazy!

    2. @jrfl said: “I don’t understand why people do build logs on hackaday.io, they are practically unreadable the way that hackaday.io is set up.”

      100% agree. I cringe when I’m about to head over to hackaday.io.

    3. I’m sorry to say that finding out the article points to a hackadai.io page, i wont follow the link. Good to know I am not the only one thinking this is a bad interfarce. ( pun intended)

      But to stay on topic: i have an old 3.5 digit nixie volt meter working on a frequency counter principle: counting pulses until a capacitor has charged.

      1. Having used the “editor” for quite a lot of projects now, I can assure you: you’re totally right.
        It’s just a pain just thinking about using the damn thing…

        I can’t even blame people for not reading (or even clicking) hackaday.io write-ups :(

  2. This is an impressive design. The use of metal package PNP germanium transistors is a fun touch.

    As is usually the case, the issue of the H P journal is an education all by itself.

    If I were to do something like this, I’d take a good look at sigma-delta ADCs, which might improve performance while making the multislope charge-discharge scheme unnecessary.

      1. Yawn. The whole project boils down to using a high res ADC on a chip, good reference, a display and a uP. To get the full advantage for the ADC, a proper layout and isolation from noisy source is important too.

        There was no documentation on calibration. It’s fine as a project, but I can’t take it as an instrument seriously

      2. LTC2400 is 24 bit Delta-Sigma, but also is $14 (half that in larger quantities). But dual-slope is really cool. Polystyrene is a great choice for the integration cap to get good linearity and low leakage. I think everything else, like switch resistance is cancelled out by dual-slope. HP is great. I will have to read that one. I just threw out 15 years worth from the 1970’s and 80’s!

        1. Hmm, yeah these days things getting really interesting with more options not just better.
          I see Analog Devices has the fairly recent AD7771 converter which is 8 channels of 24 bit each either single ended Or differential And with programmable gain too – reading all 8 channels at the same time ie Simultaneous sampling !
          Haven’t checked price yet but, don’t expect it to exclude much for the moderate to expert professionals, expect it to be more than the tinkerer hobbyist…
          Neato as ideal for all sorts of tricky or troubling data logging investigations of signal/effect relationships :-)

        2. Hmm, yeah these days things getting really interesting with more options not just better.
          I see Analog Devices has the fairly recent AD7771 converter which is 8 channels of 24 bit each either single ended Or differential And with programmable gain too – reading all 8 channels at the same time ie Simultaneous sampling !
          Haven’t checked price yet but, don’t expect it to exclude much for the moderate to expert professionals, expect it to be more than the tinkerer hobbyist…
          Neato as ideal for all sorts of tricky or troubling data logging investigations of signal/effect relationships :-)

  3. I have a pair of HP 34401As I bought for $235 and $250. The one for $235 arrived wonky. On examination I found a 1 cm spot of flux residue. I cleaned it off with isopropyl and it is my most accurate meter.

    Aside from being cheap, they are fully documented, so you can copy as much or as little of the circuit as you want. The 3478A and 3456A are also great deals at around $120.

    If you really want to learn DMM design, get one study it and then try to do better.

    Have Fun!
    Reg

  4. Oh man… I hope he pulls through – from the .io project blog: “Right now I’m typing this from a hospital ward, with great difficulities. The real lesson learned here is how is human life fragile and how important is to have your loved ones. Take care.”

  5. Well done, good project with layout oddly similar to the way I do things for quick protos, did a frequency counter that way circa 1979 though no FPGA, an Intersil chip and 1GHz prescaler in ECL. I also have heaps of HP journals and their design briefs too, nice sources of ideas and implementation. Thanks for links, will be looking into references too, tah.

    Health wise worthy dosing up on minerals of various types as many enzymes exploit good mineral fortification for immune function, immune system signalling and many other functional aspects for lymph equilibria. RDI out of date eg Cu in USA govmt states 0.9mg/day with the updated figure 0.1mg/Kg body/day. All best for recovery, health more important than anything, hey lets do the Klingon thing with the gods that caused widespread unequal suffering !

    Thanks for posting, cheers

  6. A very interesting device that will obviously be useful for industrial purposes. It is very important that it is as correctly scheduled as possible, so that people do not go around this much. A speedy recovery to the author of this multimeter and would like to see some measurements for reliability. And so, it looks good, it is clear that the person is interested.

Leave a Reply to Chris Maple Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.