Observations on the sorry state of commercial DMM offerings

      Electronics, Workshop

Gossen Metrahit Energy – die “Benz S-Klasse” of DMMs.

Digital multimeter (or DMM for short) is one of the most ubiquitous tools around, found in every electronics professional’s or hobbyist’s tool kit. Yet, one can be surprised by the fact, that in the boiling sea of portable device innovations, DMMs had stood as islands of stubborn stability: a rather expensive Fluke DMM purchased in 2013 is not substantially different from its siblings sold in 1995. Even electrical kettles enjoyed more progress.

Let’s take look for a second on what are considered “high-end” DMM features – those that impose a considerable unit price mark-up (this, surprisingly, applies equally to all breeds of DMMs, from Chinese “no-names” to posh Gossens; the actual prices differ by the order of magnitude of course, but the trend is the same):

  1. 16 bit per sample precision (aka 40000 counts in DMM jargon). Most DMMs, including expensive ones, only do 12 bits.
  2. Dual channel acquisition – simultaneous voltage and current with power calculation (a very handy ability when debugging power supply issues).
  3. Data logging at kSPS speeds and computer uplink interface. Granted, DMM is not a replacement for datalogger or scope, but this sort of features are really almost free to implement in a modern design.

Even at a superficial glance it is clear that so-called “high end” features can trivially be implemented on a US$25 worth microcontroller board. But, of course, it takes more to build a proper DMM than just picking an MCU board.

It so appears, that things are not much better on the “analog” side. For example, barely anybody (apart from Gossen, apparently) employs proper 4 terminal current-sensing resistors for current measurements. Similar issues apply to input protection: resettable (or not so much) fuses, MOVs and open-air sparc gaps are all the rage, with SMD gas discharge tube (GDT) being a fancy extra, found in some high end units. Apparently, TBUs and Trisils are yet to to find their way into the DMM input protection circuits.

I hold an opinion (which I developed after blowing a couple of DMMs trying to fix an ancient, tube-transistor hybrid Tektronix scope) that surges of various sorts are not an exceptional condition, but reality of life for any decent DMM. Thus, I would greatly prefer a protective circuit made of fast acting semiconductor components, over a “high-end DMM” MOV/GDT one (both MOVs and GDTs tend to degrade somewhat every time when barely triggered; why bother living like that? ;-)). Protection circuits found in cheap DMMs are better be left uncommented, for public decency reasons.

An issue of ergonomics, or, rather, lack thereof is also quite glaring. Most DMMs are controlled by rotary selector wheel; sometimes it only selects the measurement mode (good!), but more often it selects both measurement mode and range, so one has to constantly click through random number of positions to switch modes (very annoying at times). I recollect using a “radio button” operated handheld DMM some 20 years ago, but I have not seen any of those in ages. At any rate, good DMM must have range selection decoupled from mode selection, and it must be operable by one hand while hanging in the air (this is achievable with very limited subset of available models).

And what about functional response times? It so appears, that people seriously discuss and evaluate such things as apparent measurement delay, auto-ranging sluggishness and time it takes for the DMM to sound a “short-circuit beep” when testing for continuity. Not only response times on these and similar functions are perceivable in most DMM units, they are sometimes measured in full seconds, even when expensive units are concerned. How a relatively expensive electronic device manufactured in 2013 manages to achieve such long response delays, escapes my imagination (continuity testing and auto-ranging should take no more than few sampling periods; at any sort of decent acquisition speed those should be in imperceivable millisecond range).

Instead of keeping on with the rant (I suppose, the point is more or less clear already), I’d rather toy around with an “ideal DMM” concept, as I see it. I may even decide to build one – I wonder, whether this can be made into a kickstarter project (whatever the electronics, but I won’t be able to handle a neat, ergonomic plastic casing on my personal budget).

  1. Acquisition tract should be capable of 16 bit, simultaneous voltage + current channel sampling at 1MSPS with some more internal ADC + DACs connected to the front-end circuit. This will make the “dream DMM” into a decent data recorder and give it a bit of basic scope functionality (hard to resist this particular temptation, eh?). But more importantly, such set-up will enable all kinds of active probing – impedance measurement during continuity testing comes to mind.Indeed, It was never that apparent to me why DMM manufacturers keep pairing continuity test function with diode voltage drop test (I know why it was done in the first place, like 50 years ago, but why stick with it indefinitely?). I’d rather prefer an impedance measurement instead, broken into resistance and perceived capacitance/inductance (possible to measure and display with fast sampling “dream DMM”).
  2. Agilent U1273 with roughly 3.5″ OLED dot-matrix display.

    So, unlike normal DMMs with their “old-school” segmented displays, “dream DMM” needs to be able to display multiple value readings and may even need to plot some graphics. Thus, dot-matrix display is a must. Fortunately, sunlight visible, 3.5 inch (85mm on the wide side, to fit the unit size of Agilent handheld depicted, to take one as random example) LCD module featuring 480×320 resolution can be obtained from Alibaba for less than US$10, allowing for ample read-out space, even when multi-channel and graphing options are considered. In “traditional” mode such display can easily rival any segmented display in read-out readability and clarity, even from a distance.

  3. Fluke 117 featuring sane ergonomics.

    Mode selection in the “dream DMM” will stay with the edge mounted rotary selector wheel (I sort of like the design of Fluke 117, depicted). Still, I would add a couple more push buttons, which in combination with largish, dot-matrix screen will facilitate faster manual range selection and such. Unlike most manufacturers (who probably enjoy doing this circular traces on the pcb for selector switch to slide over in open), I would go for a sealed, integral SMD selector switch, possibly even endless and non-contact encoder type (with selected function indicated on a display). “Dream DMM” is expected to do all of its actual front end mode selection with integrated semiconductor switches.

  4. ”Dream DMM” will run of a “square”, rechargeable and easily accessible lithium battery and feature an USB slave interface, both for charging, control and data transfer.
  5. The usual feature pack (RMS voltage, peak detection, frequency counting and whatever else) is to be implemented in software on the MCU.
  6. Most importantly, everything should be fast: modern, “feature-creep blessed” measurement equipment is most often unbelievably annoying to anybody who remembers how fun it was to work with old analog stuff (not fun in the sense of measurement proper, but in the sense of everything happening instantly, without the need to wait a second for a voltage scale or timing base change).

From the look of it, such “dream DMM” can be economical at end user prices of around US$200, which is still cheaper than almost any “brand name” DMM on offer. Of course, I assume that there are no firmware development costs directly associated, because of open source magic. There are many unhappy electronics guys around who happen to know how to do embedded programming.