When a fine piece of lab instrumentation crosses your bench, you’ve got to do your best to put it to work. But even in the highest quality devices no component lasts forever, especially vacuum tubes. For some vintage instruments with vacuum fluorescent displays, that means putting up with less-than-perfect digits in order to get that sweet, sweet precision. Or not – you can always reverse engineer the thing and add a spanking new OLED display.
The Hewlett-Packard 34401A digital multimeter that fell into [qu1ck]’s lap was a beauty, but it had clearly seen better days. The display was full of spuriously illuminated dots and segments, making it hard to use the 6.5 digit DMM. After a futile bit of probing to see if a relatively easy driver fix would help, and with a replacement display being made of solid unobtanium, [qu1ck] settled in for the long process of reverse engineering the front panel protocol. As luck would have it, H-P used the SPI protocol to talk to the display, and it wasn’t long before [qu1ck] had a decent prototype working. The final version is much more polished, with a display sized to fit inside the original space occupied by the VFD. The original digits and annunciator icons are recreated, and he added a USB port and the bargraph display show in the clip below.
We think it looks fabulous, and both the firmware and hardware are on Github if you’d like to rescue a similar meter. You may want to check our guide to buying old test gear first, though, to get the most bang for your buck.
slick. like the look of dot-matrix even though it doesn’t have to be, and the higher-res elsewhere. makes it look like a cusom/oem display, rather’n just another raster-screen.
And the bargraph is cool, lots of other potential as well, a small log, ala scrolling oscilloscope comes to mind, but may be overkill.
because oled is waaaaay more ‘robust’ than VFD…….
But it’s not unobtanium, so you can keep replacing it… As often as needed! :P
As some owners of older consumer equipment with dead OLED displays might attest to…not necessarily true a few years down the line =(
So? He has the hard work done now. He doesn’t have to reverse engineer the meter again. So long as he saves all his notes then next time it should be short work to port his solution over to whatever is easy obtanium by at that time.
Not an eternal fix, but a fix that can be redone in 20 years when the OLED fails.I love the look of the VFD, but they do age out, albeit generally gracefully.
I like it. (This was one of my favourite meters, back in the day)
Agreed. Nothing is a permanent fix, but the hard work is done and any future fixes will be easier.
Actually that display isn’t an “aged out” VFD – the VFD is likely fine. However, these meters are notorious for the driver chip going bad, especially in the older versions, with the drivers going leaky and then you get spurious symbols illuminated, like in the article. The chip is unobtanium today, unfortunately (it is apparently some long obsolete MCU with integrated high voltage drivers), but there are some replacements if this happens.
There are also two variants of these meters as far as the display is concerned – the older ones use a different display board and there you are screwed if the VFD needs replacing. The newer ones have aftermarket VFDs still available from China (check eBay). I have fixed my own 34401A with a very faded display like this. Keysight also sells the entire display board sub-assemblies still but it is quite expensive (about $100).
If you want to fix your own meter, check the EEVBlog forums first – it has been discussed at length there. It may save you a butchery of an otherwise perfectly good meter.
Jan Ciger writes:
“..notorious for the driver chip going bad, especially in the older versions..”
Reminds me of a high voltage issue with scintillation counters & Cs137 circa 1980’s.
There were leakage issues in odd places, turned out IC epoxy at those times on some semiconductors slightly hygroscopic – so worth addressing driver pins ie. Given the age is comparable pethaps gradual ingress of moisture has affected operation.
Easy to fix:
Remove board, place in oven at suitable safe temp in presence of good dessicant, cool, coat with fresh PVDC or equivalent, cure at suitable temp with dessicant for appropriate period and whilst cooling…
Also reminiscent of the notorious problem with the Kenwood TS-440 transceiver displays. Why does everyone think that gooping the parts around the display drivers is a good thing, anyway?
Very nice. Bravo!
outstanding
top notch
I wonder if this would drop into other old HP gear. Someone did an amazing LED retrofit of the LCD on an HP 3458A a while back, but due to the heavy milling of the displays doesn’t seem to be cheaply reproducible.
Outstanding. This person just increased the value of the (excellent resolution) instrument on eBay 50% :).
I have test equipment that 45+years old and still works well. There really is some truth to “they don’t make them like they used to”. I have an DVM/frequency counter that uses _NIXIES_ that still works well, and with its original capacitors on top of that (HP5326B).
My Tek464’s are also both still working 100%. I bought a broken one for spare parts – 40$ of shotgunning with caps, bridges, and transistors repaired that, also. 80’s and older equipment GENERALLY (but not post-1980 TEK) uses mostly off the shelf parts, high-quality through hole circuit boards with thick plating and they were designed to be maintained…basically forever.
regarding: “There really is some truth to “they don’t make them like they used to””
Please let us not forget that there is a lot of junk from the past too, although most of that junk has been discarded by now, so most stuff we have got left of “the good old time” is the stuff that was great/perfect/wonderful or build like a tank.
The reason we think to believe that modern equipment is rubbish is simply because there is a lot of rubbish too, and the good stuff that does exist, hasn’t been able to prove itself yet. On the other hand I wonder how many people who own a HP34401 today, could really afford it when it came out.
At work I use the 34401 and I’m quite happy with it, I do want one for home to, but do I really need one (nope). So yet I’m stuck in a situation of bad comparison, comparing the (perhaps) overly priced HP/Agilent/Keysite stuff with my cheap handheld digital multimeter. Is ins’t build like a tank, but for home… it’s doesn’t matter because honestly, when do you really need those extra digits?
But perhaps the most important thing with a super accurate measurement device like this how many times do their owners calibrate it (and if they do so, calibrate it with what), so that they know those extra digits really mean something.
In my experience, with no bearing on this specific case, “good old times” usually get mentioned whenever a typical modern unit of [whatever] turns out to be of a quality / endurance spectacularly less than an equivalent typical unit from a few decades ago, usually in an apples-to-apples comparison. And it does indeed tend to happen a lot – I’m not saying there weren’t any crappy things made in the past, but the stuff that wasn’t explicitly crappy sure as hell was built with _some_ endurance in mind as opposed to today’s “just last until next year when the next model comes out since you’re basically renting all the stuff you think you ‘bought’ at this point anyway” craze.
When I hear of these fixes in precision devices I always wonder how much effect they have on that precision. Anything from loading the power supplies differently, to changing the thermal patterns in the device. With a 3.5 digit DVM I would have go at it and have fun, with a 4.5 digit DVM, I would add the caveat “be careful”. Just something to be aware of. You have a very sensitive device there so you need to think above and beyond normal terms.
If replacing the display is going to affect accuracy due to the load on the PSU, I’d say the power supply design is pretty poor. Most likely it is being supplied by a separate regulator to the analog stuff anyway. As for it affecting thermal stuff, I’d say there’s got to be far bigger issues with varying ambient conditions.
I understand and agree with your concerns in general, but in a well-designed device for stuff that isn’t touching the analog section, it should be a non-issue. Besides, how many of these devices are accurately calibrated anyway?
I would actually be concerned by display from the noise. It’s quite possible the VFD was spec’d to have enough shielding to reduce noise radiated or conducted back into the DMM. I have personally had severe issues with OLED displays in very noise sensitive designs. Because each pixel is emissive, column/row scanning inherently results in high current spikes.
Also, it’s much worse than you think. The 34401A is a lab grade, benchtop 6.5-digit multimeter. I believe in this case it was just set on a lower resolution mode to demonstrate its speed. There is no demonstration of accuracy or low noise at high resolution, which is the point of having a benchtop multimeter in the first place…
Fair enough, I didn’t consider noise. And an OLED would scan much faster than the VFD, so noise would also be at a different frequency which might make things worse too. I would expect most of it to be averaged out, as it will be at a far higher frequency than the meter samples at, but…
Nice post.
Yeah good post, good idea and nice implementation and yes like the bar graph. Towards end of vid though Display states DC when it’s low freq AC. I guess one can set the threshold where alternating polarity DC could be considered as AC and for some definable period/waveshape – maybe if a fundamental with negligible harmonics trip the display to state AC after some period of time. Hmm, all of course set by selection of measurement method as in coupling too into front end…
Everything is DC on a short enough time scale!
Gorgeous build here. I’ve got a function generator that has a lame LCD calculator type display that I’d love to upgrade. I was originally thinking I would just attempt to add a backlight to the thing, but projects like this really set the bar high.