DMM overclocking!

[eccerr0r]

I had some thoughts earlier in an earlier topic bit curious if anyone went ahead with it?

Anyway a lot of the older DMMs have clocking that ultimately determines how fast it comes up with a reading. Because of this clocking, DMMs can take several hundred milliseconds to make a reading, some DMMs up to a half a second to refresh.

I don't have any of the really modern DMMs that are much faster, but perhaps an overclock of 500ms refresh rate (2 samples/second) to 250ms (4 samples/second) would make it nicer. I suspect that jitter would worsen. Also not sure how the LCD will respond to faster updates, it may just blur out or something.

Anyone tried, and what issues did you run across? What brand/chipset meter was it? Was it worth it?

[stj]

Re: DMM overclocking!

the D/A sampling circuit will go unstable if you run it too fast - it takes a minimum time to get a value from the input

you will also dim the display a bit as it multiplexes faster - but you can compensate for that.

[eccerr0r]

Re: DMM overclocking!

There is a S&H but it's just a 1-bit DAC, there is a bit of settling time with the comparator, which would just add to, as said, the jitter. Or maybe it will read a little low for each sample?

Anyone actually tried for fun? Wonder if the reference design DMMs are more likely overclockable... the custom ones may be running at their limits already?

[stj]

Re: DMM overclocking!

update rate is a selling point, so they probably already push them,
although the datasheet for the dreamtech 0660 says clockspeed is tied to battery current.
so it's a compromise

[eccerr0r]

Re: DMM overclocking!

Well jitter is another thing that would be affected by a higher clock rate, but this could also be binned. Easier to make all devices along a line the same but some could potentially take an overclock just like any other electronic device...

[stj]

Re: DMM overclocking!

well you can try - every chip is gonna be different
it wont be fun if you have to re-calibrate the meter though - specially if it's using an eeprom

[eccerr0r]

Re: DMM overclocking!

Well integrating ADC DMMs won't have eeproms, they don't have computers!

Then there's the thing that a whole lot of DMMs end up using canned chips like ICL7106's... granted the DMMs I have, this is not the case...

TBH I've never had a SA DMM. Closest I've gotten was my PWM bench PSU project that read the output voltage/current using the microcontroller ADCs and as expected, they're fast, got ~3 digit resolution updated (on a HD44780) faster than I can read.

[redwire]

Re: DMM overclocking!

There's a limit to what your eyes and brain can read, as far as update rate. The noise/jitter/flicker of the A/D also affects readings, it can update fast but bobble back and forth, never steady. And worse if you have more digits like 4-1/2.
I like my old Beckmans 310,3030,HD110 etc. because they are very fast and with their manual ranging and low jitter, I can take 10 measurements within seconds. On ACV it has fast RMS averaging. The multimeter keeps up with what I'm thinking, where to probe next etc. as I work through some gear.
Newer DMM's are clunk clunk clunk auto-ranging and then they have to settle in, firmware averages things, 10x slower to work with. It's gross, waiting for the reading.

On the Wavetek/Meterman/Amprobe BDM40 (ICL7135 LED) I did increase the clock from 100kHz (2.5 readings/sec) to 150kHz (~4 readings/sec) which was not easy because it uses a tuning fork crystal but I found some 149.75kHz and I remember trying 192kHz it was too fast to read the display easily (~5 readings/sec) and you need a multiple of mains frequency for the clock, to reject AC hum. The integrating ramp resistor I had to change too.
It was an experiment, so it didn't act like some slow antique.

[eccerr0r]

Re: DMM overclocking!

The microcontroller ADC that I hacked up I was getting probably 8 readings per second or so, at least as far as I recall, because that was the screen refresh rate. The ADC sampling actually occurs much faster than that for other reasons, but the HD44780 is only refreshed 8 times a second. Either way, yes it's way too fast to comprehend unless the voltage/current were stable.

The meters I was thinking about hacking was my BKP 4½ digit, which also uses an ICL7135 driving an LCD. I also have a Dana 4½ digit LED which I'm thinking about hacking too, alas this is an autoranging unit.

I have two 3½ digit LD110/LD111 based LED DMMs that seem so much faster in updating than 7106/7107 DMMs which started the wondering. Alas most of my DMMs are ancient antiques, my newest one died on me (autoranging 3¾ digit) so I'm trying to just use my acquired-used 4½ digit units.

(Incidentally, if 1999 count is 3½ digit, 19999 count is 4½ digit, 3999 count is 3¾ digit, so what are 5999 and 3199 count units?)

[redwire]

Re: DMM overclocking!

I find it two things - the display update rate, and the A/D filtering/averaging.
If you have fast updates but slow averaging, you are waiting looking at numbers slowly climbing up to the final value.
If you have fast averaging but slow updates, you are waiting for that last number just to show up.

I seem to like 4 updates/sec, like Fluke 87 although its averaging is a bit sluggish. The Beckman/Wavetek just say "response time <1 second". Older multimeters are usually 2.5 updates/sec which seems slooow.

I'm running some antique multimeters because they're made in USA and new crap from taiwan or china can't beat them for reliabilty. It's so bad I wanted to make an LED DMM from scratch lol but can't find Hycon IC HY3131 to buy anywhere.

[eccerr0r]

Re: DMM overclocking!

Well on older DMMs the averaging is done by LPF, and the clock doesn't really affect the filtering though it does affect what gets read. Will get a lot more trash reads if the clock isn't matched with the filtering, which is something that will need to be looked at with the overclock.

I think my Fluke 77 isn't too bad for update rate or even autoranging. My Dana and my HP are dastardly slow at autoranging though I think the HP beats the Fluke in update speeds. I don't remember how bad autoranging was on my Mastech... On the other hand, the startup selftest of the Fluke 77 sometimes gets annoying. Then again the HP and Dana (and probably my Viz and Tek), as all LED meters, are specced after warmup period of 30 minutes or so, just like the 7107.

Then the quandary...if the all the main silicon parts are made in China and assembled into a complete unit in the USA, is it a made in the USA?

[redwire]

Re: DMM overclocking!

Older Fluke 77 is 2.5 readings/sec and 77-IV up is 4 readings/sec. I find 2.5 is too slow. For something like DCV, the LPF is matched to the display update rate. If you overclock, the LPF can be 'speeded up'.

Fluke is just an old "brand" now owned by mega-conglomerate Fortive. Milk the old cow, priority is profit.
FTC finally did an update to Made in USA Labeling Rule, it was old 1997 and updated last summer 2021/07. Some loopholes got closed.

A lot of Fluke product is from Brymen, so I would expect some changes if they get audited. I think some final assembly/test/calibration still done in Everette WA which supports "the good undergoes its last substantial transformation in the United States" for some products. The trend for profit is outsourcing as much as possible to china and no longer using union labor in the USA.

[eccerr0r]

Re: DMM overclocking!

hmm. thought the fluke 77 was significantly faster than my b&k 4½ digit, which I thought was a reference ICL7135 design. Need to locate full docs for the b&k, seems almost impossible to find free docs for it... data is available for everything but my meter...

[redwire]

Re: DMM overclocking!

BK Precision 2831D uses ICL7135 and 100kHz xtal, it's similar to the Wavetek BDM40. I changed the xtal to 149.75kHz (all I could find near 150kHz) and pin 6 resistor from 100k to 82.5k and it works fine. 192kHz xtal/33k was 4.8 updates/sec and was too fast for my liking.

[eccerr0r]

Re: DMM overclocking!

Oops. I was wrong, it uses an ICL7129 IC. Anyway it looks like I'll need to find a crystal unfortunately, else I will have to hack up a PLL or something. Probably similar crystal requirements, but I didn't look at which was installed, probably the 100KHz or maybe 120KHz.

One datasheet says it should be good to 360KHz... heh...