Multimeter strange behaviour.

Introductions, general chit chat and off-topic banter.
Post Reply
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Multimeter strange behaviour.

Post by offgridQLD »

Digitech QM1575 / Extech EX540

http://www.extech.com/instruments/produ ... prodid=530

It's a 40,000 count meter with a basic acuracty of ±(0.06% reading + 2 digits)
When I first got it I hooked it up to a friend's calibrated HP bench meter and the accuracy was well within spec.

Though lately I have notice some strange behaviour.

If you take a dc measurement and note the value. Then take the same measurement but reverse the polarity (swap the leads around the other way) you get a higher reading. It's very consistent always higher. 1.6153v correct polarity and negative polarity 1.6174v. Sure it's only 2mv but all my other meters give the same results in both directions Image

Another odd one I noticed when I 1st got the meter so it's always been there is as you hit the range button and scale down the count for less resolution it has a significant effect on the readings that doesn't make sense to me. As in the roundings don't add up.

Starting from full resolution and dropping down
1.6153v
1.608v
1.55v
1.0v

Yes the meter has fresh batteries.

Kurt
Last edited by offgridQLD on Mon, 09 Feb 2015, 06:28, edited 1 time in total.
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Multimeter strange behaviour.

Post by offgridQLD »

Also Just to confirm my understanding of the accuracy potential of the meter.

±(0.06% reading + 2 digits)

So in the up to 4v setting. Say I was to measure a reference voltage of 1.0000v.

My understanding is. I could be +/- 0.06% so it could potentially read high 1.0006v then I need to add 2 digits so 1.0008v high and the other way low 0.9994v take 2 digits is 0.9992v.

So the accuracy range potential is 1.0008v - 0.9992v difference of 0.0016v or 1.6mvImage

The only confusion around it is the (+2 digits) when you're working out your +/- range do you take two digits when its measure low and add two digits when it measure high extending the accuracy error potential by 4 digits?

Kurt

User avatar
acmotor
Senior Member
Posts: 3605
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Multimeter strange behaviour.

Post by acmotor »

The roll over error when reversing the leads is not uncommon.
At +-.06% in your example of reversing leads is around .12% (.06% each way) so basically within spec. in my thinking.

The +2 digits is +-2 digits and is just that, +-2 of the display digits
Having said that, your observation on changing range seems extreme.
I would expect only +-1 digit discrepancy and a quick check of a few meters I have confirms this.

Just another thought is... was it pure quiet DC e.g. just measure an AA battery or something. Sometimes meters can make errors depending on how they handle AC components on DC signals.
iMiEV MY12     110,230km in pure Electric and loving it !
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Multimeter strange behaviour.

Post by offgridQLD »

Yes just a AA battery. The range changes has always been something quirky that I didn't like. When we tested it a few years back we did have a variable power supply and compared it to the HP bench meter (can't remember how many count that was but it had a lot of zero's. It was good as could be expected at a variety of voltages.

Kurt
Last edited by offgridQLD on Mon, 09 Feb 2015, 11:18, edited 1 time in total.
User avatar
weber
Site Admin
Posts: 2784
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Multimeter strange behaviour.

Post by weber »

Hi Kurt,

I agree with what acmotor says. And yes, you are correctly interpreting the specs. The simplest hypothesis that explains both your polarity-reversal and range-changing results is that your meter has a fixed offset or bias of -8.5 counts (which is clearly out-of-spec), in addition to its specified inaccuracy of +-0.06% and +-2 counts.

A "true" value of 1.61635 V and a -8.5 count bias would give a reversed reading of (actual reading in parenthesis).
-1.6172 (-1.6174)
and range reduction readings of
1.6155 (1.6153)
1.608 (1.608)
1.53 (1.55)
0.8 (1.0)

So you can see that the readings you got are all within +-2 counts of these hypothetical readings.

I put "true" in scare-quotes above because it's only true in the sense of having the offset removed. It says nothing about a possible scaling error (what the 0.06% is about).

What do you read on each range with zero volts in, i.e. when you short the probes together? I'm expecting a minus sign and the low digits as 07, 08, 09 or 10. If so, you can short the probes and press the REL button to zero this out every time before taking voltage readings.

But it's not acceptable. Maybe there's a trimpot inside or some secret combination of buttons on powerup to access some digital calibration mode, to zero this out permanently. If it's under warranty, you should get it fixed.
Last edited by weber on Tue, 10 Feb 2015, 12:20, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2784
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Multimeter strange behaviour.

Post by weber »

It occurred to me that if I was designing a user interface to zero a multimeter offset permanently (so it would survive turning the meter off) I'd probably use a long press on the REL Button. Unlikely, but can't hurt to try it.

Alternatively, I might make it accessible through the PC software that come with the meter.

But really, techniques for automatically eliminating systematic offsets have been known since the pyramids were built. Image Basically by internally reversing the polarity and averaging the results. So there is probably no user interface for this because there is no excuse for it to exist in the first place.
Last edited by weber on Tue, 10 Feb 2015, 12:52, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Multimeter strange behaviour.

Post by offgridQLD »

I just tried your suggestions.

shorting the two probes showed 0.0005v fluctuating to 0.0006v . I then hit REL and it was showing zero with probes shorted.

I then measured the AA battery again.

1.6156v in correct polarity
1.6166v in reverse polarity

Though trying the next test (cycling down the range while measuring the voltage) once I push REL and set it zero I can't cycle down the range until I take the (REL) off. Image

Kurt

Last edited by offgridQLD on Tue, 10 Feb 2015, 13:14, edited 1 time in total.
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Multimeter strange behaviour.

Post by offgridQLD »

Measuring a different battery this time.

I get 1.5969v correct polarity and 1.5990v reverse polarity
After zero (REL) I get 1.5973v and 1.5985v.

kurt
User avatar
weber
Site Admin
Posts: 2784
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Multimeter strange behaviour.

Post by weber »

offgridQLD wrote:shorting the two probes showed 0.0005v fluctuating to 0.0006v .
Were they really positive? If so, that blows the simplest hypothesis out of the water. If they were really negative, then we only have to complicate it a little. See below.
1.6156v in correct polarity
1.6166v in reverse polarity
Clearly you've omitted the minus sign for the second result here, so perhaps you did for the shorted-probe readings above too.
Though trying the next test (cycling sown the range while measure the voltage) once I push REL and set it zero I cant cycle down the range until I take the hold (REL) off. Image

Darn.

I note that a -5.5 count offset is also consistent with the range-change readings +-2 counts. If the "true" value is 1.61565 we get
1.6151 (1.6153)
1.610 (1.608)
1.56 (1.55)
1.1 (1.0)

This 5.5 count offset (which is still out-of-spec and therefore warrantied) only explains about half the discrepancy in the polarity-reversed reading. But the remainder can be explained by different gains (scale-factors) for positive and negative voltages. And, as acmotor showed, both gains can be within spec. This can occur when dual-slope integrators use separate positive and negative voltage references.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2784
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Multimeter strange behaviour.

Post by weber »

BTW. I asked "What do you read on each range with zero volts in?"

On the latest hypothesis I'm expecting:
0.0004 to 0.0007
0.004 to 0.007
0.04 to 0.07
0.4 to 0.7
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2784
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Multimeter strange behaviour.

Post by weber »

For those following, I can answer my own question since Kurt brought it over tonight, along with his new acquisition (I'll let him tell you about that). With probes shorted, on every range it read -4 counts, sometimes flicking to -5. So -0.0004 volts on the 4 V range and -0.4 volts on the 1000 V range.

Since it's meant to be zero, what percentage error is that!!!? (only joking). It's out of spec as a counts error, not a percent error.

[Edit: Added minus signs to voltages where required]
Last edited by weber on Wed, 11 Feb 2015, 04:59, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Multimeter strange behaviour.

Post by offgridQLD »

The "new acquisition" Well there is a small...long story to it.

After feeling a little disappointed that my best Multimeter. The one I turn to when I want accurate and reliable readings. Perhaps wasn't so accurate or reliable now. My friend with the bench meter that I first tested it against is now 2000km away in Melbourne. I really didn't have a reliable reference to even attempt to start tweaking pots on this meter.

So with accuracy and calibration on my mind I got side tracked watching a few Utube videos. Dave from the EEVblog had posted a video on(DIY digital MM calibration) Going into detail showing his calibrated analog 5 count adjustable voltage standard hooked up to a HP3478A 5 count bench meter. I was very impressed how the HP MM followed the reference down the the last digit and continued to follow it to the last digit as he turned the knobs. Thinking to myself at the time how nice it would be to have a quality bench meter like that. On call at any time I want to check my hand held meters against. Giving me some confidence in readings.

I started watching another EEVblog video. This time Dave had upgraded his HP3478A to a bigger and better HP unit (as if a bang on accurate to 5 digits meter wasn't good enough) This new unit was a HP 3457A 6 digit bench meter. He did a bit of a teardown and overall it was very impressive.

So out of curiosity I put HP 3457A into ebay search. I didn't get one hit in AU. A few came up in the US but boy they were expensive ranging from $300 for a broken dirty looking one showing error messages on the screen. With the average around $400-$500 and a nice one with some of the optional features close to $800 - $900 . That US$ + about 100 - $200 shipping from the US Image More searching info showed that they were around $3000US new in the late 1980's a lot of money back then.

I had more or less given up on that Idea. Then I did a quick search on gumtree. To my surprise Bingo one came up and it was even in Brisbane. I sent the guy a text to work out a time to come take a look and do some negotiating. Also asking if it was working. I got a text back saying yes it did last time I used it. Then I got a text shortly after saying they just plugged it in and it went pop and smoke came out the back and it stinks. It would only be good for parts. I started thinking is this a scam was it ever working Image I said ok well I will come take a look and see.

On arrival the guy gave off a reasonably trustworthy vibe. He did say it had worked 10 years ago and today it did for a few seconds then pop. It sure did stink like it just happened. So we settled on just $30 cash for this (Good for parts HP 3457A meter) What could I lose. Some one might want it for parts.

As I was driving past anyhow on my way home. I gave Weber a text saying I had purchased something interesting (as I knew he had just purchased a reliable quality fluke MM for accurate DC voltage readings) I thought having a 2nd more knowledgeable opinion with the cover off would be great. So it wasn't long before Webers kitchen table was a mess of screwdrivers, tools and the aroma of burnt electronics Image Fuses all looked good. Nothing on the BCB to point the finger at. Using the sniffer dog tecniche it was apparent the strongest signal was coming from the mains plug cable inlet socket. Looking closer you could see a small stain of yellowish gunk on the edge of it.

I said perhaps something is inside that plug thats fried it looks a little iffy. Then webber mentioned it could be the EMI filter inside that socket. We could just try bypassing it and see what happens. I said why not and we hooked up a temporary mains cable. Webber was in the kitchen at the other end of the extension cable ( hmmm was that a tactic to distance himself from the explosion!) I hit the power button then Weber hit the GPO switch. A moment of silence then a small chime and a lovely LCD display showing full digits slowly counting down to zero Image

Weber went and got a 40Ah calb/winston cell and his Fluke meter while I sat there thinking boy was that it? Do I now have a awesome bench reference meter for $30 + a new inlet plug. We tested his fluke and it was ever so slightly so off. The cover come off the fluke and it was tweaked to match the HP (within the flukes resolution). Weber seemed confident in the HP.

$10 got me a new inlet plug and as I didn't have any tools at the house in Brisbane. Webber was kind enough to help me out out getting the old riveted socket out and the new one installed. I plan on using this as my reference from now on and will have a go at trying to dial in the suspect 40,000 count hand held meter perhaps tomorrow.

Webers fluke meter test.
Image

All back togeather and working. Self test function shows (all ok) calibration count in memory shows 36 calibrations.
Image

Accuracy specs. Rule of thumb is you need a meter 10x more accurate than what you are trying to calibrate. Looks like we have that and some with the HP.
Image


Anyone with a suspect hand multi meter or one you just want to test its still ok. I will bring the HP along to The Brisbane AEVA meeting next week and a reference voltage and you're welcome to use it to check against. I haven't had the HP calibrated but the general consensus is they dont drift much and it should be a reliable.

Kurt







Last edited by offgridQLD on Wed, 11 Feb 2015, 02:44, edited 1 time in total.
User avatar
Adverse Effects
Senior Member
Posts: 1103
Joined: Sat, 01 Jan 2011, 03:30
Real Name: Adverse Effects
Location: Brisbane

Multimeter strange behaviour.

Post by Adverse Effects »

nice score
User avatar
Johny
Senior Member
Posts: 3732
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Multimeter strange behaviour.

Post by Johny »

Well done. Good hunting skills.
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Multimeter strange behaviour.

Post by offgridQLD »

Inside the suspect Digitech QM1575 meter. 915MHZ communication card unplugged to expose more trim pots. Red and black arrows pointing to all the trim pots.

Image

Results after tweaking the bottom left trim pot. I dialed back the HP resolution to match the max resolution of the hand held meter.
Image

The biggest issue with trying to tweak the pots while powering up the the QM1575 is the 9v battery is removed when the back cover is off to gain access to the trim pots. So the only quick way around it was hold the 9v battery against the contact pads on the PCB with one hand and the other held a screw driver. While holding the meters LCD screen at a mirror to follow the reading (backwards) ....very frustrating. I think If I am going to have a more serious attempt at sorting it out I will need a fly lead and a 9v battery plug and solder it temporarily to power the meter.

Hunting now online for some more info on the meter.

Kurt





Last edited by offgridQLD on Wed, 11 Feb 2015, 12:28, edited 1 time in total.
Post Reply