Calibration of measurement equipment

[mansr]

Here's an old, cheap DMM next to a new Fluke 289 with factory calibration and the built-in meter of a power supply:

Same meters measuring the mains AC voltage (the cheap one doesn't even pretend to be true RMS):

The cheap meter is off by 1.6% for DC and 2.6% for AC.

 

[solderdude]

At lower AC voltages the small meter will be off even more. The 'rectifier' is just a silicon diode with its knee voltage not compensated. (try to measure 2V AC for instance)
It's why the smallest range is 200V AC to 'hide' this issue.
The Fluke probably calculates true RMS by taking the waveform in account as well.
Mains voltages generally are far from truly sinusoidial, they are flattened on the tops due to rectifiers in other equipment.

I work with fiber optics (daytime job) and calibration is an even bigger problem there.
Plug in an optic cable a few times in a row and every measurement differs due to many reasons.

 

[DonH56]

Most DMMs spec some percentage +/- some number of digits so aren't super-accurate (and don't really need to be). I'm surprised at all the nice Flukes folk have; I never got around to getting one, just borrowed from work, too cheap.

People using DMMs to measure audio-frequency signals should be aware of the limitations others have stated, mainly limited frequency range and lack of true-RMS measurements.

@solderdude -- In a previous job I designed a VCSEL driver and ran into the fiber optic variability issue. So much depends upon the connection, including not just the connector itself but also the mounting/insertion of the cable and how the cable itself is cut. Easy to get a few dB of variance in intensity/loss...

 

[solderdude]

a VCSEL driver with or without monitoring diode ?
Fiber optics is a nightmare (when it comes to tolerances and repeatability) for us electronics designers.

Have a habit of using work measurement gear myself... Also only have cheapies but some of them I calibrated against a calibrated meter at work.
One of my cheaper meters has a full audio BW AC range where the Fluke from work starts to drop above 2kHz or so.

 

[Timbo2]

Most DMMs spec some percentage +/- some number of digits so aren't super-accurate (and don't really need to be). I'm surprised at all the nice Flukes folk have; I never got around to getting one, just borrowed from work, too cheap. ...

For Fluke meters used in the electrical trades they are frequently found on eBay. Lots of pawn shops take them and list them if the loan isn’t paid. As a result the second hand values are both consistent and higher than you would expect.

 

[mansr]

At lower AC votages the small meter will be off even more. The 'rectifier' is just a silicon diode with its knee voltage not compensated. (try to measure 2V AC for instance)
It's why the smallest range is 200V AC to 'hide' this issue.
The Fluke probably calculates true RMS by taking the waveform in account as well.
Mains voltages generally are far from truly sinusoidial, they are flattened on the tops due to rectifiers in other equipment.

What the hell, let's do some quick tests with a function generator.

Sine wave, 50 Hz, varying amplitude
1 Vpp: Fluke 0.3522 Vrms, Cheapo 0.1 V
2 Vpp: 0.7035 Vrms, 0.4 V
5 Vpp: 1.7595 Vrms, 1.4 V

Not even close.

5 Vpp sine, varying frequency
100 Hz: Fluke 1.7623 Vrms, Cheapo 1.4 V
1 kHz: 1.7601 Vrms, 1.4 V
10 kHz: 1.7620 Vrms, 1.3 V
20 kHz: 1.7608 Vrms, 1.2 V
50 kHz: 1.6196 Vrms, 0.9 V
100 kHz: 1.1440 Vrms, 0.4 V

The Fluke covers the audio range, whereas the cheapo doesn't.

 

[solderdude]

It's a fluke when a Fluke isn't priced as a Fluke and measures like a Fluke

 

[mansr]

I'm surprised at all the nice Flukes folk have

At some point you just get tired of using crap. Besides, even the 289 is cheap by audiophile standards, costing far less than some USB cables.

 

[DonH56]

I am usually swamped at work and so have little time or energy for tinkering at home. When I was younger I did more; now, meh, maybe when I can retire. I can afford a Fluke, and the nice Weller iron I'd like, just don't need them often enough to justify the expense. There have been a few things I would really like to piddle with using one of the 'scopes from work, and have permission to do so, but am a little nervous about bringing home a DSO worth more than my house.

@solderdude -- Mostly without the monitoring diode; I designed a (quad) driver IC that included analog and digital (ADCs/on-chip microcontroller) control and monitoring circuitry. Customers could dial in the driver using whatever monitor diode they wanted or it would directly monitor the VCEL's current and voltage. Also had a temperature sensor, natch. My contribution was the HF driver and some novel signal-shaping capability. It was ahead of its time and very well received but that was 2000/2001 when everything collapsed. There was a group of about a dozen designers working on various Rx/Tx chips and all but three were laid off. One was switched to another group and two of us had projects shelved "pending business evaluation". I saw the writing on the wall and took a job as Director of Engineering at a small R&D company.

 

[solderdude]

I quit the audio/video repair business at about the same time when I noticed repair business had become replacement business. Went from component to board replacements.

Neat (driver thing). I am in the optical (railway) sensor business now. My part is optical to electrical conversion. Mostly still analog.

 

[Blumlein 88]

I quit the audio/video repair business at about the same time when I noticed repair business had become replacement business. Went from component to board replacements.

Neat (driver thing). I am in the optical (railway) sensor business now. My part is optical to electrical conversion. Mostly still analog.

I'm not in either business. I did have some training on motherboard repair in the late 90's. That could be a complex thing. You had several books the size of New York phonebooks with diagrams for each motherboard model. Certain symptoms lead to certain checks with logic probes, clock checks with Oscopes and so on so forth. I was better than average at it at the time. Some repairs were so complex it took hours to find what needed replacing or to find it was too extensive to fix the board. Others it was quick and cheap to fix. By the time I was thru that training I asked the instructor how useful any of it could be. It was by then such that around half of repairs took so many hours to troubleshoot you could buy a new board for less money. I mentioned I expected boards to cost half as much in 18 months (Moore's law). To also be twice as complex making for more deep troubleshooting procedures So the only thing anyone would do is make sure its the MB and swap in a new one. Repair was about to become a bit of history I told him. Many times now it isn't a board swap, you just junk the computer if you can retrieve data and move on.

 

[JJB70]

At some point you just get tired of using crap. Besides, even the 289 is cheap by audiophile standards, costing far less than some USB cables.

In fairness you can buy an awful lot for less than the price of some audiophile USB cables

On calibration, that can become one of those iterative questions. When I worked in electricity generation my employer spent quite a lot of money providing a calibration lab to do all calibration, span, zero checks for instrumentation on-site. Within about 18 months they'd reverted to a single centralised calibration shop as the costs of keeping the instruments in on-site labs at each plant was prohibitive (a point that was made with some force before they spent the money in the first place).