Chasing PPMs? Put that cell phone away!

[CalibrationGuy]

So I wanted to take a few pictures of our 732 array, and was reminded of something I would tell all my lab techs. PUT the stupid phone away! Here's what happened to a measurement I was doing to test a newly acquired 732a with our most stable reference (also a 732a).

Those peaks occurred when I moved close to take a picture of the "flatness" of the test. So yes, these measurements can be affected by your (and your cell phone's) presence.

Picture...

TomG.

[Andreas]

Hello,

Not only mobile phones. Also (even unused) USB-cables near measurement setup
can be a source of unexpected measurement deviations. E.g. here:

https://www.eevblog.com/forum/metrology/lowest-drift-lowest-noise-voltage-reference/msg3738565/#msg3738565

with best regards

Andreas

[EC8010]

And switch-mode power supplies to laptops. I've twice chased red herrings before remembering to unplug the laptop's power supply.

[ITArchitect]

Ah, chasing noise can be fun sometimes.  I found a similar issue with EMI coming from a wireless phone charging stand.  Pulled my favorites scope apart thinking I had issues.  Since then, all mobile devices have been banned from the lab. 

[mendip_discovery]

What would be these PPMs you be talking of? Surely you mean µV/V.

My MD was harping on that I should be doing TikTok videos, this in front of the 17025 assessor. He didn't notice my comment about TikTok users being mostly kids and creepy old men. I feel the need to draw 4 red perpendicular lines, some with blue in and some with transparent ink.

[CalibrationGuy]

Yes, of course, the joke being that a part per million in the 10Vdc, or 5Vdc references that we are building and/or testing is a tangible thing that can be acquired and/or kept! Hence calling them ppms in this context.

TomG.

[exe]

Is this because of the phone, or your presence? Would it make a difference if you put the phone in flight mode?

[CalibrationGuy]

If you use shielded cables, your presence does not usually affect the measurement unless you start touching things. Then, all bets are off. Yes, when the cell phone reaches out via radio or bluetooth, etc., that will be a larger disturbance than the EMI emitted by their new, powerful CPUs and also the inverters powering the displays, etc. It's just easier to leave them in the next room. Remember, the interference decreases by the square of the distance.

TomG.

[dietert1]

The Fluke 732A is a design from a past when cellular phones weren't that abundant. One could even call it a legacy device. Would be interesting to which degree their revisions  (B and C) solved the problem.

[CalibrationGuy]

Perhaps a savings grace is the fact that all the old stuff was built like a tank with lots of metal encasing everything, which would help somewhat. I'm not going to put my phone near my 732a reference gear to find out, though!

TomG.

[David Hess]

Test equipment used to be available with an "EMI" option for extra shielding.

In extreme cases, I have used bulkhead connectors and coaxial cable to keep RF out of sensitive signals.

[edpalmer42]

When I'm making sensitive measurements I always use double-shielded cable - preferably braid & foil rather than double braid.

I also collect the data from the equipment via GPIB to a computer that I access remotely via VNC.  I never go closer than about 2M from the equipment.

[Andreas]

The Fluke 732A is a design from a past when cellular phones weren't that abundant. One could even call it a legacy device. Would be interesting to which degree their revisions  (B and C) solved the problem.

The Fluke 732B specifies <0.18V/m as limit in the manual. See screen shot attached here:
https://www.eevblog.com/forum/metrology/emi-measurements-of-a-volt-nut/msg2684070/#msg2684070

A cell phone can generate easily 100-300 V/m when you put it into a cookies box and call it from outside.

I recommend metal shields (tinned steel) laying on the desk to shield somewhat the environment from the connection lines.

with best regards

Andreas

[CalibrationGuy]

The Fluke 732A is a design from a past when cellular phones weren't that abundant. One could even call it a legacy device. Would be interesting to which degree their revisions  (B and C) solved the problem.

The Fluke 732B specifies <0.18V/m as limit in the manual. See screen shot attached here:
https://www.eevblog.com/forum/metrology/emi-measurements-of-a-volt-nut/msg2684070/#msg2684070

A cell phone can generate easily 100-300 V/m when you put it into a cookies box and call it from outside.

I recommend metal shields (tinned steel) laying on the desk to shield somewhat the environment from the connection lines.

with best regards

Andreas

I have an ISO17025 accreditation. I bring this up because our auditor was old school and a stickler for detail. When I showed him the 732a and the Fluke 5520a, he asked me to verify the EMI field strength and demonstrate that it was below Fluke recommendations. We had to get a new meter to prove that the EMI was below the threshold. Many labs have these meters precisely because you can never fully know every source without some means of detection. Also, if you rent space like we do, you don't always know what's in the walls and above the ceiling tiles.

TomG.

[David Hess]

When I'm making sensitive measurements I always use double-shielded cable - preferably braid & foil rather than double braid.

I use double shielded cable when making sensitive receiver measurements, or RF will leak around attenuators, but have not needed it to keep EMI out of DC measurements.  I still often use it because the connector fastenings are more reliable.

[Kleinstein]

Perhaps a savings grace is the fact that all the old stuff was built like a tank with lots of metal encasing everything, which would help somewhat. I'm not going to put my phone near my 732a reference gear to find out, though!

TomG.

With cell phones  and Wifi there is now more EMI from higher frequencies. A simple shield can work well for the lower frequencies, but it only needs a relatively small slit to let the higher frequencies in. Old style shielding may not be made for the modern high frequency environment. Modern PCBs tend to be multilyer with a ground plane - this can effectively act like a shield too, just look less impressive.

Cell phones can from time to time send rather powerfull pulses (e.g. to synchronize / reconnect to base)  - enough to be audible with some audio gear. A measurement may well miss those rare pulses that only happen a few times per day.

Whatching out for EMI problems is defintely a good idea. In this context I wonder, why there is no common fiber optic computer interface with sensitive instruments.

[CalibrationGuy]

@Kleinstein: I built a poor man's fiber optic interface from a project I saw in I think Radio Electronics (yes, I'm that old). It was RS232 to infrared led through a fiber optic cable back to RS232.

Also, you are correct. Despite the massive metal enclosures, these devices also had big slots everywhere to dissipate the heat.

TomG.

[EC8010]

With cell phones  and Wifi there is now more EMI from higher frequencies. A simple shield can work well for the lower frequencies, but it only needs a relatively small slit to let the higher frequencies in. Old style shielding may not be made for the modern high frequency environment. Modern PCBs tend to be multilayer with a ground plane - this can effectively act like a shield too, just look less impressive.

I repeatedly failed EMC compliance testing due to a 0.1mm wide slit that let 400-600MHz in. Once spotted, a tiny piece of copper foil inside fixed the problem. Self-adhesive copper tape (sold cheaply to gardeners for deterring slugs) is your friend; put it over all gaps. I now do all my testing (mostly audio frequencies) inside earthed biscuit or tobacco tins - if it's good enough to keep food fresh it's good for EMC. Beware that many tins are epoxy coated and need an explicit electrical bond between lid and tin. External connections are made via chassis-mounting BNCs. And that includes power. When I'm really nervous, the power comes from 12V lead-acid batteries also in earthed biscuit tins. All my BNC cables have a clamp-on ferrite at each end to assist common mode rejection. Screening close to the DUT or the source works best, hence the use of ground planes.

I've done a number of experiments that responded to my body heat in the room and to draughts.

And if you're really unlucky, light is a problem. It took me four redesigns before I was able to measure the I-V relationship of an amber LED down to single-digit fA because ambient light easily causes an LED to produce nA of current. The only thing that stops light is metal. Black PVC tape etc attenuates but doesn't stop. Insulated BNCs let light in. Try putting a powerful torch on one side of a thick wooden door - the results will surprise.

[pdenisowski]

Cell phones can from time to time send rather powerfull pulses (e.g. to synchronize / reconnect to base)  - enough to be audible with some audio gear.

This was a big issue with GSM, but I haven't noticed audio artifacts with 3G and higher cellular technologies .... although that could be my hearing getting worse with each new generation of cellular technology 

For anyone who's either nostalgic or too young to have ever heard it, an audio sample is attached.
 

[HighVoltage]

Best way I found to see if there is any unwanted disturbance, hook up a Keithley 617 electrometer and measure a 100 G resistor. If that measurement is stable, your lab is clean.

The worst disturbance I had once was from a neighbor that had a broken power supply in his server rack.
https://www.eevblog.com/forum/chat/43-mhz-burst-noise-every-63us-in-the-lab-(how-to-find-the-source)/

[tggzzz]

Cell phones can from time to time send rather powerfull pulses (e.g. to synchronize / reconnect to base)  - enough to be audible with some audio gear.

This was a big issue with GSM, but I haven't noticed audio artifacts with 3G and higher cellular technologies .... although that could be my hearing getting worse with each new generation of cellular technology 

You will get artefacts with all pulsed radio systems, i.e. non-CDMA systems. It will audibly be worst with a TDMA system where the pulses are in the audio band, e.g. GSM at 217Hz.

Equipment has been forced to accommodate such pulses, by better shielding and/or electronically filtering out offending pulses (e.g. a 217Hz notch filter).

Obviously older equipment may be more susceptible, since that threat was never envisaged.

[dietert1]

..
Whatching out for EMI problems is defintely a good idea. In this context I wonder, why there is no common fiber optic computer interface with sensitive instruments.

Just saw an ebay offer with "Advantest R8240 Digital Elektrometer + Omron Z3GB optical GP-IB Link". The fiber isolator seems to exist as a product, yet rare. Don't know how they get the supply but remember that i once modded a GPIB socket, using one of the Gnd pins to take +5 V for an adapter.

Regards, Dieter

[CalibrationGuy]

Yes, I see that they exist. If you have the money, here's one: https://www.artisantg.com/TestMeasurement/84017-1/National-Instruments-GPIB-140-Fiber-Optic-GPIB-Extender OOF, the price starts at $2,300, I think.

TomG.

[dietert1]

Is that the price for a pair of those?
No, one should make that. e. g. using two small FPGA boards and two fiber transceiver modules. Except the GPIB bus hardware interface will be a bit nonstandard to make the connection simulate a bus cable. It is possible, see the isolating I2C drivers from TI. They also have open collector outputs on both sides.

Regards, Dieter

[CalibrationGuy]

The price is for a pair of transceivers and fiber optic cables, but still.

TomG.

[dietert1]

With DIY one could save one GPIB interface using an existing USB to fiber interface on the PC end. This could already serve to separate the measurement setup from the control computer, its displays and whatever is  connected there. At amazon.com i found a USB to fiber ethernet interface "StarTech.com USB to Fiber Optic Converter - 1000Base-SX SC - MM - Windows/Mac/Linux - USB 3.0 Ethernet Adapter - Network Adapter (US1GA30SXSC)" for $233.10.
Remains the fiber ethernet to GPIB controller on the other end. We have a NI GPIB-ENET100 and i could try to use it with a "10/100/1000Mbps Single Mode 1&4 RJ45 Ethernet to Fiber Optic Media Converter j" for $42 at ebay. The DIY could be as little as making quiet supplies for the two adapters on the GPIB end.
A better price and less work than DIY from scratch.

Regards, Dieter

Edit: Even better, the 42 buys a pair of ethernet to fiber adapters, so one would use ethernet on the PC side and save that 233 USB to ethernet adapter.

[ftg]

Cell phones can from time to time send rather powerfull pulses (e.g. to synchronize / reconnect to base)  - enough to be audible with some audio gear.

This was a big issue with GSM, but I haven't noticed audio artifacts with 3G and higher cellular technologies .... although that could be my hearing getting worse with each new generation of cellular technology 

You will get artefacts with all pulsed radio systems, i.e. non-CDMA systems. It will audibly be worst with a TDMA system where the pulses are in the audio band, e.g. GSM at 217Hz.

Equipment has been forced to accommodate such pulses, by better shielding and/or electronically filtering out offending pulses (e.g. a 217Hz notch filter).

Obviously older equipment may be more susceptible, since that threat was never envisaged.

In the land mobile radio world there is also TETRA and DMR. Both are timeslotted and the handheld terminals can have output powers between 1W and 5W.
So far worse than what GSM ever had. Thankfully these are rare.
Unless your sensitive equipment needs to exist in a place where there might be users with those close to it.
Say folks from various public safety agencies.

As for cellphones in general.
I was once characterizing the noise performance of some low noise amplifiers and binning op-amps based on their popcorn noise.
Could not have a cellphone or a bluetooth speaker in the same room with me.
Read a decent amount of books while waiting for individual measurements to stabilize.

[pdenisowski]

In the land mobile radio world there is also TETRA and DMR. Both are timeslotted and the handheld terminals can have output powers between 1W and 5W.
So far worse than what GSM ever had. Thankfully these are rare.

Great observation.  Here in the United States we have (of course ...) our own timeslotted version of this: P25/APCO.

I guess the important takeaway here is that shield / screen rooms and boxes are still the way to go if you're making sensitive measurements.

Incidentally, I had a customer once who told me he LOVED working in a screen room because it gave him an excuse not to return anyone's phone calls:  "Sorry, was working in the shield room all day and didn't get your call ..."   

[dietert1]

Meanwhile we got 2x tp-link MC100CM fiber to copper adapters and a 20 m multimode cable and it's working with the NI GPIB-ENET/100. Of course these fiber interfaces, their power supplies and the GPIB interface all have switched mode converters inside. A cheap endeavour, yet needs more work.

Regards, Dieter

[miro123]

Meanwhile we got 2x tp-link MC100CM fiber to copper adapters and a 20 m multimode cable and it's working with the NI GPIB-ENET/100. Of course these fiber interfaces, their power supplies and the GPIB interface all have switched mode converters inside. A cheap endeavour, yet needs more work.

Regards, Dieter

Few years ago I switched my home lab to optical fiber and LIFePO batteries. Later on rearranged the whole heme network. It is 80% fiber now.  I have used those TP- link converters in the past - instead of adapter I used lab PSU set to 9Vdc, later on attached to 3S - Latium battery

[exe]

I switched my home lab to optical fiber and LIFePO batteries

How often do you charge your batteries?

[miro123]

I switched my home lab to optical fiber and LIFePO batteries

How often do you charge your batteries?

Once per few moths. When BMS  reports that. The battery pack is 5S 240Ah LiFePO

[dietert1]

Meanwhile i found that the tp-link media converter is easy to run without it's buck converter if one feeds a 3.3 V about 200 mA linear supply. And it has a metal case.
The NI GPIB-ENET/100 is worse in that respect. It contains two LM2595S switching converters, one for 5V and one for 3.3 V, both with open ferrite inductors. Its enclosure has a metal base with a plastic cover.
Therefore i am studying a STM32H7 based ethernet solution - also avoiding the NI proprietary protocol and their drivers.

Regards, Dieter