Cabling, Thermal EMF, Impedance, and You!

[CalMachine]

I know there has been a thread start about what is the best material to use for cabling/wires for the lowest possible thermal EMF and noise.  But to my understanding, that thread was generally about DC applications.

I would like to start a thread about low freq signals and the best types of wiring/cabling for the most accurate measurements.

100 mV @ 1 MHz.....  Sourcing from Datron 4808 => Datron 4920(freshly cal'd 3 weeks ago by Fluke). 
with my ~1 meter silver-plated solid copper conductor, with gold plated spades = 100.5070 mV
(these materials were sourced from http://jswilley.com/Accessories.html)

with my ~1 meter RG213 N-type cable with a Pomona N-type => 4mm Banana adapter  = 100.4147 mV
(adapter is brass with nickel plating, I believe)


My first question is, at what frequency would/should you switch from using shielded twisted pair => coax?

[dacman]

At 1 MHz, differences in cabling, connectors, and connections are very noticeable.  RG-213 has about 30 pF of capacitance per foot.  The 4-wire silver-plated solid-copper wire from J.W. Solutions is really designed for 4-wire resistance measurements (using dc reversal techniques).  How were the gold-plated spades attached?  Were they crimped or soldered?  If soldered, was cadmium (low thermal) solder used?

Shielded twisted pair is good for lower voltages and lower frequencies.  This type of cable from Pomona has a rating of 300 Vac max.  I would put the frequency at 100 kHz max.  Coax should have a higher voltage rating.  Shielded balanced line (or twin lead, not twisted) would have a higher voltage and frequency rating than twisted pair (I believe due to lower capacitance).

Thermals are really a dc phenomenon, but the 4920 is dc coupled.  I most usually use coax and high quality (usually Pomona) connectors, and don't concern myself with thermals a great deal, when doing ac measurements.

At 1 MHz, cabling and connections are going to have a much greater effect than any (dc) thermals.

[CalMachine]

At 1 MHz, differences in cabling, connectors, and connections are very noticeable.  RG-213 has about 30 pF of capacitance per foot.  The 4-wire silver-plated solid-copper wire from J.W. Solutions is really designed for 4-wire resistance measurements (using dc reversal techniques).  How were the gold-plated spades attached?  Were they crimped or soldered?  If soldered, was cadmium (low thermal) solder used?

Shielded twisted pair is good for lower voltages and lower frequencies.  This type of cable from Pomona has a rating of 300 Vac max.  I would put the frequency at 100 kHz max.  Coax should have a higher voltage rating.  Shielded balanced line (or twin lead, not twisted) would have a higher voltage and frequency rating than twisted pair (I believe due to lower capacitance).

Thermals are really a dc phenomenon, but the 4920 is dc coupled.  I most usually use coax and high quality (usually Pomona) connectors, and don't concern myself with thermals a great deal, when doing ac measurements.

At 1 MHz, cabling and connections are going to have a much greater effect than any (dc) thermals.

The spades were crimped on to the conductors.

I'm not sure why I had assumed thermals were a noticeable error source.  The constant changing polarity nature of AC cancels out any thermal EMF errors?

Thanks!  Very informative.  The RG-213 cable I am using came with the 4920M that we purchased recently, it came with a really nice quality female n-type to male banana adapter.  I've tried to get more of these from pomona, but it seems as if they don't produce it anymore.

[ManateeMafia]

I have been thinking about making a chart for similar issues between different calibrators so I started a basic one. Only catch is that Fluke and Datron/Wavetek had different cutoff values for different frequencies and voltages.

Looking over the 4808 manual for what you are sourcing, I see they recommend screened 2-wire cables (looks like twinaxial) for no external sense. With external sense use coaxial or two twinaxial cables. Either way I would at least try external sensing to see if that improves your measurements.

Let us know if you see any improvements. I have been making and buying test cables for a while. It looks like I need to build some twinax and triax cables for calibrating. I have bought several dozen spade lugs from jswilley and have crimped all connections without solder. The cable is Belden 8719 as recommended by Fluke for their 5440A-7003 cables. So far I have made 3 sets and you can't have too many. I have the 7002 versions but they don't like the Pomona binding posts because they are not deep enough. It is ironic since the plugs appear to be made by them.

I also bought a few of the Pomona 1740 adapters. I got them from ebay a couple of years ago. The only one I can find online is at qservice.com
An alternative is the E-Z Hook 9414. Specs from here  http://www.testpath.com/pdf/ezhook/e-z-hook-8013-datasheet.pdf

[dacman]

I use a lot of RG58 cables for ac.  Another cable type I've used is the Pomona 1658-T (RG214/U) , which has a slightly larger outer diameter and same pF/ft as RG-213.

Pomona sells adapter kits that can be used to construct the adapter (5698 and 5748).  The center conductors are gold plated and the outer shells are silver plated, which can tarnish, but I do prefer the silver over nickel or stainless.  I also use the Pomona adapter.

I"ve also bought some Pomona 2BC cables, cut them in half, and installed Pomona 73050 connectors to have double banana to N(m) cables.  (A cable with these ends could be constructed using the Pomona adapter kits and 5749 adapter cable.)

[CalMachine]

I have been thinking about making a chart for similar issues between different calibrators so I started a basic one. Only catch is that Fluke and Datron/Wavetek had different cutoff values for different frequencies and voltages.

Looking over the 4808 manual for what you are sourcing, I see they recommend screened 2-wire cables (looks like twinaxial) for no external sense. With external sense use coaxial or two twinaxial cables. Either way I would at least try external sensing to see if that improves your measurements.

Let us know if you see any improvements. I have been making and buying test cables for a while. It looks like I need to build some twinax and triax cables for calibrating. I have bought several dozen spade lugs from jswilley and have crimped all connections without solder. The cable is Belden 8719 as recommended by Fluke for their 5440A-7003 cables. So far I have made 3 sets and you can't have too many. I have the 7002 versions but they don't like the Pomona binding posts because they are not deep enough. It is ironic since the plugs appear to be made by them.

I also bought a few of the Pomona 1740 adapters. I got them from ebay a couple of years ago. The only one I can find online is at qservice.com
An alternative is the E-Z Hook 9414. Specs from here  http://www.testpath.com/pdf/ezhook/e-z-hook-8013-datasheet.pdf

A chart made by someone with lots of experience and knowledge about cabling and proper impedance matching would be extremely handy.  I would hope it would have varying voltages and varying frequencies?  As well as 2 wire and 4 wire?

I actually did come across thsoe EZ Hook 9414s just the other day.  I plan on purchasing a few to try out.  I'm also going to look into getting some Belden 8719 to try out. 

I use a lot of RG58 cables for ac.  Another cable type I've used is the Pomona 1658-T (RG214/U) , which has a slightly larger outer diameter and same pF/ft as RG-213.

Pomona sells adapter kits that can be used to construct the adapter (5698 and 5748).  The center conductors are gold plated and the outer shells are silver plated, which can tarnish, but I do prefer the silver over nickel or stainless.  I also use the Pomona adapter.

I"ve also bought some Pomona 2BC cables, cut them in half, and installed Pomona 73050 connectors to have double banana to N(m) cables.  (A cable with these ends could be constructed using the Pomona adapter kits and 5749 adapter cable.)

I think I'm in love with that universal adapter kit/cable!!

[ManateeMafia]

...
A chart made by someone with lots of experience and knowledge about cabling and proper impedance matching would be extremely handy.  I would hope it would have varying voltages and varying frequencies?  As well as 2 wire and 4 wire?

I actually did come across thsoe EZ Hook 9414s just the other day.  I plan on purchasing a few to try out.  I'm also going to look into getting some Belden 8719 to try out. 

I think I'm in love with that universal adapter kit/cable!!

I would be lying if I said I had a lot of experience and knowledge. Right now it is more of a long term experiment that is just beginning. I am collecting the information from the following operator's manuals

1. Datron / Wavetek 4808
2. Fluke 5700A Series I
3. Fluke 5700A/5720A Series II
4. Fluke 5730A
5. Fluke 5500A
6. Fluke 5502A
7. Fluke 5520A
8. Fluke 5522A

I will add some generic notes on the specified cables if the manual lists them or the Fluke website (ie 5440A-7002).
Most of the help will come from people who actually use these calibrators .

The first four are almost complete. I can post that data while working on the next four. It will be broken down by functions and ranges. It is making my to-do list of cable building bigger as I write it.

[CalMachine]

I did a little bit of testing this morning with different cables and setups.  I also wanted to see if there was much of a difference between Channel A and Channel B on my 4920.

I wanted to stick with the same nominal of 100 mV @ 1 MHz for this test.  I rounded up 6 different cables to use.  I plan on getting the twin-axial Belden 8719 here in the future to make some cables with.  That will be thrown into the mix as well 

1) 40 GHz RF/Microwave cable with necessary connectors.
     We have several of these high quality M/A COM serialized RF cables that I wanted to try.  They seem to have yielded the best results.  But at what frequency would you start using this?  And at what voltage would you stop using this?

2) Solid Core Silver Plated Copper Twisted Pair.
     The materials for this cable is from JSwilley and I hand made it myself.  I really like these cables and I think they turned out fantastic!  It is somewhat of a nuisance getting the cable onto multiple bindings, though.

3)  RG213 N-Type
     This cable came with the 4920M we purchased recently.  The Pomona N-type => Banana adapter came with it as well.  I really like this cable and it seems to yield pretty decent results.

4)  MultiContact Low Thermal Gold Plated Banana Cables
    These are some really good single banana cables.  I use these for a lot of applications.  They were given to us by Jay Klevins from Ohm Labs when we purchased our 2 standard resistors.

5) Pomona 5291A-60
    I despise the retractable sheath cables... they slowly push the banana plug out of the jack.  This particular pair is too long for my liking.  The insulation is also PVC, which I don't think has as great of electrical/thermal properties as PE or PTFE (Correct me if I'm wrong).

6) Pomona 2BC-60
   This is a pretty good higher voltage AC cable.  It did yield high results compared to other cables, however.  This one is also a little too long for my liking.

I have attached 3 recent calibration certs with the reported 100 mV @ 1 MHz value and the data reported from Fluke for the 4920.  Which cable, in your mind, performed the best?  So far, I'm liking the measurement results with the 40 GHz RF cable.  I only wish they could withstand a higher voltage 

[quarks]

I have tested quite a few cables
https://www.eevblog.com/forum/metrology/diy-low-emf-cable-and-connectors/msg880768/#msg880768
and besides the low EMF (DC) tests, I found RG400 to be very good for AC

[ManateeMafia]

This is a list I made that I have been fiddling with for a couple days. This is just information straight out of the operator's manuals. There is a lot of room for improvements. The second page eventually will have the Multi-Product calibrators. I will add as I get time from current projects.

[dacman]

This is a list I made that I have been fiddling with for a couple days. This is just information straight out of the operator's manuals. There is a lot of room for improvements. The second page eventually will have the Multi-Product calibrators. I will add as I get time from current projects.

There is a caveat with the 5700A series of calibrators.  This is out of a 5700A/5720A manual in the secondary ACV performance section:

External Sense ...................................................... Selectable for 2.2 V, 22 V, 220 V, and 1100 V ranges; 5700A/5720A <100 kHz, 5725A <30 kHz

This is out of the 5730A Operators Manual:

External Sense ...................................................... Applicable for 2.2 V, 22 V, 220 V, and 1100 V ranges; 5730A <100 kHz, 5725A <30 kHz. Specifications are the same as internal sense.

The statement in the 5730A manual is more precise because External Sense is selectable at, for example, 10 Vac 1 MHz, but it has been my experience that the output isn't within the 2-Wire tolerance limits.

[dacman]

CalMachine, unless those single cables are twisted together slightly, the readings will vary greatly.  Moving them slightly can vary the readings greatly.

[ManateeMafia]

@dacman
Thanks for the info. I imagine it took a while to sort out the cabling for a 3458A calibration?

I asked about that same cabling a year ago and was told by the seller that it is not twisted pair.

[CalMachine]

This is a list I made that I have been fiddling with for a couple days. This is just information straight out of the operator's manuals. There is a lot of room for improvements. The second page eventually will have the Multi-Product calibrators. I will add as I get time from current projects.

Great list, thanks!  I've purchased some Belden 8719 that I will be making some comparison measurements with.  Stay tuned.

CalMachine, unless those single cables are twisted together slightly, the readings will vary greatly.  Moving them slightly can vary the readings greatly.

I will twist the MC leads and throw those measurement results into the mix.  I think I'm going to drop the 60" pomona low thermal leads... too long for my liking.

[dacman]

@dacman
Thanks for the info. I imagine it took a while to sort out the cabling for a 3458A calibration?

I asked about that same cabling a year ago and was told by the seller that it is not twisted pair.

I used to use 5440A-7002 cables for the 3458A but they have a birdcage banana that spins around the banana shaft that is a contact that sometimes needs attention (it is made of non-plated copper).  I also did not like the repeatability that I was seeing.  Knowing about the statement in the 5700A/5720A manual about using spades with the 5720A only, I switched to Pomona 1756 cables (one high, one low, shields are shields, not a conductor) and have never went back to the bananas for DCV (unless it's not a critical measurement).  I like to use Pomona 2BC cables for ACV, although for some I have attached an N connector.  I like the 8508A-LEAD kit from Fluke, but it is pricey and precious.

[ap]

Have always used twisted shielded pair teflon wires with spades for the 3458A. And low emf banana cables (Pomona), twisted that together, for dc and higher voltage measurements (also for the 5440B). Why?
The Pomonas are convenient, you just plug them in, and they are insuated well, and if the measurements are not critical, they are perfectly fine (at least in my environment). For more critical measurements twisted shielded cable reduces both magnetic and electric field noise (you dont get that with coax cable allone), and the teflon insulation avoids leakage current in high ohms measurements. The Belden wire referenced above I found is PVC insulation (if I am not mistaken, maybe wrong subspec...?), definitelly dont like that. Mil spec wires are pretty good. They are also pretty rugged (strong outer braid).

[CalMachine]

I see this thread has seemed to have derailed to DC applications    I mostly wanted to start this thread for people to look to for AC applications.  The ranges I had in mind would include 1 mV @ 10 Hz through 700V @ 100 kHz.   Basically the AC capabilities of any 4808 / 5720. 

Does anyone know why good high quality RF cables are not used for lower voltages? Say < 100Vrms.  I plan on testing the RF cable against the Belden 8719 cable I made, here in the next day or 2. 

[TiN]

Just minor note, 5700/5720A can do 700VAC 100kHz only with help of 5725A booster.

[CalMachine]

Just minor note, 5700/5720A can do 700VAC 100kHz only with help of 5725A booster.

Hmm really?  I could have sworn it could do 700VAC @ 100 kHz....  Oh well, maybe that's why very few labs are able to give accredited measurements at that level.  The 4808 can, that's why I am including it in here.  Also, I've found a way to get the 4808 to source 30V @ 1 MHz, instead of the previous max 20V.

[CalMachine]

Picked up a few steals on Ebay this weekend!  I plan on throwing these into the cable testing mix.

[ManateeMafia]

The Fluke one will be ok until the plugs start popping out while you insert it. They are good on the deep binding posts like the Fluke 742A but my 3458A spits them back out at me. It can be a bit frustrating but it is hard to pass on a deal like that.

[CalMachine]

The Fluke one will be ok until the plugs start popping out while you insert it. They are good on the deep binding posts like the Fluke 742A but my 3458A spits them back out at me. It can be a bit frustrating but it is hard to pass on a deal like that.

I will cut the retractable sheaths off, I despise them.

[dacman]

It also has contacts that spin, and if they build up resistance they can cause measurement errors, especially for Ohms.  (They need lubricated and moved every once in a while, in my opinion.)  I apologize for tainting the thread, but I do use those Fluke cables for ACV sometimes.  (At my day job, there are some of those Fluke cables that have been there longer than I have, and some of them have the sheath cut off.)

[CalMachine]

It also has contacts that spin, and if they build up resistance they can cause measurement errors, especially for Ohms.  (They need lubricated and moved every once in a while, in my opinion.)  I apologize for tainting the thread, but I do use those Fluke cables for ACV sometimes.  (At my day job, there are some of those Fluke cables that have been there longer than I have, and some of them have the sheath cut off.)

Multicontacts Banana spring always spin, right?  I've not had any issue with their banana connectors.  Thanks for the heads up though, I'll be on the lookout for weird resistance measurements using cables with spin-able banana springs.

[dacman]

Several companies use bare copper terminals (not just Fluke), but they are more subject to resistance buildup than gold-plated ones.  Most all Fluke high end calibrators and DMMs use bare copper terminals.  That in theory will produce a lower thermal than gold plated ones.  (An example of a DMM that uses gold plated terminals is the 3458A.)  If DeoxIT is used on bare copper, the terminal can turn green over time, so I just use a lubricant, if anything.