Ferrites/cable advice

Simon,

You inspired me to get out a type-k thermocouple and measure the temperature of the ferrites.  The ones on my mains cables and my longest Ethernet cables (5m) are both about 0.2°C above the temperature measured from an unused ferrite placed fairly close.  I do seem to have high levels of radiated RFI at my house.

No doubt dear Huge of your properly measure, but thermocouple can't generally be repetitive in the measurements within 0,x K due to contact resistance and pressure applied. Apart this, my concern are on the reason of such differences in temperature, how to distinguish between the Joule effect in the main cable under use (and the conduction to the ferrite) and the intrinsic heat generation into the ferrite by RFI? Sorry if I'm not clear in my question and, please, apologies my comments if are not pertinent.

Edit: sorry, you are measuring at Ethernet cable, I so suppose the Joule effect is negligible. 

Huge, genuinely fascinating 

you could perhaps wind severeal turns of hookup wire around your leads and connect to a 'scope and see if you have a hash on the display.... I have done this on the DC leads from SMPS, and I have several Apple devices showing high RF in this regard..

bicela posted:

No doubt dear Huge of your properly measure, but thermocouple can't generally be repetitive in the measurements within 0,x K due to contact resistance and pressure applied. Apart this, my concern are on the reason of such differences in temperature, how to distinguish between the Joule effect in the main cable under use (and the conduction to the ferrite) and the intrinsic heat generation into the ferrite by RFI? Sorry if I'm not clear in my question and, please, apologies my comments if are not pertinent.

Edit: sorry, you are measuring at Ethernet cable, I so suppose the Joule effect is negligible. 

Hi bicela,

I am aware of these effects...

All measurements were repeated at least 5 times and any outliers discarded, repositioning the thermocouple each time to reduce the effect of mechanical stress and statistical variance.

Measurements of the surface temperature of the mains cable were taken (underneath pipe insulation) and these were indistinguishable from the ferrite not excited by the cable.

Although not tested, I don't believe that leakage currents were significant as the measurement of the surface of the wire would have been equally affected.

But still given that 0.2°C is very small and even with these controls it's at the limit of detectability.  I believe the effect to be present at the limit of detectability; probably more than zero, but definitely less than 0.5°C.

The reason for picking a type K thermocouple (rather than a more sensitive semiconductor junction sensor) was so that the time taken to settle was sufficiently low that all the sets of three measurements could be taken in a time scale variations of ambient temperature were not significant within each set so the differences measured within each set would be valid.  Taking multiple measurements could then be used to reduce statistical errors and eliminate outliers between the separate sets of three.

First paragraph should have read...

All measurements were repeated at least 5 times in sets of three (energised ferrite, non-energised ferrite & cable surface using the same thermocouple) and any outliers discarded, repositioning the thermocouple each time to reduce the effect of mechanical stress and statistical variance.

but I messed up a cut and paste edit 

I am genuinely impressed by the time Huge is finding to experiment with this. I would have liked to play around more but juggling work and a baby the best I could manage was another shot to see if I can get better results than my previous bodged attempt.

Removing the ferrites from all signal carrying cables was task one. Then redistributed them to PowerLines (despite having misgivings about putting them on an earthed power cable), the DC input from the NAPSC, the ethernet cable, other devices like LED lamps and the phone etc. Finally the only signal carrying cable I attached them to was the individual wires of the speaker cable closest to the 250.

On first listen the groove was back and that mild compressed sound I had on the first attempt was completely absent. I couldn't honestly say one way or another that it was better than no ferrites but it certainly isn't having a detrimental impact now.

I am glad Huge convinced me to have another go.

Huge posted:

First paragraph should have read...

All measurements were repeated at least 5 times in sets of three (energised ferrite, non-energised ferrite & cable surface using the same thermocouple) and any outliers discarded, repositioning the thermocouple each time to reduce the effect of mechanical stress and statistical variance.

but I messed up a cut and paste edit 

Thank you Huge for the precise description and the kindly answer, alright.

I suggest a trick, instead to try to approach with multiple measurements that are all affected by contact problems and other effects is maybe better (when the delta T is so limited) to arrange a differential measurement, one Tc fixed on the "cold" ferrite and the other to the ferrite with RFI, of course I suppose you have possibility to use the acquisition of the difference of signals when all is off and when one ferrite is affected by the supposed thermal effect. Still remain the uncertainty of external convective boundary condition, anyway.

Of course I'm with you on detectability limits, but the trick may let you to measure under another approach the same.

Your knowledge is always wide, consider my reply as aid and not an advise that you don't need. Good night!

Hi Mau,  I guessed you might be interested in this  -   temperature & ferrite rings a few bells for you.        In my previous life in refrigeration equipment design & mftr we had a number of test calorimeters.  They were required to both verify conformity in random production units,  prototype & design changes & a very complex unit for official EU & USA standards verification.   Whilst this calorimeter had many K type's,  their accuracy was not approved for anything other than air & material temp monitoring.  The critical areas used T type & each of those had to be doubled up & were questioned if they disagreed by >0.1'K ('C) 

Dear Mike, I listen the bell very loud here!  But we are going OT 100%!
Thermocouples are one of my long term works, last develop is a thermocouple painted, one small line of Platinum and one of Gold, this is very fast because practically massless... As done 50 year ago by others, revamped now with newer acquisition systems.

Ah! that makes three of us who worked with calorimetry.  In my case Differential Scanning Calorimetry to analyse the thermal properties of phase change behaviours of polysaccharides.

I also built a computer controlled desk top water bath that was stable to +/-0.3mK (control limit, so max deviation from set temperature rather than RMS), so we could measure the volume changes with time and relate this to the thermal properties.  It used a quartz crystal thermometer as the sensor.  It also weighed 150kg.

At the moment I'm just working with the tools I have.  I no longer have access to the more precise stuff!