Resistive heating using high freqency power supply in threaded bar

130mikep

Hi i've come across a problem at work and was hoping someone could help me to understand what is going on exactly.

power supply: 6kW, 9.2A, 5890V, and running at approx 50kHz.

this power is routed through ∅3mm Kanthal AF bar, threaded at one end, and what we see is resistive heating through the bar, but almost twice as hot at the thread as anywhere else.

my first thought was skin effect, so i tried reducing the diameter of the bar but it was (almost) the same temperature at ∅2.4mm.

My question is can anyone explain why the threaded section causes so much more heat?

NascentOxygen

When the threaded bar is screwed into its threaded socket, much of the thread of one does not contract the thread of the other. For example, the peaks and valleys of each won't be in contact with the other, so no current goes through those areas. If you examined the thread surfaces with a microscope you would probably find there is a lot of area not making good electrical contact. This means the current is squeezed through narrowed regions, giving higher current density and hence higher resistance and generating more heating. If it were possible to have precision tooling produce smooth near-perfect thread on the bar and the socket, I'd expect to see little extra heating.

What is the electrical contact at the other end that results in no significant heating?

Carl Pugh

Very wild guess:
When the bar is threaded it work hardens and the surface becomes magnetic. The current flows on the surface of the bar, so the threads have much more resistance.
Again this is a very wild guess.

jim hardy

here's my guess:

At that frequency i would think " skin effect " causes direct heating only very near the surface. Eddy currents in iron keep the magnetic flux near the surface.

http://books.google.com/books?id=ncA...ion%22&f=false


So heating occurs at the surface and progresses inward by old fashioned thermal conduction.

The threaded part has more square inches of surface area per inch of length than does the non threaded part. So more heating per inch of length goes on there.

sound logical?

see also:
http://www.uie.org/webfm_send/441 page 11

sophiecentaur

The answer is not obvious, I think.
When you say "twice as hot", what does this mean? Is it twice the reading of your thermometer in degrees (unspecified scale), or just 'feels really hot'? Are you assessing the temperature by its effect on your finger? If you are, then the greater contact area between metal and skin on the threaded bit could have an effect.

I may be confused by the dimensions of the bar. Do you mean 3mm? This would correspond to around an M3 screw (?). I tried to estimate the expected change in resistance when the bar is threaded. The thread depth of this will be about 0.5mm so the cross sectional area of the conductor would be significantly different when threaded. I can't think how to work out precisely how the helical slot would affect the resistance per m but the ratio for two smooth wires with diameters equal to the max and min thread diameters is about 0.6 (2.5²/3²) - corresponding to 1.6 times the power dissipation for the same current. If the effective resistance ratio is more like 0.8 (i.e. somewhere in between 0.6 and 1) then the power dissipation would be 1.25 times as much. Would this be significant?

Assuming that there would be measurably more power dissipated then you could ask whether the effective cooling would be more or less due to the threads. Would it be more due to the greater surface area or less due to turbulence slowing the flow of air in near contact with the bar?

It would be useful if you could measure the actual temperatures involved and, possibly, the resistances of plain and threaded portions.
Are the threaded and unthreaded portions in the same 'thermal environment'?

130mikep

okay, thanks for all the replies! my email didnt tell me anyone had replied so sorry for the delay...

ok i'll try and go through everything so far, but im dont know how to quote you all seperately...

@NascentOxygen:

it isnt a connection issue weve tried threaded connection, brazing over the top, silversoldered etc. it is purely the thread being there creates heat somehow...

@Carl Pugh:

i was thinking along these lines too, the magnetic field of the thread could disrupt the field of the bar?

@jim hardy:

thanks for the links, ive had a quick read and looks interesting, i'll get back to you on them!

@sophiecentaur:

all of the measurements have been taken using a thermal imaging camera and when i say twice as hot, i mean on a bar half M3 threaded and half ∅3mm bar (your assumptions were correct) were approx 320°C and 180°C respectively.

An m3 screw has approximately a 2.4mm core diameter, this is why i cut a section down to ∅2.4mm, with only a small change in temperature, this leads me to the assumption that either th increased surface are, or the indroduction of a coil "wrapped" around a bar has created the heat.

Once again thanks for all the help :)

sophiecentaur

I sort of predicted a probable ratio of power dissipation of about 1.25 due to difference in resistance. Your two temperatures are about 300°C and 160°C from room temperature. That would imply a ratio in heat losses of 1.9. So something isn't as you'd expect. It's either resistance or thermal loss that we're getting wrong.

I did wonder about the resistivity going up with temperature but I just looked at the data sheet - it only changes by 4%, even up to 1000°C - dang!! Having found the datasheet, it appears that the resistance ratio would only be 1.58 for two bars of the different diameters so, my estimate must be fairly close (somewhere in between, as its a mixture of CSA)

You could always measure the voltage gradients (V/mm) along the bar and check that the ratio is, indeed the 1.25 I suggested. Two prods and a DVM should give an accurate enough idea. It's probably better doing it that way than using a DVM just to measure the resistance as it eliminates the problem of contact resistance.

Other than that it just has to be magic!

NascentOxygen

So, if you were to create a short length of thread midway along a bar and leave that thread exposed and unused, but silversolder both the electrical end connections, you can demonstrate the threaded portion glows much hotter than elsewhere?

130mikep

yes, if you could easily thread the centre of a bar, this would be the case, but the thread pretty much has to start at one end.

i think ive attached a pdf of an image of our old connection, and you can see the peak temperature is at the end of the thread, furthest from the connection. you can see the silver soldered nut connection is creating hardly any heat, its literally being generated in the thread...

@sophiecentaur i think i understand what you mean (I'm more mechanical than electrical i have to admit!) so i'll get on that to find the voltage drops per mm over threaded and non threaded parts and find the ratio, is that right? and that will tell me whether the heat is likely due to lack of cooling or resistive issue? and yeah the material is quite consistent thoughout the temperature range, one of the main reasons we use it :)

Attached Files

Thermal image of connection1.pdf (42.9 KB, 7 views)

NascentOxygen

Have we an explanation for the cool region between two warmer regions?

Attached Thumbnails

 

sophiecentaur

I wondered whether it made any difference having the rod mounted vertically or horizontally and if the air circulation at one end is different from what happens at the other. The important thing is not to include the contact resistance between the power source and the rod in your voltage measurements.
But at least an electrical measurement (you understood me correctly, I think) would eliminate one possibility.
Another thought - how uniform is the temperature distribution along the rod? Are the 'high' and 'low' regions distinct and uniform within themselves?

NascentOxygen

Now that we know that exposed thread runs hot, it must be generating exceptional heat here because the greater surface area of the thread more rapidly dissipates heat to the air, like the fins on a heat sink.

NascentOxygen

So, we're all stymied here? Pity, it was getting intriguing .....

uart

My guess would be that the skin depth is less than the depth of the thread, forcing the current to flow in significantly more convoluted path (a zig-zag path) through the threaded area and therefore increasing the resistance in this region.

NascentOxygen

http://www.physicsforums.com/showpos...9&postcount=10