Thermal potting methods

[sdk_imported]

I need some creative ideas for getting the heat out of the windings of a brushless dc motor. I'm considering potting the windings with a thermally conductive epoxy or thermoplastic. Does anyone have experience doing this? Are there any other methods of improving the conduction path from the windings to the laminations?

 

[jbartos]

Suggestion: Many good thermal conductive materials are also good electrical conductors, which is obviously unvelcome characteristics. What is left are non-conductive materials, e.g. plastics as you mentioned. Normally, the windings are cooled by gas or liquid through some cooling system paths.

 

[gjones33]

Hi sdk,

Can you please provide more information on the motor and application, I may be able to help.

Cheers
G

 

[UKpete]

I haven't anything exotic to suggest, just attention to detail:

1. Impregnation.
VPI with a suitable resin gives lower winding temperatures than simple dipping - proper VPI is a vacuum cycle followed by a pressure cycle to push the resin in and eliminate air pockets - as described in the FAQ section of:

http://www.pdgeorge.com

Insulation materials should be chosen to suit VPI - e.g. use of un-calendered (i.e. not rolled) slot liner material as it is better at absorbing the resin.
I have also seen brushless motor windings fully injection moulded with loaded nylon; more expensive on tooling but it does get over the problem of resin drain-off and curing times.

2. Slot fill.
High slot fill makes the stator more difficult to wind but it does increase the thermal conductivity between winding and laminations (and reduce the copper loss of course). Also, a slot with a lot of space in may suffer from resin drain-off during curing and therefore poorer thermal conductivity, unless special jigs etc are used.

3. End-windings
Windings have much better thermal conductivity in the axial direction than in the radial direction (where the heat has to flow through insulation) so it may be advantageous to pot the end-windings, or use some form of direct cooling to suck the heat out from the slots. Any potting resin should perhaps have some flexibility to cope with thermal cycling. As mentioned in the previous post, it is advisable to keep electrically conductive materials a few mm away otherwise eddy-current losses may become significant (for the same reason the stator core is laminated) - particularly important for high speed/high frequency machines.

4. Laminations
To reduce the radial temperature gradient a good fit is required between the lamination stack and external housing - this usually means grinding the o/d of the stack. The interference fit should still be good with the machine at operating temperature.

As a final comment, I think simulation using FE or lumped parameter modelling is worthwhile - I know some errors creep in as thermal resistances and heat transfer coefficient's are difficult to reliably predict, but it can be a real eye-opener if you get the chance to experiment in this way.

You may find that potting doesn't make such a big difference, and that most of the heat flow is radially outward through the lamination stack.
For typical software see:

http://www.motorcad-usa.com/

http://www.cedrat.com/software/Motorcad/Motorcad.htm

http://www.speedlab.co.uk/

 

[sdk_imported]

Thanks for the great feedback UKpete.

I have done a fair amount of FEA and lumped parameter modeling. However, my modeling is only as good as the data I put in. My understanding of how to model the thermal resistances INSIDE THE SLOTS is limited. Do you have some rules of thumb for modeling the thermals resistances inside the slot (i.e. contact resistances, airpockets, etc.).

On a related topic, I am getting mixed feedback on the use of epoxy insulation in place of paper. The epoxy offers thermal benefits over paper. This motor operates from 200 to 400 volt supply. Would you be confortable using epoxy insulation for prototypes? How about production runs?

Gjones33 - this motor runs at 30,000 rpm and delivers 7 kW of shaft power. It will likely be a 6 pole 9 slot configuration, but nothing is finalized.

 

[gjones33]

Hi sdk,

Is it necessary for the motor to be totally enclosed or can it be ventilated, in which case the latter would be the best option especially if the winding overhang can be placed in the airflow.

For high speeds, the fewer the poles the better because hysteresis and eddy current losses are reduced.

If variable speed is required then I suggest a sensorless drive using a single pulse per phase rather than pwm would be a good deal more efficient, reduce heating in the stator windings and cost less. I think you will find the following up-dated link useful.

http://www.sensorlessdrives.com

Cheers,
G

 

[UKpete]

sdk

I go along with gjones - 4 pole might be better from a losses and cost point of view. On the subject of drives, using a non-PWM regime may be better - all that high frequency chopping you get with PWM does put extra losses in. A non-PWM drive might, for example, control speed by varying the DC link voltage.

For some reason I'd assumed it was totally enclosed; the machines I worked with were because there were no internal oil seals for the bearings. Air cooling would certainly keep the rotor temperature down. Have you considered oil-spray cooling? What rotor sleeve do you use?

I do have some measured slot thermal resistances, but not here at home. I'll try to post some data tomorrow.

 

[UKpete]

I have some measured data on slot thermal resistances. It does depend on the slot liner and the fill factor, but with a typical .23mm liner and 35% copper fill, the measured thermal resistance is 240W/m^2K. With a reasonable fill, a first approximation would be to assume a solid mass of copper in the slot.

I notice that in the MotorCad software (one of the links I posted above), the slot is modelled as a set of concentric cylinders with insulation in between. I don't have this software now but I did have a demo version a while back - as I recall it had a lot of empirical data in it for htc's and thermal resistances.

 

[jbartos]

Suggestion: High resistance metal powders that have higher thermal conductivity than various plastics could potentially be applied, when solidified by adhesives.

 

[UKpete]

sdk

your question:
"On a related topic, I am getting mixed feedback on the use of epoxy insulation in place of paper. The epoxy offers thermal benefits over paper. This motor operates from 200 to 400 volt supply. Would you be confortable using epoxy insulation for prototypes? How about production runs?"

I'm not sure I understand what you mean by epoxy insulation. It is common to use epoxy as an impregnant but not as a solid insulation e.g. for slot insulation. The way I see it, there are three classes of materials thermal insulation-wise: air/void (poor), insulation (intermediate) and conductor (good). Epoxy and Nomex paper falls into the second category.

 

[sdk_imported]

UKpete,

I am still interested in this topic. You mentioned that you've seen motors potted with loaded Nylon before. Do you have any additional information about this (i.e. name of motor manufacturer, issues surrounding nylon potted motors, etc).

Thanks,
Scott

 

[UKpete]

sdk - I have looked up my old notes and have some info, but really you need to talk to an injection moulding expert.

I know that injection moulding is used to encapsulate stators in a number of motors in mass-production, the particular example I saw was from Sankyo Seiki. It was a 120W 3000rpm 24V brushless dc motor with built in electronics, and was used in an embroidery machine. The complete stator was encapsulated in what looked like plain black nylon, so that neither the end-windings nor the laminations (on the inside and outside) were visible anywhere. I believe that other small motors from other manufacturers used in automotive applications are also injection moulded.

I have only seen one large machine (with a stator outer diameter of 130mm) injection moulded, and that was a prototype for use with a 50kW high speed micro-turbo generator of Japanese origin. I think that used a nylon loaded with something to improve the thermal conductivity.

I was involved in some injection moulding trials on motors with airgap windings, about 75mm o/d, the plan being to provide strength to the winding, bond it to the stator core (slotless) and provide some thermal conductivity. To get good penetration of the nylon-12 into the winding the key to success was to pre-heat the winding/stator assembly before placing it in the injection moulding machine, to about 190°C, with the injection resin even higher I think. This nylon resin was loaded with another material for another purpose, this had to be electrically insulating otherwise you will certainly get inter-turn insulation problems because of the extreme force used in the process.

Hope this helps.

 

[sdk_imported]

Thanks for you input. I've been asking around I am finding a few instances where aluminum nitride filled polycarbonate is used for injection molding. Does this sound familiar to you?

Your comments about heating up the stator before injection molding make a lot of sense to me. It's hard for me to imagine that injection plastic can get clear down into the slots of the windings.

In your test cases, did the 190 degC preheat temperature cause any issues with insulation on the magnet wires? Also, did you find an improvement in thermal conductivity in your test cases?

Thanks.

 

[UKpete]

I'm not familiar with that material but that doesn't mean much. I think an important attribute is that the shouldn't be abrasive, this would damage the wire insulation as well as the injection moulding machine.

The 190degC was used with a 0.28mm Kapton wire (aka HML, polyimide) which has a high abrasion resistance and also a high temperature rating, something like 220degC continuous - but it is expensive. If you use a more standard wire enamel like polyester say class H (180degC) the 190degC won't do any harm for periods of up to a few hours, although the enamel will be softer and less resistant to abrasion. You may well find that you get good consolidation of the winding when it is pre-heated to a lower temperature, depending on how soft the injection material is at that temperature. If the operating temperature of your winding is not high, you could use a plastic with a lower melting temperature. It's a bit of a mix-and-match, trial-and-error thing.

The brushless dc motor that I mentioned did have plastic all through the slots, I cut it open but didn't see any voids anywhere. But it was only a small stator, about 30mm core length.

What wire diameter do you want to use? Finer wires have a thinner enamel and are obviously easier to damage.

sdk, please let me know how you get on. I think it is a technique that is becoming more popular and I'm interested to know.

 

[jbartos]

Suggestion: Nomex, mentioned in UKPete posting, is used in transformers, e.g. PowerCast by Square D. Visit

http://www.dupont.com/nomex

etc. for more info