Hardware temperature control tsdz2
- Thread starter andrea_104kg
- Start date
hemo
100 W
I see copper scouring type pads are being utilised to fill some of the larger air voids where thermal pads may prove to be not so good, what are folks thoughts about rolling up and and mashing aluminium foil into shape to fit these larger air voids or even clean old aluminium take away trays/cartons ?
scianiac
10 kW
- Joined
- Oct 24, 2018
- Messages
- 669
hemo said:
The problem with most mashed up options is the thermal conductivity of the material is almost irrelevant once you add so many gaps of either air or even thermal compound. The result would probably be comparable to just a brick of pads which is what most people do, if you really wanted to put something in there and didn't want to cut some thick blocks of aluminum like some have done, maybe figure out a way to orient the sheets radially so the heat travels from one end of a sheet of aluminum to the other with only 2 contact points.
Honestly I haven't done anything besides the heatsink plate and a few pads on the opposite end of the motor and it seems to cool well enough for moderate-heavy use, the whole system just isn't built to wring max power out of it constantly, better to have a good gearing selection and run the motor at reasonable power levels and RPM ranges.
gordone said:
Where are you putting the thermistor? It should be mounted to the motor so it's inside the sealed case and shouldn't need any waterproofing. Not that it will hurt, although most thermistors are sealed anyway.
I have copied the next question from here
Silicon pads between the outside of the stator and casecover are less effective than an aluminium heatplate between motor and case.
But for some indication of the difference between inside and outside, read there
FYI
It is also possible to measure inside the case with an external temperaturemeter.
In that case you have to cut the sensorwire and run it through the wire grommet behind the chainring from the inside.
After gluing the sensor on the bare motor, you solder the wire back to the temperature meter.
It all depends how and what you use for better heatconductivity.bpratl said:
Silicon pads between the outside of the stator and casecover are less effective than an aluminium heatplate between motor and case.
But for some indication of the difference between inside and outside, read there
FYI
It is also possible to measure inside the case with an external temperaturemeter.
In that case you have to cut the sensorwire and run it through the wire grommet behind the chainring from the inside.
After gluing the sensor on the bare motor, you solder the wire back to the temperature meter.
dzid_
1 mW
- Joined
- Jul 18, 2022
- Messages
- 11
Does anyone have motor case temperature, battery current, voltage and wheel speed and cadence dataset?
I was thinking it of simple temperature estimation.
The algorithm would estimate three states - winding temperature and motor temperature (where we measure) and case temperature.
The temperature initialization of all the states could be done via sensor or via winding resistance measurement if someone doesn't have a sensor (higher resistance means higher temperature. But it can only be done when motor speed is 0, by activating a single phase for a second).
Constants would be:
- winding heat capacity (to have a good margin, copper wire mass could be used. Fixed, not fitted to data)
- motor casing heat capacity (either fitted or guessed based on steel mass)
- winding to casing conduction/radiation coefficients (whatever values fit good to data) - this should be the same for everyone
- casing to air - mainly convection depending on the air speed
- motor heat dissipation coefficient to outside case (for those without a sensor). This would be different depending on a cooling solution and would have to be adapted online (basically estimated temperature would have to be compared to temperature obtained from winding resistance - it doesn't have to be super precise - we just change parameter up or down if there is temperature discrepancy. It would be required to activate a motor phase for a second when not pedaling to take resistance measurement. Tiny battery energy loss.).
The only problem I see is that the uncooled motor cools a lot through radiation (ΔT⁴) and cooled via conduction (ΔT). Perhaps a different model could be selected in the configurator for people with heat pads. Single variable adaptation with a single known state is easy on controller because doesn't require memorizing time series.
......Or am I overcomplicating this? Because if we can (sometimes) measure winding resistance then we know the winding temperature, which is what we are really interested in. The big assumption is that the rider sometimes takes brakes from pedaling and we could actively measure the resistance.
We could also give try to measure Resistance dynamically knowing BEMF from cadence, etc. I have no experience with how reliable this would be. https://build-its-inprogress.blogspot.com/2019/10/motor-temperature-estimation-without.html
Anyway, we could give it a try if someone had a bunch of riding data saved.
I was thinking it of simple temperature estimation.
The algorithm would estimate three states - winding temperature and motor temperature (where we measure) and case temperature.
The temperature initialization of all the states could be done via sensor or via winding resistance measurement if someone doesn't have a sensor (higher resistance means higher temperature. But it can only be done when motor speed is 0, by activating a single phase for a second).
Constants would be:
- winding heat capacity (to have a good margin, copper wire mass could be used. Fixed, not fitted to data)
- motor casing heat capacity (either fitted or guessed based on steel mass)
- winding to casing conduction/radiation coefficients (whatever values fit good to data) - this should be the same for everyone
- casing to air - mainly convection depending on the air speed
- motor heat dissipation coefficient to outside case (for those without a sensor). This would be different depending on a cooling solution and would have to be adapted online (basically estimated temperature would have to be compared to temperature obtained from winding resistance - it doesn't have to be super precise - we just change parameter up or down if there is temperature discrepancy. It would be required to activate a motor phase for a second when not pedaling to take resistance measurement. Tiny battery energy loss.).
The only problem I see is that the uncooled motor cools a lot through radiation (ΔT⁴) and cooled via conduction (ΔT). Perhaps a different model could be selected in the configurator for people with heat pads. Single variable adaptation with a single known state is easy on controller because doesn't require memorizing time series.
......Or am I overcomplicating this? Because if we can (sometimes) measure winding resistance then we know the winding temperature, which is what we are really interested in. The big assumption is that the rider sometimes takes brakes from pedaling and we could actively measure the resistance.
We could also give try to measure Resistance dynamically knowing BEMF from cadence, etc. I have no experience with how reliable this would be. https://build-its-inprogress.blogspot.com/2019/10/motor-temperature-estimation-without.html
Anyway, we could give it a try if someone had a bunch of riding data saved.
I must say it sound interesting, but reading the linked blog, I think it isn't that simple and a lot of work for only an estimation of the temperature and I wonder if you can determine the resistance accurately enough with just 8bits.dzid_ said:
stratohunter
10 mW
- Joined
- Apr 12, 2022
- Messages
- 34
I built the following cooling structure with 1.6mm thick aluminum sheets. The only power tools used are an angle grinder and a drill. I selected soft 3000-series aluminum to ease the work. The first picture shows all the components.
To make the curvature for all these components I cut the aluminum sheet into strips then took the motor lamination to roll them into shape. The 4 small "tiles" which goes onto the motor lamination should ideally be roughly 2mm thick (my guess) but I stuck 0.5mm thermal pads onto 1.6mm ones to make them aligned with the shape of the motor. The middle "brace" was simply cut and rolled. The outermost "brace" was about 2cm longer than the middle brace to clip a temperature probe in place, and was ground into a wedged shape to adapt to the tapered motor cover, so that maximum contact was ensured. To make this part, I ground an aluminum strip into the rough cross section, rolled it into the curved shape, then ground more by trailing and error, e.g. try closing the motor cover and grind further if it won't close. The cover should be closed fairly easily to accommodate for thermal putty as described in the next step. I'd start with a 1~1.2mm thick al sheet if I'm making this part for the second time.
Then I pasted some 5 W/mK thermal putty between each layer of aluminum. Using too much would result in a hard time closing the motor cover.
Finally I drilled a hole on the bottom of the motor cover and inserted a temperature meter probe through it. This probe was wrapped with thermal putty and went into the space between the outermost aluminum brace and the motor lamination as shown in the last picture.
I've been riding with this whole setup for more than 3 months and about 1000 km in the hot summer, and it worked exceptionally well. I could continuously draw 12~14 amps or frequently hit 16 amps without hitting 95 degC in 30 degC weather. For cooler days I'd just keep my speed over 40 km/h and forget about it. Probably this is why I ruined my brass gear but that's another story.
To make the curvature for all these components I cut the aluminum sheet into strips then took the motor lamination to roll them into shape. The 4 small "tiles" which goes onto the motor lamination should ideally be roughly 2mm thick (my guess) but I stuck 0.5mm thermal pads onto 1.6mm ones to make them aligned with the shape of the motor. The middle "brace" was simply cut and rolled. The outermost "brace" was about 2cm longer than the middle brace to clip a temperature probe in place, and was ground into a wedged shape to adapt to the tapered motor cover, so that maximum contact was ensured. To make this part, I ground an aluminum strip into the rough cross section, rolled it into the curved shape, then ground more by trailing and error, e.g. try closing the motor cover and grind further if it won't close. The cover should be closed fairly easily to accommodate for thermal putty as described in the next step. I'd start with a 1~1.2mm thick al sheet if I'm making this part for the second time.
Then I pasted some 5 W/mK thermal putty between each layer of aluminum. Using too much would result in a hard time closing the motor cover.
Finally I drilled a hole on the bottom of the motor cover and inserted a temperature meter probe through it. This probe was wrapped with thermal putty and went into the space between the outermost aluminum brace and the motor lamination as shown in the last picture.
I've been riding with this whole setup for more than 3 months and about 1000 km in the hot summer, and it worked exceptionally well. I could continuously draw 12~14 amps or frequently hit 16 amps without hitting 95 degC in 30 degC weather. For cooler days I'd just keep my speed over 40 km/h and forget about it. Probably this is why I ruined my brass gear but that's another story.
Attachments
+1pxl666 said:
Indeed it will better, because the heaconductivity from aluminium is a lot better.pxl666 said:
But a lot better will be to insert a 1mm aluminium or copper heatsink plate (on both sides some heatpaste) between motor and motorcase.
In that case you get a better contact with the complete housing and not only with the thin casecover.
I have copied the next question from here to prevent fragmentation about the same subject
So you can run with higher power.
How much? That depends of the measured temperature.
I hope you can measure the temperature too.
Because the heat dissipation is improved, you keep the motor temperature lower, which has a postive influence on the function of the motor.GrampaPete said:
So you can run with higher power.
How much? That depends of the measured temperature.
I hope you can measure the temperature too.
Last edited:
stratohunter
10 mW
- Joined
- Apr 12, 2022
- Messages
- 34
pxl666 said:
Just measured them today. My "middle brace" is 162mm long, 33mm high and 1.2mm thick. My "outer brace" is 196mm long and 29mm high, and has a trapezoid cross section, being 0.4mm thick on one side and 1.3mm on another. But again that's rough handcraft and the motor casing is known to have variations, so take it with a grain of salt.
On a Polish forum about Tsdz2 I found a very nice comparision of the use of different materials.
It isn't the same as using the tsdz2 under different conditions like off road or long runs with high speed on flat roads, but because the measurements are all made under the same conditions you can see how much they differ.
The measurements are done by markkar
The heat conductive material is place between motor flanges and case, as in the wiki.
The heatsinks are placed outside on the case and casecover.
NB
With the aluminium shims, the stator is also covered with aluminium.
With heat pads this isn't the case, which is a pity for 1:1 comparision,
What is now: stock vs. as wiki vs. full alu covered.
The results:
Tsdz2 stock 81°C / 178°F max.
Tdsz2 with heat pads 63°C / 145°F max.
Tsdz2 with aluminium shims (stator too) 54°C / 129°F max.
Tsdz2 with aluminium shims and heatsinks 53°C / 127°F max.
As you can see the use of heat pads does help a lot (-18°C), but the use of aluminium shims is even better (-27°C)
Adding heatsinks for better airflow helps a bit, but the difference isn't that big.
SOURCE
It isn't the same as using the tsdz2 under different conditions like off road or long runs with high speed on flat roads, but because the measurements are all made under the same conditions you can see how much they differ.
The measurements are done by markkar
The heat conductive material is place between motor flanges and case, as in the wiki.
The heatsinks are placed outside on the case and casecover.
NB
With the aluminium shims, the stator is also covered with aluminium.
With heat pads this isn't the case, which is a pity for 1:1 comparision,
What is now: stock vs. as wiki vs. full alu covered.
The results:
Tsdz2 stock 81°C / 178°F max.
Tdsz2 with heat pads 63°C / 145°F max.
Tsdz2 with aluminium shims (stator too) 54°C / 129°F max.
Tsdz2 with aluminium shims and heatsinks 53°C / 127°F max.
As you can see the use of heat pads does help a lot (-18°C), but the use of aluminium shims is even better (-27°C)
Adding heatsinks for better airflow helps a bit, but the difference isn't that big.
SOURCE
ornias
100 W
- Joined
- Jul 18, 2021
- Messages
- 172
pxl666 said:
I see no problem with that (and his price of about 450Eur), as he does quite thorough research before shilling a product.
Though I see some potential improvements:
- Also add a conductive ring around the top area of the motor (like some users on this forum are doing)
- Increase the size of the "top plate", to include the area between the screws
- Increase the size of the "bottom plate" to form around the complete "flat section" on the motor (to add additional themal capacity and increase the area to for more efficient heat flow into the part)
- Cut away the plastic cover, like shown elsewhere, to expose the bottom of the motor housing to fresh air
Another area I think we can make progress is cooling the stator ends:
- I don't think the stator ends currently contact the motor housing.
- There are papers online about how stator end cooling can improve motor efficiency considerably
- By just adding some conductivity between the motor housing and the stator-ends (pads, plates+glue), combined with the top and bottom cooling plates. We could significantly affect cooling rate. As the heat from the stator ends gets vented right at the stator ends itself
I'm only interested in his comparision.pxl666 said:
Of course the results could be manipulated for commercial use, but his comparision looks legit
Exactly this is described in the wiki (step 2) tooornias said:
wouldn't the best solution to this to just fill housing with oil ? or machine fills from aluminium that ,with some thermal paste , ensures full heat transfer from windings to cover . I also considered drilling holes in both sides of cover to let the air flow but then motor would must have been sprayed with water repellent lacquer and I'm not really fond of these ...
ornias
100 W
- Joined
- Jul 18, 2021
- Messages
- 172
Elinx said:
Ahh yes this.
Sorry, i'm still working on restructing the wiki mess. I barely can find my way through the not-semi-duplicated content at this time :')
Anyhow, that's missing the top part though... that can also use some additional thermal interfacing.
ornias
100 W
- Joined
- Jul 18, 2021
- Messages
- 172
pxl666 said:
For the area between the motorhousing/case and stator/motor: No.
Engine oil thermal conductivity is about 0.145 and there isn't that much movement in that space with fluid.
base aluminium has a thermal conductivity of about 237
To give an idea: even quite basic thermal pads have a better thermal conductivity than the oil has.
This also explains why the user doing this already, still reports overheating-ish (frequent 70 degrees) whereas previously linked test reports better performance with heat-transfer plates.
However:
For inside the motor itself, this MIGHT be a solution. Assuming one can 100% seal the motor inerts.
Though one has to take into account that the bearings might not like this and this also might actually increase friction inside the motor as well.
Though most heat comes from energy losses in the stator windings, so the heat will be mostly concentrated there as well. If you've a higher heat buildup in the rotor, you, most likely, have more severe other issues or extremely great cooling on the stator
Has anyone tried cutting inlet and exit air ventilation holes in motor side cover? Cover would no longer be water-proof, but if you ride in dry conditions, air flow should control motor temperature. Or, perhaps cut round hole in face of motor cover? Openings could be covered over with a foam or felt filter material to keep out crud…
-
-
Attachments
Topic starter andrea_104kg did a sort of this in 2019 :wink: . He share his results there too.Piper J3 said:
This was before a more effective methode was published into the wiki:
-- heat conductive material between the motorflanges and case --
Replacing all airgaps by heat conductive material, like copper and add heatsinks outside the case for improving the airflow cooling, is imho the most optimal methode.
Similar threads
