FW: Integrated Temperature Sensor
Michael A. Nixon
One more time – with feeling…
From: Michael A. Nixon
[mailto:mni...@teknastructures.com]
Sent: Tuesday, December 30, 2008
6:43 PM
To: 'noha...@googlegroups.com'
Subject: FW: Integrated
Temperature Sensor
Posted properly.
Mike
From: Michael A. Nixon
[mailto:mni...@teknastructures.com]
Sent: Tuesday, December 30, 2008
2:40 PM
To: 'Per Hassel Sørensen'
Subject: Integrated Temperature
Sensor
Per,
I would like to start with the assumptions for this sensor:
1 sensor or multiple sensors (just one sensor mounted on the stator frame or one sensor per phase) – Assuming one sensor and uniform heating as a baseline assumption.
Put all the “interpretation” of the sensor data into the processor board and minimize electronics in the hub motor.
Minimize wire connections to the hub motor.
Temperature Range - -55F to 125F (I know – I’ll have to start using that C thing with you Europeans…)
Magnetic Environment - ?
Moisture Isolation – Shouldn’t be any larger problem than the stator windings in general. How to specify?
Shock – Sensors I have used are solid state so the sensor electronics should not be the limiting factor on this specification.
Based on this I am thinking of a one-wire digital temperature sensor that I have used in the past from Maxim. I’ll get some specs and samples.
I almost took a job offer from Schlumberger here in the US a few years back in much the same field – downhole well evaluation electronics.
Mike
From: Per Hassel
Sørensen [mailto:perh...@gmail.com]
Sent: Tuesday, December 30, 2008
12:48 PM
To: Michael A. Nixon
Subject: Re: Links to reference
material
Michael,
You have a very impressive background in advanced electronics and research, I am certain you will become a very valuable contributor to this project!
There is one task that you might want to have a look at immediately, if you feel it is suitable for you:
Since the development team are spread out it wise to keep the controller design identical to the developer kit as long as possible, but one peripheral needs to be added to this development base. We need to monitor winding temperature inside the motors to avoid overheating the magnet wire insulation. This is very important for reliability and performance issues - we can put more current into the copper when needed if we know exactly when to back off.
We need to find a suitable temperature sensor and that detail should be settled as soon as possible since it needs to be integrated into the motor windings. This sensor must be able to withstand the harsh environment (temperature, moisture, shock/vibration) plus the electromagnetic fields inside the motor. It also needs to be a small and low cost solution that is easy to interface.
If this sounds interesting and you have some spare time, it would be very helpful, and we will of course help you as much we can! Please let me know how you feel about this task.
While I am writing I might as well ad a few words about myself:
I have a BsC in automation from Bergen University, Norway in 1988
I have been working with PC's and other microcomputers for 20 years until 1999 when I moved into power electronics and electromechanical design.
I now work in R&D for a Norwegian company that makes downhole, subsea and topside instruments for oil & gas production. That measn I have acess to very good tools for such a development project.
Kind regards,
Per
PerBear
On 31 Des, 01:45, "Michael A. Nixon" <mani...@hotmail.com> wrote:
>1 sensor or multiple sensors (just one sensor mounted on the stator frame or one sensor per phase) – Assuming one sensor and uniform heating as a baseline assumption.
One motor sensor goes to the associated ESC, and the two remaining
goes to the master MCU. I agree on the assumption on uniform heating.
But as you can see from the temperature simulations of the motor
casing (in the file area), there is a slightly higher temperature
close to the wire inlets due to less heatsink pins resulting in
slightly less cooling. It could be wise to put the sensors on the
motor windings closest to the wire outlet. It will also give the
benefit of short sensor wires reducing noise pickup.
> Put all the “interpretation” of the sensor data into the processor board and minimize electronics in the hub motor.
is high. By the way, I would just like to mention that the nohassel is
not a hub motor to avoid any confusion.
> Minimize wire connections to the hub motor.
balanced or screened/coaxial/transmission line is needed - I believe
we need to sort this out during the first tests. It is also possible
to find the coil temperature by measuring resistance in the winding.
That would really minimize wire connections!
> Temperature Range - -55F to 125F (I know – I’ll have to start using that C thing with you Europeans…)
can use google to translate between F and C - enter 125f in c and you
get the C value).
Especially during testing/debugging we must be able to follow
temperature up to a minimum 125C, preferably 150C. This is socalled
extended automotive temperature class ("under bonnet"). The magnet
wire insulation handles a maximum of 180C, limiting maximum winding
temperature to 150C should give us some margin.
DoE in the USA has done a nice reverse engineering job on the Toyota
Prius main motor and has made a thermal model of the windings in that
motor. In that report they mentioned that the motor winding
temperature was limited to 174C by a cutout mechanism but I have not
found out how the Prius measures the winding temperature, only how DoE
did it (they used thermocouples). You may want to have a look on the
report, it can be found here: http://www.ornl.gov/~webworks/cppr/y2001/rpt/122586.pdf
> Magnetic Environment - ?
The sensor needs to be noise immune and it must not radiate/conduct
electrical noise from motor or controller. This is maybe the biggest
challenge for the temperature sensor peripheral.
> Moisture Isolation – Shouldn’t be any larger problem than the stator windings in general. How to specify?
same laquer that is used in AC motors and transformers and oven
hardened after dipping.
> Shock – Sensors I have used are solid state so the sensor electronics should not be the limiting factor on this specification.
handle everything the motor itself handles.
> Based on this I am thinking of a one-wire digital temperature sensor that I have used in the past from Maxim. I’ll get some specs and samples.
range and with non-conductive housings since metal housings could show
wrong temperature since they can get warm due to the induced currents
from the stator fields.
Kins regards,
Per
Michael
>
> Agree - there is not much room inside the motor, and the temperature
> is high. By the way, I would just like to mention that the nohassel is
> not a hub motor to avoid any confusion.
>
> > Minimize wire connections to the hub motor.
>
> It could be that the environment is too noisy for single lines and a
> balanced or screened/coaxial/transmission line is needed - I believe
> we need to sort this out during the first tests. It is also possible
> to find the coil temperature by measuring resistance in the winding.
> That would really minimize wire connections!
>
digital signal out of the sensor, it would minimize the "filtering"
needed. Of course, if there is so much noise that it masks the
digital output, that would be problematic.
On lead length/quantity - with the one wire sensor, all four can be
placed on a single data line so only one (shielded) data line exits
the motor. Each sensor has a unique "address" on the bus.
On measuring winding resistance - I am no motor designer but doesn't
the Back EMF (a dynamic variable) affect the observed resistance?
Certainly there could be enough processing to account for these
variables but is that a first-order consideration?
> > Temperature Range - -55F to 125F (I know – I’ll have to start using that C thing with you Europeans…)
>
> The magnet wire will become much warmer than 125F - or about 50C (you
> can use google to translate between F and C - enter 125f in c and you
> get the C value).
> Especially during testing/debugging we must be able to follow
> temperature up to a minimum 125C, preferably 150C. This is socalled
> extended automotive temperature class ("under bonnet"). The magnet
> wire insulation handles a maximum of 180C, limiting maximum winding
> temperature to 150C should give us some margin.
> DoE in the USA has done a nice reverse engineering job on the Toyota
> Prius main motor and has made a thermal model of the windings in that
> motor. In that report they mentioned that the motor winding
> temperature was limited to 174C by a cutout mechanism but I have not
> found out how the Prius measures the winding temperature, only how DoE
> did it (they used thermocouples). You may want to have a look on the
> report, it can be found here:http://www.ornl.gov/~webworks/cppr/y2001/rpt/122586.pdf
>
operating range.
> > Magnetic Environment - ?
>
> The sensor needs to be noise immune and it must not radiate/conduct
> electrical noise from motor or controller. This is maybe the biggest
> challenge for the temperature sensor peripheral.
>
> > Moisture Isolation – Shouldn’t be any larger problem than the stator windings in general. How to specify?
>
> I think we should have some form of conformal coating - probably the
> same laquer that is used in AC motors and transformers and oven
> hardened after dipping.
>
> > Shock – Sensors I have used are solid state so the sensor electronics should not be the limiting factor on this specification.
>
> Agree. Unless it is mems or quartz sensors, they should be able to
> handle everything the motor itself handles.
>
across a transitor junction to measure temperature. I will have to
find out the specific methods used.
> > Based on this I am thinking of a one-wire digital temperature sensor that I have used in the past from Maxim. I’ll get some specs and samples.
>
> That sounds interesting, can you get them for extended temperature
> range and with non-conductive housings since metal housings could show
> wrong temperature since they can get warm due to the induced currents
> from the stator fields.
PerBear
I just went through the DS1822 datasheet. I would personally prefer a
higher max limit and lower cost, like LM73 from National semi, but I
think the main issue right now is how well it works glued to the coil
while the motor is working hard . The electromagnetic field is strong
and has a high dV/dt, so I belive testing is required.
Do you have a brushless RC motor & ESC that you could use for testing?
Kind regards
Per
Michael A. Nixon
take a look at a good inrunner. Have any recommendations?
Per Hassel Sørensen
DC bus voltage. If you have a fast ESC and 40-50V battery voltage it
should be fairly similar.
Per
2009/1/2, Michael A. Nixon <man...@hotmail.com>:
Michael
Sorry it has been a few days. I ordered a new ESC (Castle Creations
Phoenix HV 110 - 12s LiPo controller) and an AXI 5345 motor for a test
rig. In addition, I have order samples of both the LM73 and the Maxim
parts. I'll let you know how things proceed.
M