Voice Coil

[TheReaper]

I've seen several posts on this site regarding voice coil temperature measurements, but haven't been able to find anything that solves my problem.

I'm trying to measure the temperature of a voice coil in a non-invasive manner, i.e. no holes drilled or thermocouples added to the actuator (the voice coil is sealed within a metal housing), and have been unable to get the accuracy I desire using a well known method.

The method involves monitoring the change in resistance of the actuator and calculating the change in temperature/new temperature based on the temperature coefficient of resistivity of copper. My design involves wiring a 0.1 Ohm power resistor in series with the device and and measuring the voltage dropped across it. With a known source voltage, known power resistor value and known voltage dropped across the resistor, the resistance of the voice coil can be inferred, but upon comparing the results obtained using this design with data captured using a thermocouple in contact with the voice coil a 15%-20% error is seen.

The amplifier circuit passed inspection (100mV input - 4V output) and the resistance of the power resistor holds steady at 0.1 Ohms, but I am still unsure where the error may lie.

Thank you for any ideas/suggestions!

[forr]

Hi,
I have plans for such an apparatus.
I think to feed the driver through an amplifier, AC coupled using a cap, in parallel with a directly coupled DC current source, and then measure the heavily low-pass filtered voltage across the voice coil which would be almost DC.
As tempco is about 0.4% per °C, the temperature is easily deduced.
There is an article by Keith Howard in Stereophile on the subject.

[TheReaper]

Thanks for your reply.

I've read that article (and at this point probably most articles on the subject...) and considered that method, but before I resort to building any more circuitry I want to make sure that the method works and is accurate. At the moment I'm using a sine wave at 1kHz as the excitation source for the actuator, having measured the 1kHz impedance for the device using a databridge (4-wire method) and knowing that 1kHz is in the linear region for the actuator - surely it should be safe to assume that the temperature will have equivalent effects on the voice coil in this region as it would at ~DC?

Thanks again for your response.

[DSP_Geek]

Originally Posted by TheReaper

Thanks for your reply.

[...] At the moment I'm using a sine wave at 1kHz as the excitation source for the actuator, having measured the 1kHz impedance for the device using a databridge (4-wire method) and knowing that 1kHz is in the linear region for the actuator - surely it should be safe to assume that the temperature will have equivalent effects on the voice coil in this region as it would at ~DC?

Linear region - do you mean unaffected by other impedances such as the transducer resonant peak or series inductance? 1 kHz is well into the inductive region for woofers and midbass drivers.

You may wish to measure the driver impedance across the working band and tune your measurement frequency to the minimum impedance point.

[TheReaper]

DSP_Geek, thanks for your reply.

Based on impedance sweeps of the device I know that 1 kHz is within the linear region for the impedance just after the resonant peak of the device. What I forgot to mention in my original post was that this is a full range flat panel device to which I am referring (apologies for quite possibly posting in the wrong place... I had seen VC temperature based threads in this section and thought I might be more likely to get advice than trying to find the proper home, possibly foolishly...).

I have designed a 6th order Butterworth LP filter in LTspice that appears to have the desired response (120dB/octave cut-off after 1Hz), with which I may be able to isolate the ~DC component of the response and calculate the change, but in the mean time I've been trying 1kHz excitation as the device has a similar impedance at 1kHz as it does when measured as Re. I do like your idea for tuning the input signal and may attempt this.

Cheers