How to keep part at constant temperature

Looking at the specs at:

Page 9 of this data sheet states: TEMPERATURE OUTPUT AND CALIBRATION It is common practice to temperature-calibrate gyros to improve their overall accuracy. The ADXRS401 has a temperature-proportional voltage output that provides input to such a calibration method. The voltage at TEMP (3F, 3G) is nominally 2.5 V at 27°C and has a PTAT (proportional to absolute temperature) characteristic of 8.4 mV/°C. Note that the TEMP output circuitry is limited to 50 ?A source current.

Is this not enough ??

Do you think this is the right device for your application ??

Also: I would think a couple of power resistors inside a small metal can would help. An 8-pin cpu could make a stand alone temp controller.

Simple, but why ??

donald

If you feed back on each gyro's temp sensor, you can use a local resistor to heat each gyro chip. The loop will probably be tricky to stabilize. Use duty cycle modulation to drive the heaters to avoid e^2 nonlinearity problems. I like to glue a thin pcb to a big aluminum plate, slap a Minco heater or a few power fets on the bottom and a surface-mount thermistor on top, close that loop at 60C, and populate the board with surface-mount parts. That rig has good loop dynamics.

You could also just measure the temperature and apply a fudge factor to the output signal.

But I don't think you can get anywhere near usable stability for your application (the inertial navigation thing, right?) even if you do temp regulate these parts. Probably not by 3-6 orders of magnitude.

John

l to

Is this not enough for what? In the datasheet it says that the amount of sensitivity varies with temperature. The exact amount that it changes varies part to part. So I can't just read in the temperature and extrapolate what the actual rate out should be, unless I calibrate each part that I use.

I am not sure. I need a high accuracy low cost (

I don't understand why I wouldn't be able to keep it fairly stable. I mean - outside temperature should be relatively constant - so it should be able to slowly ramp up the duty cycle to the heating resistors till they get it hot enough. I've never made such a device so I'm speaking entirely theoretically, though.

Except that according to the Analog datasheet - if I'm reading it right

- some parts keep the same sensitivity when temperature changes. So I'd have to calibrate it for each part - which would be difficult and annoying.

Yup - same IMU project.

Thanks,

-Mike

I've used one of the earlier devices (301?) and temperature compensated it. My procedure was to let it warm up in a draft free environment, capture a long average (with no rotation) then turn it 180 degrees in a jig, careful not to go too fast, and measure that turn. Then I'd do it again after putting it in the fridge for an hour, and then in the freezer. The result was pretty good. Far, far better than not compensating it.

One problem I found was that the device self-heats farily significantly on startup, and the thermal conductivity of the die was such that the actual gyro part would be warmer than the temperature output indicated until the part stabilized thermally, which took 5 or more minutes. Since my application was a backup (normally off, battery powered) flight instrument, that behavior was really annoying.

How about attaching the multiple MEMS to the different faces of a cube a thermostatize this cube with a peltier ? A hundredth of a degree is doable with an analog solution.

Rene

Un bel giorno Mike Noone digitò:

I did this with an accelerometer. I glued a SMD resistor (2512 or bigger, I don't remember) and a NTC above the component and covered the three components with silicone. The control loop wasn't too hard to set; fortunately in my case the ambient temperature was changing very slowly, and a PI algorythm with a very mild integrative part was enough.

However, I agree with the people that told you to give a try with the temperature compensation formulas. In my case the component didn't have neither the internal temperature sensor nor the specified calibration formulas, therefore I hadn't chance.

Yes, you can keep it stable, but the amount of energy wasted would be significant. I built a insulated cube of 6 inches to keep the inside temperature at 40C (15C above environment). My rough estimate is at least 3 to 4W energy wasted. 60C is not a simple task.

Have you checked the power budget on this flight instrument?

Yes, temperature compensation table would be much easier.