EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: cosmos on July 14, 2013, 08:26:27 pm
-
Looking for ideas:
I am thinking about stabilising a low cost reference (REF5050 maybe for starters) by putting it in an oven (reducing 3ppm tempco)
I know there are better internally heat regulated references but I want to see if there might be a simple low cost circuit that can improve the simpler circuits a bit.
Ideally I want to have 2 extra wires (isolated power so it does not interfere with the ref) enter the regulated oven, a few simple actives (opamp + sensor?) and a few resistors for heating.... the absolute temperature it regulates to is not so important (range ~45C) but it needs to be stable and repeatable (say to 0.1C to have a starting point for discussion).
Any good ideas or sources I should look at?
Good reasons not to try it at all?
-
Pid, pid and next pid
-
Use a transistor as heater and temperature sensor. If you get something (with a lowish hFE) that will stand 30mA through the base you have a two wire heater 20mW heater, just switch off the heater current and measure the Vbe - that will tell you the temperature.
You can do the same with diodes, just not as nice a package.
-
I did that with a crystal years ago to make it more stable, using a NTC thermistor and a NPN transistor to make a simple heater. Thermistor got lower resistance as it heated and stole base current from the transistor, which was biased when cold to about 100mA, when it was hot ( about 60-70C) it then was at about 10mA. Not very fast, not very accurate temp wise but it worked well to keep at the set point with a 7812 regulator supplying the current for it. It worked well to keep the crystal stable, and as a plus had no extra noise. I just taped it to the crystal with the NTC on top and covered it all in a small piece of shaped styrofoam and a cardboard outer box to keep it.
-
You can try a ready to use crystal heater such as:
http://www.kuhne-electronic.de/en/products/crystal-heater/qh-40-a.html (http://www.kuhne-electronic.de/en/products/crystal-heater/qh-40-a.html)
I've build an LM399 based 10V buffered reference (see datasheet from LT) using cheaper 5ppm/K smd resistors with the crystal heater on top of them for thermal stabilization.
After preaging the complete circuit (300h @ 120°C) and afterwards soldering the electrolytic cap the circuit is now quite stable (started at 10.00179 and settled at 10.00182) since many weeks (24/7). All I can see is an influence by humidity (10.00181-10.00182V @ RH<40% and 10.00182-10.00183V @ RH>50%) but can't tell if this is due to a sensitivity of my board or the HP34401A. The temperature in the lab is very stable (<±1K).
In most cases the supply voltage of the heater fits the supply voltage of the reference. The heater is already assembled on a ceramic substrate, perfect for mounting it to the reference package with thermal glue.
-
Any good ideas or sources I should look at?
Good reasons not to try it at all?
If using REF5050 I would use the temperature output of the reference as temperature sensor.
A cirquit for a LT1019 can be found in LT AN42 page 15 (figure 66).
since the REF 5050 does not have a internal heater resistor you will have to place
several SMD resistors under and around the reference pins.
By the way: in newer datasheets the heater pin is no longer mentioned on the LT1019. The heater current will shift the ground pin internally in the range of around 1mV and thus affects the output voltage unintentionally.
But with a additional ground pin for your heater resistors this will be no issue for external heater resistors.
Of course you will not reach the stability of a hermetically tight case where the heater and the sensor are on the same chip. And with plastic cases the reference will be affected by humidity.
With best regards
Andreas
-
If you want to use the LT1019 with internal heater, more info on pins etc is in this datasheet http://www.farnell.com/datasheets/60321.pdf (http://www.farnell.com/datasheets/60321.pdf).
In a DIY project where space is no limitation, I find it more attractive to separate the temp-sensor/heater from the device that is to be kept at constant temperature. You can then use a cheap temp-sensor IC driving a simple resistor heater and obtain stability at around 0.1 degrees (sensor output is 10 mV pr degree) http://www.ti.com/lit/ds/symlink/lm335.pdf (http://www.ti.com/lit/ds/symlink/lm335.pdf). As there is little or no thermal contact between the heater and the heated device this arrangement depends on the thermal isolation of the inside of your (multilayer) box from the varying outside environment temperature.