Minimalist precision thermostat - AT30TS750A

A while ago there was a thread that included a thermostat/heater for an incubator. I still do not know where the schematic came from, but I had the feeling that it was from a few years ago. I was thinking about it and wondered if I could do "better".

Here is what I have come up with so far:




So far, I have only bread boarded it.


...and that is without the 78L05 and the IRL540 is only turning on an LED. The AT30TS750A is a pretty impressive chip...and at only US$1.30. Yes, you have to program it, but in this day and age...so what?

I also checked out the the older DS1626 and the max31723.

The only heater I have available is one of these (12V and measured at .75 A mA, although the IRL540 should be able to handle considerably more than that in this design).

I think that I will try to solder up a test version today.

Very simple.
Not sure about the fet drive, is the chip o/p open drain pull down, and the 10k a pullup?, if so switching time will be slow, 1k might be better.

dr pepper said:

Very simple.
Not sure about the fet drive, is the chip o/p open drain pull down, and the 10k a pullup?, if so switching time will be slow, 1k might be better.

Click to expand...

The temperature sensor is rated at 4 mA, so 1k is a marginal overload. I don't see any reason why slow switching would matter. It'll only happen every few minutes.

I don't see any need to regulate the 12 V.

Appreciate the comments...working on it right now...

Diver300 said:

The temperature sensor is rated at 4 mA

Click to expand...

Glad to see you notice that. I looked long and hard to find out how much current the Alert pin (which is open drain and with programmable polarity) could sink. I too saw the testing specs at 4mA and figured...well I know it can do that...and I actually worked on the programming using a pulup resistor and LED connected directly to the alert pin so I could see the behavior of that pin.

As for the 'regulated', well I put that in the schematic because I am using a 12V lab bench supply, which is regulated. I suppose I would use a larger value for C1 if the 12V PS was 'less' regulated.

I also thought about using a zener with a 390 Ohm, 1/2W resistor for the 5V source to the chip. It draws so little.... I did some homework on the issue because I have often read that zeners as regulators can be problematic. After a good deal of reading (like here), I was all set to use one but changed my mind because 5V regs are so cheap and easy, I couldn't see how a zener would be 'simpler'.

The sensor is nice. When it powers up, it will use NVM stored values for high and low temperature registers. You can write them programmatically with or without writing them to NV memory. Up to 12 bit resolution for temperature (programable 9-12 bits with conversion time shorter at lower resolution). Also with shutdown mode and interrupt (or comparator) mode. It also uses a fault setting...so, for eample, if you have it set for an alert when the temp reaches a low limit (IOW to turn on the heat), you can program it so it takes not 1 'read at/exceeding limit' but 2, 4 or 6 such 'triggers'. I have it set to 2, ostensibly to avoid any noise.

You can specifically write to NV memory, but with a single command you can write current values for High/Low and configuration registers to NV memory.

You can also lock out changes, using a software lock and even a hardware lock. The latter prevents making changes to NV memory...ever again.

Although I am playing with it as a simple thermostat for a heater, I can easily see how it could be a valuable general thermal shutdown device as well.

I made an incorrect assumption, I thought using a micro that the system would use pwm at some frequency instead of just on/off.
If your only switching very slowly then 10k probably would be fine.

DrG said:

As for the 'regulated', well I put that in the schematic because I am using a 12V lab bench supply, which is regulated. I suppose I would use a larger value for C1 if the 12V PS was 'less' regulated.

I also thought about using a zener with a 390 Ohm, 1/2W resistor for the 5V source to the chip. It draws so little.... I did some homework on the issue because I have often read that zeners as regulators can be problematic. After a good deal of reading (like here), I was all set to use one but changed my mind because 5V regs are so cheap and easy, I couldn't see how a zener would be 'simpler'.

Click to expand...

My thought was that the heater doesn't need a regulated voltage. It would be quite happy on half-wave rectified mains, and significant voltage variations won't matter.

I agree that a 78L05 is about as simple as it gets. Zeners are often worse regulators and often use more power. You have to dissipate a lot of power in the resistor. However they are more robust against short-term overloads or spikes and against reverse polarity.

If you do use an unregulated supply, I would add a diode like this:-


That way C1 is only supporting the temp sensor and the regulator during voltage dips. If you don't do that, the heater will pull the voltage down which means you need a much bigger capacitor, and you risk overheating the capacitor with large ripple currents.

dr pepper said:

I made an incorrect assumption, I thought using a micro that the system would use pwm at some frequency instead of just on/off.
If your only switching very slowly then 10k probably would be fine.

Click to expand...

Understandable. It is a stand-alone thermostat...once you have programmed it with I2C (or, I guess using the factory per-programmed defaults). Of course, you do not have to use it standalone - it also functions as a run-of-the-mill I2C temp sensor.

Having built various temperature controlled boxes (biltong, incubator etc.) I find the most convenient heating elements to bo 50W 3ohm resistors - available very cheap from China. 12V into a 3ohm resistor is ~50W.

Mike.

I soldered the circuit to a board (below).



Did some very preliminary testing by placing the board (sensor side down) a few inches above the heating pad and watched the power supply Amps display. It seemed to work as it should - nice.

The AT30TS750A is socketed for easy removal and transport to a programming bread board to change settings as desired.

The screw terminal block on the side has one terminal connected to +12 and the other to the Source pin on the IRL540. This makes it easy to attached a polarized load, like a relay or a motor to open and close a vent (I know those would require a flyback diode).

The other screw terminal is for the power supply - I have an old laptop PS that should work wlel for further testing.

I need to try it out in an enclosed container with the temperature monitored and recorded to see how well it does. I realize that there are a number of relevant variables concerning the heating element and insulation and ambient temperature - still I would like to see some data. Not sure when I will get around to that, but I would like to see some results.

I did some testing this morning. Nothing rigorous for sure. Just slapped the heater in an old food storage container and attached the circuit to the the inside top. I also tossed a DS18B20 in there and measured the temperature over ~80 minutes. The setting was: Thigh=27C and Tlow=25C. Ambient was ~22.5C. I am good with the results below:



The "final" schematic is below (just added the screw terminals):



Also attached is the code I used to explore the chip and set the NVM in case others want to do some investigating as well. The code is absolutely not optimized and is, frankly, a little sloppy, but it worked for me to do what I wanted. I have no interest in writing a library or a GitHub entry.

All-in-all, the AT30TS750A is a pretty nice chip and I am glad to have added to my education, repertoire, skill set, whatever.

Looks Ok for the amount of parts.

An impressive result, what happens if you change the limits to be one degree apart I.E set Tlow to 26C.

Mike.

Pommie said:

An impressive result, what happens if you change the limits to be one degree apart I.E set Tlow to 26C.

Mike.

Click to expand...

I don't know how impressive it is - it is what I expected, or at least hoped for, given the small size of the container and how little ventilation was present - I didn't add it in the post, but the ds18b20 is read every 10 seconds. That graph is from the built in serial plotter in the Arduino IDE.

I have put it away and don't want to test that specific scenario, unless it is particularly important to you. I would think that a great deal depends on the characteristics of the heater.

Would it chatter? I suppose it could in the sense that it is going to switch the polarity of the alert pin accordingly whenever the comparison meets the requirement.

As I recall, the data sheet says only something to the effect that Thigh has to be greater than Tlow in order for alerting to work properly. With 12 bit resolution, you could have a .125 degree C difference between Thigh and Tlow. Additionally, you could set the fault number to 1 (I had it at 2) so that one detection at/exceeding a limit triggers the alert. I don't recall a specific clock speed - meaning how long between conversions in free running mode. Conversion time is in there though as I remember specifically that lower resolutions are faster.

Just as an aside: I have no specific use for that circuit. I just wanted to investigate it, as I said. Now there is a small box that contains the circuit and a manila file folder with my notes and the schematic and data sheet.

I do have some interest in making a specialized food dehydrator but I would want better control - something analogous to a profile that you see with IC ovens. But, I would also like to have some humidity control (likely de-humidification). It is on the very long list.

BTW: I don't know about you folks, but when I do a project - even a simple one like this - I get junk spread out everywhere - it always takes me an hour or more to clean up and put everything back - I need a maid.