Controller for phase change units

[Dafut]

Hi guys,

Me and a friend are thinking of developing a controller for phase change units. We have some features in mind but maybe you guys have soms extra ideas.
So what kind of functionalities would you like to see on it?

Functionalities we want to have in the controller so far:

-Display, ofcource to show info on it, like temps.
-Ability for up to 4 temp sensor
-Ability to first let the unit reach a certain temp and than boot up the computer system.
-Monitoring program for the computer, cross OS compatible with mac windows and unix. In the
program you will be able to control when the second (and/or third) stage starts on a cascade. You will also be able to log data.

We haven't been able to find an accurate and affordable temperature sensors yet (one that will go beyond the -100 degree).
So if you know one we could use in the controller, please let us know.


If al things run great we are planning to make the controller available as prebuild kit in the near future.

[ultralo1]

Here is a good reference for choosing the correct temp sensor
http://www.omega.com/temperature/tsc.html

[Moc]

PT100 probes...

[mytekcontrols]

We haven't been able to find an accurate and affordable temperature sensors yet (one that will go beyond the -100 degree).

Since you guys are rolling your own controller, which I presume you will be using an embedded processor and writing your own operating code, I would recommend going to a thermocouple front end. It wont be cheap, but it will be extremely flexible and reliable. And what it takes in extra cost to create the front end, it'll make up for in the ability to use inexpensive TC wire for the sensors (for sub -100C temperatures; do a cost comparison of TC wire vs. RTD's, or specially calibrated low temperature sensing diodes).

Here is a device I am using to read and linearize the TC's: SEM203/TC Signal Conditioning Module. It costs $65 in single quantities, but has an excellent temperature range (down to -200C), stability, and is capable of reading types T, K, and J thermocouples.

It requires a 4-20ma input, so I use a RCV420JP 4-20ma Receiver IC to translate this to 0-5VDC which is easily read by most embedded processors such as a PIC16F877A's 10-bit analog inputs. It costs $7 in single quantities. (you will also need a negative supply for the RCV420JP such as a SRS-0505 DC Converter Module)

Of course you will also want to have multiple TC inputs (I believe you mentioned 4). This will require a MUX chip with dual differential analog I/O to switch between the TC pairs (don't want to just switch one wire due to crosstalk problems). I use a SN74HC4852N IC for this purpose, which is a 4x2 channel analog MUX IC. It costs $0.64 in single quantities. You'll need to provide an isolated power supply for the MUX (+/- 5VDC), this can be provided by a DC to DC converter module be run off of the same +12VDC supply that runs the SEM203/TC module such as a NMV1205SC which goes for $11 in single quantities.

WARNING Don't try to share the same negative supply used for the RCV420JP. You need a completely isolated power source for the TC MUX circuit to work properly. You will also need to implement an opto-coupled interface from the embedded processor to the MUX address lines (TC selection). Yes everything needs to be isolated when dealing with grounded TC's. The SEM203/TC module already has built-in isolation for it's single TC input.

All together we are talking about $90 to create this fully isolated 4 channel TC to 5V logic front end. This seems pricey, but also keep in mind that it can be easily expanded to 8 or more temperature inputs by simply adding additional SN74HC4852N's for $0.64 per multiples of 4 channels a pop. Also it will yield +200 to -200C capability with half a degree resolution (w/10-bit analog input).

Here is the basic TC front end circuit shown without the MUX switching (Type "T" TC shown, but can be "K" or "J" types):

[SexyMF]

I've been looking at sensors to redo my controller. Thermocouples are it when dealing with -80 or colder temperatures. Unfortunately the CJC and amplifier ICs are relatively expensive on top of the probe itself.

Consider your market. Generally cascade builders have money to play with but single stage builders do not.

So, if you expect to sell these, are people going to pay the few hundred bucks to make it worth your while?

I've decided to try an RTD sensor this time (over a -80 -> 150C thermistor) and aim for the single stage unit only, with evap and condensor temps.

Best of luck.

[danavyor]

Thanks for all the comments so far.

As we've been looking at the several different options for gathering temperatures. We decided it would be wise to create two different versions. One for cascade systems and one the simpler single stage machines.
As the single stage systems won't go beyond the -70 degrees anyway, we can relay on a cheaper sensor system for those. And the TC front end described by mytekcontrols would be a good solution for
the more complex cascade systems.

The developing of the hardware won't be a problem, as we are embedded engineers. It's more the question are there people interested in having such monitoring and controlling modules for there coolers.
And what are the key features and functionalities they want to have in the controller.

[teyber]

Thanks for all the comments so far.

As we've been looking at the several different options for gathering temperatures. We decided it would be wise to create two different versions. One for cascade systems and one the simpler single stage machines.
As the single stage systems won't go beyond the -70 degrees anyway, we can relay on a cheaper sensor system for those. And the TC front end described by mytekcontrols would be a good solution for
the more complex cascade systems.

The developing of the hardware won't be a problem, as we are embedded engineers. It's more the question are there people interested in having such monitoring and controlling modules for there coolers.
And what are the key features and functionalities they want to have in the controller.

I vote for a function to be able to start the phase change without having to have it hooked up to a computer. I think the main issue for your customers is cost

[lism]

The things you mention you want in your own controller is already packed with the controller of the Vapochill XE.

I'm not sure if rebuilding this controller is what you need. (Sup Dafut :P )

[danavyor]

The things you mention you want in your own controller is already packed with the controller of the Vapochill XE.

I'm not sure if rebuilding this controller is what you need. (Sup Dafut :P )

We've looked a several different controllers on the market. And these are all related to either n00bs or the semi pro class. The only controller worth mentioning is the G-Froster. But that one has only two sensor inputs and 'only' goes down to -150. And G-Froster hardware can't be mentioned of being sleekly developed

We are aiming for the more professional class with heavy loads and low temperatures. But we most definitely want to create a prototype at first. So we can find out what's worth developing.

And of course it's a hobby project in the first place. But if the controllers are worth manufacturing at scale, we'll be seriously thinking of bringing them on the market.

[ultralo1]

How about a hot gas bypass control to help eliminate the cold bug issues?
A differential control ability for the liquid cooling people out there?
Future ability for expanding the number of sensors? This would be great for autocascades.
Possible alarm setpoints?
Maybe do soft ware that would be temp range specific, the needs of a single stage unit are different than that of a cascade which are different that AutoC.

Keep in mind I am not a OCer, i am a refrigeration guy. Some of these suggestion comes from that point of view

[teyber]

We've looked a several different controllers on the market. And these are all related to either n00bs or the semi pro class. The only controller worth mentioning is the G-Froster. But that one has only two sensor inputs and 'only' goes down to -150. And G-Froster hardware can't be mentioned of being sleekly developed

We are aiming for the more professional class with heavy loads and low temperatures. But we most definitely want to create a prototype at first. So we can find out what's worth developing.

And of course it's a hobby project in the first place. But if the controllers are worth manufacturing at scale, we'll be seriously thinking of bringing them on the market.

i think if you take this expensive you will miss your market. Builders don't want to include a $150+ controller most of the time, and a cascade controller, i don't see hardly any cascade builders.

I think your best selling product would be a cheap, simple maybe -50c or so, slim pickings controller for the $75-125 ballpark.

cheers

[ruffus]

i tend to agaree with teyber i know that my self have just started building units for other people (not just for my self) and all of them would love to have a controler but yet none of them want to but out the $165 or so for one so they end up just asking for a rocker switch as it only cost $2. but if i could come up with something cheaper i bet i could get them to go that way