Hi all,

I'm seeking some opinion on the best best way to warm milk for my rfid calf feeder? At the moment I have a drum full of water with a copper coil in it that the milk flows through, the water is heated by a fish tank heater (300w) and the milk comes out warm.

It seems a waste of energy to run the heater 24hrs so I would like to run the milk through a stainless steel tube that heats up to warm it. It would then only need to run while a calf is feeding.

As far as I can see I've two options for heating the tube? One is to wrap a heater element round the pipe and use PWM to control the heat output.

The other is to put the tube in a coil and use induction heating to heat the tube.

I'm going to use a couple of thermocouples to make sure the milk isn't too hot,(if they don't agree with each other everything will stop, don't want to harm the calves!).

I've done some basic sums based on maximum milk flow and temperature difference and the most power I could require is a bit less than 300w as this is absolute maximum I think 300w will be enough to allow for losses?!

This will all be controlled with a 18fxxxx PIC so I have pleanty of A/D channels and PWM avalible.

I've not had much to do with power switching before , to me induction heating the tube seems attractive, whichever way I go I'm going to need power transistors.

What do folks think? Basicly I'm looking for the pros and cons of each method. (or a new idea I've not thought of!)

Cheers Geoff

 

Have you considered Heat Trace cable?
Can actually be inserted in the pipe and is self regulating.
There are a couple of manuf out there HeatLine and Raychem, are a couple.
If you need to explore it I am sure they have an engineering dept that would tell you if it was viable.
Max.

 

Have you considered running the milk through a tankless water heater? That's basically what you're asking for, hot (warm) fluid on demand.

 

You would probably want to wrap the pipe with this.

 

Have you considered running the milk through a tankless water heater? That's basically what you're asking for, hot (warm) fluid on demand.

+1 There are many commercially available units at lots of different capacities.

Biggest problem I see with any solution is maintaining cleanliness. Water is easier to handle than milk!

I'm not convinced that your current solution isn't ideal already. If you're worried about efficiency, just insulate your heat reservoir. The shear thermal mass of the reservoir is a big safety factor to avoid any chance of hot milk to a calf.

 

I agree with wayneh. Just insulate your present water drum with some thick insulation which should give a low power loss. (You can check the energy usage with a Kill-A-Watt meter, if interested). That's no different then the way a standard water heater works and avoids the complication of an elaborate heater and control loop combination.

 

Thanks for the swift replies!

I think the heat trace cable may not be able to supply enough heat? The most powerful on the rs site is 60w/m, If I were going to use a larger bore pipe I could spiral it round but I was intending to use 6mm id pipe, so I dont think I could get enough around the pipe?

I had thought about a tankless heater, but as Wayne said I think keeping it clean would be the problem, most of them seem to be copper, which is fine for water but I would like to use sodium hydroxide solution to keep things clean and germ free, I think it might eat out the copper a bit?

Maybe I should stick with what I've got? It does work!

Thanks for the input, I'd never heard of heat trace cable before. Any more thoughts are welcome!

Cheers Geoff

 

The heated build beds on 3D printers use DIY heaters. One common method is to wrap the heated platter (in your case - pipe) with Kapton tape (Cheapest on eBay or Amazon) to electically insulate it. Then wrap with resistance wire (various types available on eBay).

I would use a ATX computer power supply - 500 watt should work fine with 20 amps on each of two separate 12volt output rails.

Two lengths of resistance wire giving 200 watts at 12 volts (two lengths of resistance wire giving 3 ohms of resistance wire at 4 amps each) = 200 watts each.

Be sure to crimp-type connectors for the terminals on the resistance wire - solder doesn't stick well.

 

I do quite like the sound of gophers suggestion! I've not played around with heating stuff before, I'm going to use thermocouples to sense the temperature, is it worth using pwm to control the heat or am I better just to write a closed loop routine using the input from the thermocouples to decide when to turn the heater on and off? I think I said before a pic will be controlling the heat so either method is avalible.
Cheers Geoff

 

I guess an LM35 is also an easy temperature sensing option. They are like voltage regulators that output 0.1 volt for every degree C. 105C = 1.05 V.

The only issue with all temperature sensing options is how fast they can respond to changes in environment - they all have a time constant (how long it takes the measuring device to heat up). I don't know how the TO92 package compares to your thermocouple.

 

I solved this problem the easy way; the cow heats the milk to the exact temperature required for the calf's comfort.

 

I solved this problem the easy way; the cow heats the milk to the exact temperature required for the calf's comfort.

That is amazing. How do you calibrate the cow's set point temperature?

 

The only issue with all temperature sensing options is how fast they can respond to changes in environment I don't know how the TO92 package compares to your thermocouple.

Time constant is not specifically named on the LM35 data sheet. After reading it all, my estimate of an LM35 which is held tightly against the pipe, and insulation is added all around the area, is 3 seconds. A thermocouple, being only a tiny dot where two metal wires are welded together, is a lot faster than 3 seconds. Does 15 seconds to 99% accuracy bother anything? If not, the LM35 is a viable choice.

I agree with post#5. The time constant of a tank of water is a big safety factor because it can't possibly go very wrong in 15 seconds. This makes both sensing devices a lot faster than the time required to make a significant error in applying heat. This method makes the brain work simpler, too. There is no need to pwm the energy source at a thousand samples per second when it takes 40 seconds for the energy to change the temperature of the tank by 1 degree. The concern about the efficiency of the tank of water is misplaced. I once did the calculations for a 30 gallon water heater with R13 added to the insulation that came with the tank, and the answer was $25 per year in heat loss to the air around the tank. Can you heat the milk with less than 30 gallons of warm water? Can you insulate the water tank to about R-20? Can you afford $2 per month in wasted energy?

OK. That's my two cents.

 

Thanks guys, I appreciate the input, the time factor had been nagging at me a bit, I thought a thermocouple would be quicker but had wondered if it would be quick enough to avoid harming the calf if anything went wrong? I think maybe I will 'polish' up my water tank system, it is nice and simple.

ramancini8 I solved this problem the easy way; the cow heats the milk to the exact temperature required for the calf's comfort.

He he, nice work, let me know when you've sussed out how to get 25 litres of milk out the calf leave it with the 2 it needs and put the rest in the tank!

 

Time constant is not specifically named on the LM35 data sheet. After reading it all, my estimate of an LM35 which is held tightly against the pipe, and insulation is added all around the area, is 3 seconds.
...

I do quite a few temperature control systems, and often use LM335 and LM35. In the standard TO-92 plastic pack they respond quite well, there's no problem clamping one to a metal object and keeping the metal object within a fraction of a degree, even using simple setpoint on-off control.

I've even done XTAL ovens using this type sensor to keep the temp within about 1/50th of a degree C, with a more complex temp control algorithm.

The big issue is the same in all temp control apps it's the thermal lag of the total system. Even though the temp sensor will respond quite quickly, if the heater is far away, or needs to first heat a big mass, it can take multi seconds from the heater being powered up to the temp rise finally getting to the case of the LM35. The time constant of the LM35 is quite small compared to the total system time constant.

If you can get the LM35 reasonably close to the heater you can get cycle times down to a few seconds even using simple on/off setpoint control around one ADC value.

 

Time constant is not specifically named on the LM35 data sheet. After reading it all, my estimate of an LM35 which is held tightly against the pipe, and insulation is added all around the area, is 3 seconds. A thermocouple, being only a tiny dot where two metal wires are welded together, is a lot faster than 3 seconds. Does 15 seconds to 99% accuracy bother anything? If not, the LM35 is a viable choice.

Wow, I always like a good estimate.
Top of page 6 shows response rate in air (3.5 minutes) and in oil bath (3.5 seconds) - I would agree that will be almost th same as close contact to a copper pipe.

My concern was mainly how big the heater element was and how much thermal momentum would be generated before the LM35 (or any sensing element) would be able to shut down the heater without it boiling the milk.

Sound like all is well sized and my concerns were unnecessary.

Cheers.

 

Thanks again guys, its interesting to think about even if I don't use it!

I've been reading through the post's and a idea has just occured to me, if I had a sensor picking up the temperature of the milk coming in I could make a good estimate of how much power I might need to put into the heater? I know how much milk is being pumped and at what rate because its measured out to the calf (closed loop system). The output sensing becomes less crucial then (still needed!), its a bit more complicated but would give a extra layer of saftey if I decide to use a pipe heater rarther than my current water tank setup! I have ir sensors to make sure there is milk in the pipe.
Cheers Geoff

 

I very much disagree. Big and dumb is safer than quick, and quick to go wrong when it breaks, and you KNOW it will eventually break. Still. it's your calf and it can squeal or moo or whatever they do when they get their tongue burnt.

 

I have experimented quite a bit with using PICs as thermostats with an LM34 as the sensor. I have an LCD so that I can continuously read the A/D value, and find that the LM34's response time is very fast, almost instantaneous. I don't foresee a problem in your application.

It is also my opinion that, despite the preference of the scientific community for Centigrade measurements, the LM34 is inherently more precise in absolute terms than the LM35, simply because one degree C is so much larger than one degree F.

As you are writing the code for the thermostat, don't forget to build in some hysteresis, unless you are planning to implement PWM.

 

Just be sure to program in some "safetys". Perhaps a completely separate over-heat detector circuit and something to detect, "power is maxed out and no heat is happening".