Ice Detector Design

[ADWSystems]

I'm looking to build something that will detect when I have a layer of ice build up.

I'm thinking of using an LM34DZ Farenheit temp sensor hooked to a resistor with a -Vs supply (as shown in the datasheet), so I can read 0 deg F and below. As the ice builds out into a room temperature area, I think the temperature will stay around 60-70 degF until it builds out enough to actaully touch the sensor at which point the readings will drop to 0. The IC is in a TO-92 package. I'm thinking of orienting it such that the top is pointed toward the ice build up, thus preventing the ice from reaching the leads.

What are your thoughts?

 

[crutschow]

It sounds like that should work. But the temperature will drop to 32F (0C) or below when it touches ice.

Is this ice build-up on the inside of a window?

 

[ADWSystems]

Obviously it was late when I posted that (11pm). Oops. No it's for detecting ice build up on the coils of a dehumidifier. I had to install a filter in front of the coils, so I can't see them any more. I've been using a mechanical timer to cycle it to melt the ice. But I'm thinking of connecting the temp sensor to a comparator to drive a contactor to turn on/off the dehumidifier.

That will change the fact that I won't need the -Vs either since it will operate between 30 and 70 deg F. I'm thinking I'll add 5x gain to amp the signal from 300-700mV to 1.5-3.5V. If using a 5v supply, I might as well use some of the range.

 

[Diver300]

You don't need to go down to 0 °F. Freezing point is 32 °F so the LM34DZ will output 320 mV at freezing point. It can't output less than about 50 mV, but that is 5 °F which is far colder than you are interested in, so you don't need the negative supply.

 

[ADWSystems]

It was 11pm when the idea occurred to me. I usually work in deg C and forgot (at that hour) the freezing was 32 degF. Your correct, I won't need the negative supply.

Anybody think this won't work?

 

[Diver300]

I didn't see that you had said in post 3 that you wouldn't need a negative supply.

It should work, but the voltage changes are quite small so noise might be a problem.

 

[ADWSystems]

I was thinking of an IA Opamp with 5x gain. Instead of 320 and 600-700mv, I'll have 1.6 and 3-3.5V. I have some IAs from another project, but I'll have to figure out if they can accept 300mV in. I don't remember off hand which parameter I need to check on the datasheet.

 

[crutschow]

You need to look at the common-mode voltage range to see if it will work with a single supply down to within 300mV of the negative (ground) supply.

The task likely requires a rail-to-rail type op amp.

 

[Brevor]

I have a refrigerator that does exactly this. It has a temp. sensor next to the evaporator coil that triggers the defrost cycle.

 

[ADWSystems]

Apparently. Most IAs I've found that aren't rail-to-rail are (V-) +1.0V and then some.

For achieving the 5x gain, what is the best mthod? Single transistor, OpAmp, or IA? I can't remember why I can't use a simple single transistor to get my 5x gain.

 

[Mr RB]

Obviously it was late when I posted that (11pm). Oops. No it's for detecting ice build up on the coils of a dehumidifier.
...

As another possible solution, I would attach the sensor to the coils, then use a closed loop temperature control that keeps the coil always above 0'C, probably around 2'C.

It should give exactly the same result, that the dehumidifier runs at the maximum possible power level while never allowing ice to form.

 

[alec_t]

I can't remember why I can't use a simple single transistor to get my 5x gain

You can use it. Vbe for a transistor is ~600mV but drops at ~ 2mV per degree C. So the drop is a significant fraction of the Vbe voltage and would normally be considered an error. Since in this application you don't need to know the actual temperature except when it is zero Centigrade you could just use this Vbe drop to detect icing and omit the LM34DZ.

 

[ADWSystems]

I was thinking of replacing the LM34 with a transistor. I was wondering about the design trade-offs between using a transistor, vs OpAmp, vs IA. I'm drawing a blank right now.

 

[crutschow]

A transistor has a significant Vbe temperature coefficient which means it's DC output will drift with ambient temperature changes.

An IA is used when you need a differential input to cancel common-mode voltages (which you don't have in your application).

A good way to amplify the LM34 signal using only one supply is with a rail-to-rail type op amp configured as a non-inverting amplifier.

 

[ericgibbs]

I was thinking of an IA Opamp with 5x gain. Instead of 320 and 600-700mv, I'll have 1.6 and 3-3.5V. I have some IAs from another project, but I'll have to figure out if they can accept 300mV in. I don't remember off hand which parameter I need to check on the datasheet.

hi,
For as simple Set point of 0C/32F all you need is a comparator circuit with a wide hysteresis.
So as the LM34 input drops below say 1C/33F, it switches OFF the de-humidifier and it remains OFF until the LM34 temperature reaches say 5C/41F

 

[ADWSystems]

My concern is there is only 80mV between 32F and 40F. Not much room to install hysteresis.

Only the LM34 will be subject to the temprature change. The LM34 will be mounted as a sensor, away from the main circuit. The remainder of the circuit will be kept at room temperature.

 

[ericgibbs]

My concern is there is only 80mV between 32F and 40F. Not much room to install hysteresis.

Only the LM34 will be subject to the temprature change. The LM34 will be mounted as a sensor, away from the main circuit. The remainder of the circuit will be kept at room temperature.

hi,
This is an option using a LM393 comparator and a LM34 temperature sensor.

 

[Diver300]

Apparently. Most IAs I've found that aren't rail-to-rail are (V-) +1.0V and then some.

For achieving the 5x gain, what is the best mthod? Single transistor, OpAmp, or IA? I can't remember why I can't use a simple single transistor to get my 5x gain.

Use an Op-Amp. The transistor's bias voltage will change a lot as the transistor temperature changes, so you will be measuring the transistor temperature not the sensor temperature.

(I don't know what an IA is in this context)

 

[ericgibbs]

(I don't know what an IA is in this context)

hi Diver,
IA is used as an abbreviation for Instrumentation Amplifier.

E.

 

[crutschow]

hi,
This is an option using a LM393 comparator and a LM34 temperature sensor.

It appears that both comparators need to trip at the exact same point for proper operation of the circuit. But, in practice, they will not due to differences in their offset voltage.

To resolve this you could use U2 as just an inverter by connecting the negative input of U2 to the output of U3 and adding a resistive divider to set the threshold of the positive input of U2 at about 0.3V.

Or am I misunderstanding the circuit operation?