PT1000 versus DS18B20, which one?

[Lobobo]

Hi. I am doing some home heating system controller what will be controling water solar collector, boiler and hot water cylinder. Temperatures will be checked and depending on readings the heating source (Solar or Boiler) and heating target (house or HWC) will be changed relitevely. I am confused what sensors is better to choose PT1000 or DS18B20? Guys, any advice? Maybe somebody knows any advantages or disadvantages?
Thanks.

 

[premelec]

Please be more specific about temperature range and accuracy you will require as well as distances from controller to sense unit. There are many temperature sense units that will handle mid-range temperatures. Things get difficult if you need to read very high temperatures. Accuracy from one sense unit to another for interchanging is sometimes an issue - the more accurate the more expensive the sense unit. There have been many discussions on this forum and you might do well to do a search on your topic of temperature transducers. The 18B20 is easily accommodated by PICAXEs and the PT1000 will require some special conditioning and consideration. Every transducer has some advantages or disadvantages - that's why there are so many

 

[g6ejd]

The PT1000 will need some conditioning of its value (resistance), it looks easy to mount onto various surfaces. It has no accuracy to state, as you will have to calibrate it against a known reading, the final temperature will depend on circuit parameters used to read it.

The 18B20 will give you an accurate temperature out of the box, to within a resolution of 0.0626C and accuracy (they say it's laser trimmed) should be good. Not quite so easy to mount.

Distance between sensor and PICAXE (as noted/infered by premelec) will be a determinant of your choice, for example long wires favour the PT1000, although in a good installation (shielded cables, good grounds) the 18B20 should operate well.

Have you thought of making a differential thermostat, to measure internal/external temperatures and their rates of change, such that you can predict if heating can be turned down in advance of some solar heating to the outside of the property, or turn down the heating if the temperature is climbing too fast towards the target temperature, that sort of thing.

 

[Svejk]

A PT1000 will require additional circuitry to make it work, an op amp or a dedicated IC will suffice. Advantages are good linearity and wide operating range. On the other hand, DS18B20 is used in Picaxe community, therefore you will have a lot of support here. Another sensor to consider is MCP9700. It has same operating range as the DS18.. but its price is lower.

 

[MFB]

As mentioned above, more details required for a firm recontamination. However, I have found the AD590 temperature-to-current sensor to be easy to mount and operate at remote locations. It is housed in a 3-pin metal transistor type can and often only needs one resistor to convert the current to a voltage for an adc input. The measurement range is similar to other semiconductor temperature sensors.

 

[Captain Haddock]

Search for Matherp's thread on thermocouples too (thermocouples made realy easy) as they can give a wide temperature range, you will need to use an eeprom for the lookup table but his thread gives the code for that too, it realy was very easy for a fairly good accuracy sensor, the ADC used has 4 channels and is i2c so 4 thermocouples will have excellent relative accuracy.(-90 to +999 degrees C should be wide enough for you).
Thermocouple extensions need the proper wire to be used.
As said before on here a good evacuated glass tube collector can exceed the range of an ds18b20 if the pump isn't running.

 

[Lobobo]

Thanks guys. I see the point, digital ones well calibrated, easy to program, but it's range is low limited, analog sensors are a little more complicated to programm but with wide working temperature range and needs calibrating. Anyway as I understand solar may go above 125'C, so I'll need PT1000 or other recommended ones. PT family is very popular sensor, so I'm going for this one probably as it will be easier to get help if needed.

 

[premelec]

Let us know if you find a simple conditioning circuit - I noticed there is some code in the wikipedia listing for resistance sense units. FYIW copper has a similar temperature coefficient as platinum and you can get 200DegC fine copper wire insulated and make a temperature sense unit with that. Don't forget to include the lead wire resistance in your calculations if you have very long wire runs to the sense unit... If you don't need much accuracy it should all work out well...

 

[Svejk]

I knew I had it posted somewhere. Here it is.

 

[optic]

I've just got 10 ds18b20 up and running the max difference in reading was 0.5 degrees between them.highest reading average -22.175 lowest -21.706
The best pair recorded -21.861 and -21.859 as the average temp in a freezer for 13 hours taking 23000 ish readings. where they do differ the most is their different reaction time to large temperature changes. take 10 sensors out of the freezer at -20 the you will see a lag between them upto 1.8 degrees as they get back upto room temp.