I'm seeking an inexpensive temperature sensor. They will be buried 100+ feet from the monitoring point. +/- a degree or so; range say -20 to +25C.
Eons ago, I recall a simple current loop design using LM35's and a mA DPM. But a look at the contemporary datasheet has no mention.
What approach should I be looking into now? Suggestions welcomed.
-- A host is a host from coast to coast.................wb8foz@nrk.com & no one will talk to a host that's close.......................... Unless the host (that isn't close).........................pob 1433 is busy, hung or dead....................................20915-1433
1K thin-film RTDs are accurate. Thermistors are pretty good.
LM35 doesn't like to drive capacitive loads, like cables. And I'd be concerned that some spike would make one latch up, which they do.
An LM35 below 0C needs a negative power supply, another latchup hazard.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Over 100' of wire s gonna be the important expense. MAX31820 will do half-C accuracy, but is precise to
0.2C if you care to calibrate the individual sensors. It needs only two wires, and gives a serial number in addition to temperature (that should simplify the monitor-point wire labelling issues). They call it 'one-wire', but that'd depend on a local ground stake.
Interchangeable thermistors can be pretty accurate - +/-0.2C without calibr ation for a dollar or so.
A Kelvin connection would make sense for a 100-odd feet of cable.
Thermistors are horribly non-linear, but the Steinhart-Hart relationship is good to about a thousandths of a degree Kelvin if you've calibrated the th ermistor that carefully - a well stirred ice-bath made with distilled/condu ctivity water gives you a 0.0C point that is that accurate.
Semiconductor sensors are noisier and mostly aren't tightly calibrated.
Platinum film resistance sensors are the platinum standard, but you tend to need AC excitation(or at least reversing DC) to get much sensitivity - sig nal voltages are low - and Kelvin connection would be a must with 100+ feet of cable.
-- Bill Sloman, Sydney
David - if you can design the input circuitry, a silicon diode should do.
Hul
David Lesher wrote:
Hul Tytus should learn not to top post - it's easier to understand a respon se if it follows the question, and he should be aware that while silicon di does are cheap, the voltage drop across them at given current isn't tightly specified.
You do have to at least a single point calibration. For most silicon diodes you can run at at low enough current that there isn't much resistive drop across the diode, but if that is big enough to signify, calibrating the dio de becomes a bit more complicated.
Semiconductor temperature sensors are designed to hide this kind of problem .
-- Bill Sloman, Sydney
What about an AD590 and a little DPM.
It's current operated so the 100' should not be an issue.
I'm not sure of the status of the AD590 as I have not used them for years.
The Analog Devices web-site says that it is still in production, but that more recent designs may offer better performance. It draws enough current that self-heating may be a problem. Accuracy isn't wonderful.
-- Bill Sloman, Sydney
The long-wire, though, benefits from either digital readout or current-mode operation. At ten cents a foot, wire cost is going to swamp any transducer investment, and four-wire or three-wire cable loses to two-wire.
A thermistor or RTD can be paired with a low-V regulator (MIC5232 would be good) to do this (just regulate 1.2V into the RTD as a load, the regulator input current is inversely proportional to resistance).
The problem is, none of the easily available regulators is as accurate as the RTD, so needs another calibration. You'd need a small circuit board for RTD, regulator, capacitors.
The RTD/thermistor solutions will all take calibration and extra circuitry for two-wire cable. The MAX31820 is an all-in-one two-wire device that comes precalibrated.
If quantities were high enough, you could consider an optical fiber and use phosphorescence lifetime for temperature measurement. It takes a blinky blue LED and suitable phosphor, just look at the phase shift of the red phosphorescence with respect to the blue excitation. Jacketed plastic fiber has to be cheaper than wire if you can get past the nonrecurring engineering costs. A speck of phosphor is the least expensive sensor imaginable.
That would be a nice solution.
I found many encapsulated 1-wire devices but not found an off-the-shelf display solution.
-- A host is a host from coast to coast.................wb8foz@nrk.com & no one will talk to a host that's close.......................... Unless the host (that isn't close).........................pob 1433 is busy, hung or dead....................................20915-1433
Unless you want accuracy. The resistance of the wire itself is sensitive to the temperature at each point along the wire, so it's four wire if you need accuracy.
Regular 100R RTD's are rated at 1mA current, which is 0.1V across the sensor at 0C and 0.1mW of dissipation and self-heating. 1.2V is 12mA and 14.4mW of dissipation and 144 times more self-heating. Dubious advice.
Thermistors can have much higher resistances than platinum RTD's, and are ten times more sensitive to temperature change - temperature changes along a two wire cable introduce much smaller errors in the signal you are working with.
phosphorescence lifetime for temperature measurement. It takes a blinky blue LED and suitable phosphor, just look at the phase shift of the red phosphorescence with respect to the blue excitation. Jacketed plastic fiber has to be cheaper than wire if you can get past the nonrecurring engineering costs. A speck of phosphor is the least expensive sensor imaginable.
The speck of phosphor might not cost the people who package and test it all that much, but it won't be cheap to buy in a form that is easy to use.
-- Bill Sloman, Sydney
Oh, I don't know about that. A worn-out pink grinding wheel can be hammered into a LOT of rubiy bort, and a dab of epoxy will couple it to the fiber.
nd use
all that much, but it won't be cheap to buy in a form that is easy to use.
ered into a LOT of ruby bort, and a dab of epoxy will couple it to the fibe r.
That doesn't strike me as easy to use, or likely to give the same sensitivi ty from one fragment to the next.
And if you are measuring temperature, you need to pay attention to the glas s transition temperature of the epoxy. On one project I worked on, the off- the shelf epoxy had a glass transition temperature of about 60C which wasn' t a great idea for a part that was supposed to cope with an ambient of up t o 85C (in an industrial washing machine for beer bottles).
It wasn't hard to find one with a glass transition temperature of 125C once we'd had our noses rubbed in the fact that it mattered.
-- Bill Sloman, Sydney
Bill - you might profit with a better news reader.
The more automated usenet readers, in my case Tin, break all the messages into the various "news groups" that have been chosen and displays their names. When one is selected, a list of different "subjects" is shown excepting those previously deleted by the operator. When one of these is selected, a chronological list of the "articles" is shown by the sender's name and a mark indicating it's been read or not. Unfortunantly, many of the messages themselves show each previous message, in full text, and wading through to the end is neccessary to read the sender's message. If the author would put his message first, the hinderence of including all messages would be avoided yet still provide information for those using an email reader or a most basic usenet reader or a minimal usenet server.
Hul
snipped-for-privacy@ieee.org wrote:
I use Google groups most of the time, and Thunderbird once in a while. That isn't the problem.
The correct answer is to snip the irrelevant parts of the preceding text.
-- Bill Sloman, Sydney
So 1-Wire DS18b20 probes are readily available in multiple forms. I've been seeking the other end. I thought I'd found the ideal unit: but the manufacturer claims it works only with their probes.
I'm dubious about that, but I wonder: Does someone else make something of that ilk that does expect DS18b20 sensors. I seek to monitor 15+ sensors, and a http/telnet interface.
-- A host is a host from coast to coast.................wb8foz@nrk.com & no one will talk to a host that's close.......................... Unless the host (that isn't close).........................pob 1433 is busy, hung or dead....................................20915-1433
I've thought of something like this myself. I want to build a temperature rope that I can drop into water and monitor the temperature profile. So a sensor every foot or third of a meter or something like that would do. In my case I want the data transmitted back if local enough or stored and retr ieved periodically if more remote.
What is stopping you from rolling your own? The Dallas one wire interface is very simple. The packaging aspects perhaps?
Rick C.
There are 1-wire interface libraries for PIC micros, and there's
1-wire support available on the Raspberry Pi. At the low level it's a pretty simple protocol; using it with multiple devices on the same wire requires some software logic to handle the enumeration of them, but it's not fundamentally very difficult IIRC.
I believe all that is handled by the libraries. The low level protocol of sending and receiving a message is very trivial. The only slightly messy p art is enumerating and categorizing the various device types. If they are all the same you only need to keep track of the serial numbers which is ver y simple.
Rick c.
Thanks, but having done such before, I know how large the gap is between the lab work's bare bones and final deliverable.... and it's more than this is worth.
The good news is I've found 2 inexpensive Ethernet 1-Wire bridges for use with DS18B20 temp probes.
directly supports 22 with its simplest UA, a web page; plus more with some code.
And at nettech, Engineering eduacted their sales person; the Does work with standard DS18B20 probes.
So I can choose based on other factors.
-- A host is a host from coast to coast.................wb8foz@nrk.com & no one will talk to a host that's close.......................... Unless the host (that isn't close).........................pob 1433 is busy, hung or dead....................................20915-1433
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.