There's been a lot of talk of temperature probes recently. One thing that's been bugging me is that most folks seem to use thermocouples rather than thermistors. Is there a particular reason for that? Faster response time? In general, given equally priced meters to read the signal, you'll get a more accurate reading out of a thermistor, since the signal is easier to read, assuming good tolerances on the thermistor itself. Accurate thermocouple amplifiers are more expensive because you're amplifying a very small signal. Case in point, the thermistor-based probe I use (found via a CoffeeGeek forum post):http://indoorhealthproducts.com/912.htm$20 (free shipping, even) buys you +/-8F absolute accuracy in the critical coffee-roasting temperature range, with 1F resolution (which is actually sometimes 2F -- the meter skips certain steps, presumably those which fall in between steps on the analog-to-digital convertor). To get +/-8F out of a thermocouple amplifier is a bit more expensive. Your typical digital multimeter with thermocouple input will not do that well, even if it's a high end $120 meter. Granted, dedicated thermocouple meters in that price range can do as well or better, but not for $20. So is it the response time that makes people prefer thermocouples? Seems that would mostly be a function of probe size and material, forget whether it's a thermistor or thermocouple inside the probe. I design circuit boards as part of my job (despite the complete lack of formal training for that sort of thing), and I've been planning to build myself a computer interface for profile monitoring and control. I'm just wondering if I should build it for thermocouples or thermistors.
Randall, it is easy to snake a TC into the bean mass, and they also come as handy rigid probes, which are easy to mount. Another reason they are often used is that they are connected to a PID controller. Dan
Nevermind, I just went shopping for thermistors a bit -- you can get accurate thermistors, and you can get thermistors that work in the temperature ranges required for roasting, but one that is accurate and can also take the heat is hard to find and probably pretty expensive. So I guess that answers it. On Sat, Jun 30, 2007 at 02:36:14PM -0400, Randall Nortman wrote: <Snip>
And that is basically what I was going to say. I would find +/- 8F accuracy was to inaccurate for my taste. I have tested the one Tom carries ( I compared against a couple NIST calibrated thermometers) and they are accurate from 32 to 600 F within 1 F at the lower ranger (under 212 F) and within 2 F at roasting range. Worth the extra $10 in my book. At 12:27 6/30/2007, you wrote: <Snip> John Nanci AlChemist at large Zen Roasting , Blending & Espresso pulling by Gestalthttp://www.chocolatealchemy.com/
There is this to say about that: If you're after absolute precision, SCRAP the thermistors,and ditto the RTD sensors. Think it through- Thermistors and RTD's have temperature sensing elements that change resistance with changing temperatures. They also have leadwires that do the same thing, so spin the wheel to determine exactly where you're reading minute temperature variations. The readout device has to have a very stable reference current source, that feeds the thermistor or RTD through the leadwires. The sensor could be a NTC device, and the leadwires PTC, or you could have a three or four wire system. What a debacle, and it's still based on the assumption that leadwire errors cancel. Reasoning like that brought down the Challenger. Good in theory; won't hold water. Thermocouples develop a temperature-dependent potential at one point- the junction. You could read a thermocouple from any distance, since it's a very low impedance source, Direct Short, and all. I hope you aren't greatly worried about building a high gain instrument amplifier to read the thermocouple and drive the readout device. Can you say "Operational Amplifier?" The 741comes to mind, and you couldn't hold $5 worth in both hands. High gain, high input impedance, precision- 8th grade science fair project. Cheers -RayO, aka Opa! Why reinvent the wheel?
I suspect the one I have is much better than +/-8F, but that's the spec. I think I looked up the specs on the one Tom carries, and it certainly wasn't +/-2F. I could be mistaken, but the point is that you usually get somewhat better than the specs, which give you the worst case. (Or at least they're supposed to.) Anyway, it seems to be very consistent from roast to roast as long as I don't move it, which is more important than absolute accuracy. When I get around to building my computer interface, I will probably go with a thermocouple, as appropriate probes are readily available. It will be a little extra expense on the signal amplifier, but probably worth it. On Sat, Jun 30, 2007 at 02:00:44PM -0700, Alchemist John wrote: <Snip>
On Sat, Jun 30, 2007 at 03:30:28PM -0600, raymanowen wrote: <Snip> Thermocouples have a similar problem, in that there are at least two other junctions of dissimilar metals in the system -- where the thermocouple wires connect to something else. So you need a compensated connector, and ideally you need to know what temperature the connector is at and all that. And don't forget that it will pick up noise along the transmission lines, and if you add a little bit of noise to a millivolt signal you have problems. <Snip> Nah, you don't need a current source -- a stable voltage and a resistor bridge (i.e., Wheatstone bridge) will do the trick. If you can get by with less resolution, you don't even need a full brige -- just put the thermistor in a voltage divider circuit with one other resistor. That assumes you have a digital processor at the other end that can do all the calculations necessary to turn that into a temperature reading. But anyway, I have convinced myself to go with a thermocouple simply because of the lack of readily available high-temperature thermistor probes that are also reasonably accurate (without calibration). <Snip> The classic 741 has too much offset error to amplify a millivolt signal reliably, without calibration. There are opamps that will do the job, sure, though the good ones can run around $5 each in small quantity, or even more for the really good ones. You have to get one that has low offset and good linearity even at really high gain -- if you want a really good, accurate signal. Better is to buy an integrated instrumentation amplifier -- decent ones again can be had around $5 in small quantity, but running up to $20 for the really high-end ones. I could probably get by with a $5 model. <Snip> Because existing wheels all have a critical design flaw: Somebody else built them, and they're going to charge me if I want one. How much for a good, accurate, dual thermocouple amplifier/digitizer that outputs to a PC in real time? How about with a couple of 15A SSRs, also controlled by the PC? What if I want to download my own control program to the thing so I can untether it from the PC? OK, but now let's stick a wireless interface on there so I can have it talk to the PC without the obnoxious tether? Sure, I can buy something like that, but not as cheaply as I can build it myself. And that's building just one. If I can find half a dozen other coffee freaks who want PC control of their roaster and/or espresso machine, I can build seven of them for barely more than what it costs to build one, and everybody wins. Not that I'm trying to build a cottage industry here. I just want it for myself. But you can bet that when I get around to building it, I'm going to try to find others here or on coffeegeek.com who want the same thing, so I can bump up the volume and lower the per-unit cost. For now it's a pipe dream. Actual work has me far too busy to have time for hobbies. Maybe in a few months...
Randall Nortman wrote: <Snip> Hum, well, don't know about most, but any probe would get in the way of wok roasting. :) So, I don't use either. Sight, smell, sound guy here. Seems to work every time. That El Salvadorian I ordered, Tom, is really good. Been sipping on it all day. -- Rick Copple
The Hearthware Precision air roaster uses a thermistor to regulate the heat output into the roast chamber. I wonder what spec they use for that one? The air temp is supposed to be 500F entering the roast chamber according to a Hearthware technician several years back. ********************* Ed Needham "to absurdity and beyond!"http://www.homeroaster.com*********************
Randall, You might be interested in this if you're thinking of setting up you're own amp to read TC voltages. Dallas/Maxim makes a DS2760 chip that was designed to monitor laptop batteries. It measures voltages to +/- 15 microvolts (from memory). The voltage data can be read via a 1-wire interface. I'm using this module to monitor the bean mass temperature using a K-type TC. Personally I use TCs just because I have used them for years at work, -Pete ------------------------------------------------------- There's been a lot of talk of temperature probes recently. One thing that's been bugging me is that most folks seem to use thermocouples rather than thermistors. Is there a particular reason for that? Faster response time? In general, given equally priced meters to read the signal, you'll get a more accurate reading out of a thermistor, since the signal is easier to read, assuming good tolerances on the thermistor itself. Accurate thermocouple amplifiers are more expensive because you're amplifying a very small signal. I design circuit boards as part of my job (despite the complete lack of formal training for that sort of thing), and I've been planning to build myself a computer interface for profile monitoring and control. I'm just wondering if I should build it for thermocouples or thermistors. Finding fabulous fares is fun. Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains.http://farechase.yahoo.com/promo-generic-14795097
I tossed mine so long ago; my memory says it looked like it could have been a diode? John
"...it could have been a diode?" Some reverse biased junctions are extremely temperature sensitive. Diode? Yes!! -ro
On 6/30/07, raymanowen wrote (among other things): "....Thermistors and RTD's have temperature sensing elements that change resistance with changing temperatures. They also have leadwires that do the same thing," 3-lead RTD's and a self-calibrating monitor (i.e. a GE Multilin) use the third lead and the common lead to keep track of lead resistance and compensate for same. I think. -- Larry J Whatever you do will be insignificant, but it is very important that you do it. - Mahatma Gandhi