After reading all of the posts about the accuracy of the controller's temperature indications in the MES40, I decided I should perform what I consider to be a good test of mine.
I just got this unit, it's a 2011 model, purchased just a few days ago at Sam's Club. (They do not have the newer version at Sam's yet).
I did this testing last night, so pardon the pictures all taken with flash between midnight and about 03:00.
First, I was incorrect in my recollection of the location of the sensor in my smoker. It's actually positioned quite low in the cabinet, and pretty much right above the heating elements. The overtemperature cut-off "snap-disk" control, however, is located higher, and to the left as shown here:
I sometimes see the controller temperature sensors for lab ovens located very near the "outlet" of the heater. I believe this is done to make it easier to "tune" the controllers for good stability (low or no "ringing and overshoot", but doing this comes at the expense of overall accuracy of the control system because the sensor sees the output of the heater, and thus, underestimates the temperature in the main body of the cabinet. However, this may also be seen as a benefit because it assures that no part of the cabinet will ever be driven above the setpoint.
This may be the same strategy that MB has used here. They may feel that it's more "conservative" or "safe" to set things up so that there will be no areas in the smoker that are ever much above the selected setpoint. But this does mean that other areas may be quite a bit cooler than what the controller "thinks" they are.
As I've pointed out in other posts, you will be very surprised at the variations in temperature between different locations in a typical oven or incubator, even a laboratory incubator or oven.
Regardless of all of that, the question in my mind was: Does the temperature controller in my smoker read accurately?
I had questioned the readings people were getting and posting in this thread and others because my experience with this sort of thing has taught me that to test any temperature sensor, you must be certain that the sensor under test is really at the temperature you think it is at. To know this, you must have a way to thermally couple the sensor under test to the reference sensor. Only if they're both known to be at the same temperature can you draw any valid conclusions. And this is not as easily done as one might guess. Just having a probe within a few inches of another probe, in air, can mean very little. I often measure differences of 20 degrees C or more in "good" ovens with probes that are only a few inches apart. It's particularly noticeable when you have no active stirring of the air (by a fan or blower). But you will even find variations that are annoying in incubators or ovens that ARE stirred by fans. It's a constant source of concern in a laboratory environment where tolerances are tight.
Anyhow, while variations in temperature within a smoker are, of course, a great concern, I'm not addressing that here, except to point out that just putting a probe into the smoker and comparing what it reads against what the smoker's controller reads will always be an exercise in frustration and won't tell us if the smoker's probe/controller is accurate. We haven't controlled all of the variables, so it's not a valid scientific experiment.
OK, so I wanted to make my testing BE valid.
To that end, I did the following:
First, I found a temperature probe that was very tiny and had a cable on it that was small in diameter so I could just slam it in the door of the smoker. Then I tested the calibration of this probe and its readout to make sure it was accurate.
I used a dry-block calibrator that I built a few years ago for the purpose of calibrating thermometers and electronic sensors. Basically, it consists of a large piece of machined aluminum that is surrounded by a "band heater'. The heater is controlled by a little PID controller and solid state relay. That controller reads a platinum RTD that is embedded into the aluminum block from the bottom, very near the bottoms of all of the various "wells" in the block. That whole assembly is mounted to two sheets of teflon to provide thermal insulation along with mechanical stability. Then all of that is surrounded by fiberglass insulation so the heater will not need to work hard once the block is up to temperature. The block is big and massive, and well insulated, so it has a long thermal time constant. The aluminum is a good thermal conductor so that, at least when the temperature of the system is not changing much, the temperature throughout the block should be quite uniform. It won't be perfect, but it's darn good!
The idea is that I can place electronic probes into various "wells" in the block, and know that they're all going to be at pretty much the same temperature. There is also a central well that I can fill with thermal oil for testing liquid-in-glass thermometers or odd-sized/shaped electronic probes.
I use a precise electronic calibration thermometer as the "transfer standard" to compare against the sensor(s) under test. That reference calibration thermometer system is calibrated and certified annually and the certification is NIST traceable. It's rated to be within + or - 0.015 degrees C from -50 to +200 °C.
I used this setup to test the thermocouple probe/readout system before using that thermocouple/readout system to test the smoker's probe/controller.
It was good enough. I tested it at room temperature, too, and it was dead-on. Great!
Then, at home, I attached the tiny thermocouple to the smoker's sensor:
I just held it up against the smoker's sensor and wrapped that all up with a strand of tinned copper wire to get the two in good contact.
Then I wrapped all of that up in aluminum foil. The idea was to make sure that both sensors "see" the same temperature at the same time.
It looks crude and ugly, but it ought to do the job well enough.
I had enough wire on the thermocouple to reach up out of the smoker, but just barely. :)
So I shut the smoker's door and started testing. I first checked at the ambient temperature, and was pleased to see that both systems agreed well enough:
But then, as I increased the temperature, what I found was kind of strange. The sensor/controller is quite non-linear.
I allowed the temperature to overshoot the setpoint and then settle slowly down so that I had the lowest possible rate of change of the temperature in the smoker when making the comparison readings. The lower the rate of change, the less likely issues with the thermal time-constants of the two sensors will cause errors. We know that the tiny thermocouple I attached to the outside of the smoker's probe will react more quickly than the sensor within the smoker's probe can because there must be some "insulation" between the smoker's probe's sensor and the outer body of the probe itself. So if the temperature is rising or falling rapidly, there will be a time lag between what my tiny thermocouple sees and what the sensor inside the smoker's probe will see.
Anyhow, when things were reasonably stable at any given temperature, that's when I took a picture to record the temperature readings.
Here's what I got:
Reference Thermometer Reading ----- Smoker Controller Reading
66.8 ----- 66
100.9 ----- 97
131.5 ----- 145
150.3 ----- 166
191.6 ----- 203
245 ----- 262
262 ----- 280
At room temperature, the smoker and reference agreed.
At 100 degrees, the smoker read a bit low.
Then, at all temperatures above that, the smoker read considerably higher than the reference system.
This agrees with what I think most other people have reported here.
If the controller and its sensor were meant to accurately display what the sensor is actually seeing, then the design or implementation is fairly poor. It's pretty easy to make an electronic thermometer that is quite accurate without spending much money, particularly if you already have a microcontroller at your disposal. And we know they do because the system has a remote control.
But to give MB the benefit of the doubt, it's possible that they intentionally bias the readings at different temperatures to compensate for what they believe the actual average smoker temperature will be at those temperatures with typical meat loading, etc. Who knows?
I have not taken things apart to see what kind of temperature sensor is used. It could be a thermistor, RTD, or thermocouple. Thermocouples are cheap, rugged, and stand up to a wide range of temperatures. But they require either a rather baroque analog circuit to "read" them, or a microcontroller programmed to perform the calculations using the multi-order polynomial expression that relates their voltage output to their temperature. Further, you need another temperature sensor to compensate for the thermoelectric voltage generated at the "cold junction". This is actually easily done with modern sensor ICs and a microcontroller. Analog Devices (and maybe others) make ICs specifically meant to "read" thermocouples, so using one of those can provide an easy solution. But if all of this isn't done properly, you'll get bizarre readings from a thermocouple. They're non-linear and there will be an unpredictable offset generated by the cold junction connection.
Thermistors sometimes require compensation for their non-linearities, but that can be done very successfully, too, with the right circuit or controller programming.
RTDs are excellent, and again can be read very accurately if things are done correctly.
Semiconductor temperature sensors can be great, but normally won't handle the high temperatures we'd require for a smoker controller probe.
So the real questions I have now are:
Do we really care about the inaccuracies we're seeing in the MB smoker's sensor/controller system?
Are those "errors" actually carefully programmed adjustments meant to compensate for the probe's position in the smoker, and they actually make the system more accurate? Or are these errors really just errors that we should try to "fix"?
The thing I like about the MB system is the remote control/readout. I really do like being able to check the temperatures without leaving my easy chair (or even bed) when doing a long smoking.
I could easily install a different probe and a good PID controller. You can get PID controllers off of EBAY for amazingly cheap. I haven't tried any of them, but even from Omega, I can get a very good PID controller for under $100. A solid state relay and a heatsink, some wiring, a sensor, and you've got it all done. Going the high-dollar route, you might have $150 into the mod. Shopping on EBAY for a cheap PID controller, you could probably do the job for under $50 for everything.
But you'd lose the remote control feature (not the readout). That might be perfectly acceptable. You could wire things so that the remote control could switch the whole thing on and off and still read their sensors, but you'd have to set the temperature on the new controller. That wouldn't be so bad! And many PID controllers will do "ramp and soak" which would let you program a routine of different temperatures at different times with "ramping" of the temperature to your desires.
The problem with any proportional control is that you WOULD need a separate smoke generator because with the heater properly proportioned, the temperature of the heating element would often be too low to make the chips smolder. The advantage of the factory system is that it does cycle the heater on and off, with quite a bit of hysteresis, so that when it comes on, it comes on for long enough to heat up the chip pan and make them smoke.
But I still wonder if there is a method to MB's madness with the apparently crazy non-linearity of their system. Maybe it is meant to compensate for something. I kind of doubt it, but who knows? I wonder what MB has to say about this?
Maybe the point is simply that there are other variables that will render things to be so inaccurate anyhow, that the controller/sensor errors are small in comparison.
The pork butt that I smoked in mine came out fine. And the internal temperature did what I expected it to do for the smoker temperatures I thought I was getting.
Next, I suppose I should check the probe sensor and its readings!
If you've read this far, you're either really into this, or a glutton for punishment (as well as smoked meat). Sorry for the long post.