Need a reliable thermometer

[iratherfly]

If I never have to calibrate a thermometer and it's reliable enough for me not to destroy cheese than I may just go for that more expensive one...  I seriously doubt that I can calibrate a thermometer based on another thermometer based on a barometer; that's a bit too crazy-laborious. Otherwise, I don't feel that eyeballing frozen water is more accurate than just relying on an NSF/HCAAP certified thermometer, know what I mean?

It's calibration that made me stop almost completely using pH meters and now (I mean, by the time you calibrate them with two liquids and rinse, do the two liquids and the rinse again, clean the probe, stick it in your milk and wait 10 minutes for the a stable number to display - the cheese is doomed. Especially once you acidify the milk to find out it only changed 0.1pH and you can't tell if it's suppose to do this or just a bad pH calibration - unless you have a $1000 food lab meter).  I am now looking into getting a no-calibration ISFET meter.  Leaving the pH meter behind actually made me a better cheesemaker. I learned to look at the curd, feel it, smell it and judge it based on artisanal methods (such as measuring how much whey is on the curd).  Unfortunately I can't say that I am similarly capable of sticking my finger in milk to tell the accurate temp; I am not *that* good. 

[linuxboy]

ISFETs still require calibration. They're just easier to clean. And they are still subject to fouling of the junction. The key difference is that there's no glass bulb. I use an ISFET probe, and you're right, the setup would be over 1,000 new. But, I like good instrumentation as an engineer.

I agree, a good NSF thermometer shouldn't vary and shouldn't need constant calibration. For me it's no big deal to use a thermometer with known bias curve because I already did the work, so it's now a 30-second task and an additional way to confirm.

[humble_servant7]

What particular brand or model analog thermometer do you happen to have, linuxboy?

Just curious...

[linuxboy]

The alcohol lab one? It's unmarked, have no clue. I think it came from a surplus sale? Very basic, just red alcohol, annealed glass. it's surprisingly accurate, though.

[humble_servant7]

An alcohol lab thermometer is analog. And it functions based on the established properties of thermal expansion, which functions linearly based on temperature. Meaning that if you use RO/distilled water, and boil it for the high reading and put in ice water for the low, and record the barometric pressure and elevation (as you said), you can develop an exact bias curve for that specific thermometer. It will have that same bias curve from now on. You might need to make slight adjustments in the future based on the barometer, but those are minor.

It's so good because it's analog.

All thermometers need all the adjustments based on the bias curve of each one. But with calibrateable ones, the bias can be adjusted for. When you use a bi-therm (the metal dial ones), it suffers from the exact same situation as analog mercury/alcohol thermometers - the reading bias. But, unlike the lab thermometer, a bi-therm may develop different bias curves over time, especially as it is calibrated. So it needs to go through the entire effort or recalculating everything to be used as a standard, whereas with a lab thermometer, you just need to adjust for daily differences in the barometer.

With digital probes, they are actually more reliable, especially if of high quality and especially if the circuitry accounts for calibration fluctuations. But same sort of idea if used as a standard - the degree of drift may change, especially if a cord is kinked or it's dropped, or other some such. This rarely happens with good units and easy to spot by using two thermometers to compare.

Yes, there are good, reliable bi-therms, alcohol/mercury thermometers, and digital ones. Most of them are costly.

Okay, forgive me for needing the further explaining.

So... what based off of the readings we get from it beng in boiling water, the ice water, as well as the recordings for the barometric pressure and elevation, by developing a "bias curve"-- we can now use the readings from this thermometer to compare to all other thermometers and calibrate accordingly?

Is that correct?

If not then I'm not really understanding what functions the "bias curve" serves for?

and, in your last sentence you say that alcohol/mercury thermometers are mostly very costly, but i thought it was the alcohol ones tht you were telling us that cost  you only around a couple of bucks?

forgive me for the questions-- I dont mean to be a PITA-- just trying to understand.

[linuxboy]

Ok, so, for analog thermometers that are not adjustable, they are made according to a standard to read accurately. But, unless you use the thermometer in the exact conditions it was made, it will read differently. And you can't adjust the markings, they are usually right on the glass. So you have to create a chart and manually adjust for your environment in terms of air pressure and in terms of any other bias that comes into play.

That specific bias is unique for your situation. If you calculate it, that analog thermometer should be able to accurately tell you the temp because the bias curve remains the same for that thermometer in your situation. And it then serves as the authoritative source.

Many digital thermometers have a similar version of this idea. In their programs are usually weights that help to automatically recalibrate based on readings.

Notice I said most are costly. It depends on what you need. If you need waterproof, reliable, shock-resistant meters, that will cost. If you want a basic lab one, those are cheap. But the basic lab thermometer is not absolutely foolproof. If you drop it from a high height, it can break (hence, not reliable/durable). Or, the alcohol inside may separate and have air gaps in between (again, during a make, not helpful at all). You can usually fix the latter issue, but still. To have a constantly reliable meter, those usually run $50+. Just because something is NSF doesn't necessarily mean much except the company paid to get it NSF certified. All meters have their drawbacks.

I usually use two bi-therms, calibrated manually and against a lab meter. The bi-therms are durable. I sometimes will glue a bi-therm after calibration to prevent the nut from moving.

I make it more complicated than it has to be. I have a *ahem* fondness for complicated things.

[iratherfly]

Haha, you sure do Linuxboy!
So now that I understand it better...  How often do you think I should check for calibration my good, calibratable NSF analog and digital thermometer? And the probe of my pH meter? (Extech

Also, I am building a Sous Vide and Cheese vat machine.  It would be made of a digital PID controller that takes temp readings from a connected high-end food-grade thermocouple (immersed in the liquid) and switch on/off a waterproof heating element (like the one you immerse in a cup of water to boil it for tea) to match a user-requested temperature and keep is steady.
For sous vide sake, the heating element will be in the water and there will be an aquarium type of propeller pump that keep the water circulating to assure even temp throughout the vat. For cheesemaking, I will use a double boiler configuration where the heating element and propeller pump will be in the vat and the thermocouple will be immersed in the milk insert (the double boiler).  Hmmm... doesn't this sound like your kind of project?  Anyways, how often do you think I should calibrate this thing?

[linuxboy]

good, calibratable NSF analog and digital thermometer? And the probe of my pH meter?

Thermometer = every week (or if it's constant, every 1-3 months, as needed, just to be sure)
ph = every make.

if your thermocouple is in a well (no contact with food, does not get gummed up), there shouldn't be a need to calibrate beyond the initial one and an annual/period check to confirm. K/J probes and a good circuit board go a very long way. In the handheld temp meters, it's usually the flexible wire or probe that gets damaged and causes wrong readings. Or a meter falling to the floor, or other shock. Thermocouples are generally reliable if stationary and enclosed/protected.

I can help however you need in terms of design suggestions, but not much time to build one of these myself. I've done PID projects before, but not lately. I've never done power electronics/industrial EE work commercially, but I know the basics. My brother has a master's degree in this (servos, PID, industrial controls, etc), so I could ask for his input, too.

[Sailor Con Queso]

IRF - I deal with a large digitally controlled vat every day. As background, I have a 40 gallon make pot inside a 70 gallon heated water vat. Based on my observations, I have a couple of suggestions for you.

First, I do not feel that a circulation pump of any kind is necessary. I tried this in the water vat and found NO difference in the rate of heat transfer to the milk pot.

Second, the thermocouple should go in the water bath and not the milk side. FarmerJD convinced me of this. You don't need to know what the temperature of the milk is, you want to know where it is going to end up. For example, assume that your milk is 80F and you crank up the heater to warm it up to 90F. The milk gets to 90F and the PID with thermocouple in the milk cuts off the heater. Great, right? No, the water bath had to be much hotter to force the milk to reach an equilibrium of 90F. The water may be 120F and will keep heating your milk. Result? You will end up dumping the 120F water and lose control of the maintenance temperature. Then your milk drops to 86F and you're back to the balancing act. With the thermocouple in the water, you simply set it a couple of degrees higher (because of heat loss) than the desired milk temp and let it do its thing. Right? Well not really. That can take all day to stabilize depending on your volume. So what I do is preheat my water vat WAY higher than finish temp, say 150F. I pour milk into the make pot and then drop the digital controller back to the finished temp +2-3f. The 150F water makes the initial heat transfer MUCH quicker, but the water will stabilized at the finished temp.

[humble_servant7]

Let me know if that's not clear of if you need help with figuring out how to calculate the bias curve for a thermometer. There's probably some guide you can find online for it, too. I just learned the lab methods long ago, so I still follow the same ones.

this would be absolutely great if you could post this.

Way better solution than spending $100 on a Thermapen.

[FarmerJd]

So what I do is preheat my water vat WAY higher than finish temp, say 150F. I pour milk into the make pot and then drop the digital controller back to the finished temp +2-3f. The 150F water makes the initial heat transfer MUCH quicker, but the water will stabilized at the finished temp.

This is the fastest way to do it, but make sure you do a little math to make sure you don't overheat it. Sailor has found what his setup needs but your number may not be as high depending on how much water you have compared to your milk.

[humble_servant7]

So what I do is preheat my water vat WAY higher than finish temp, say 150F. I pour milk into the make pot and then drop the digital controller back to the finished temp +2-3f. The 150F water makes the initial heat transfer MUCH quicker, but the water will stabilized at the finished temp.

This is the fastest way to do it, but make sure you do a little math to make sure you don't overheat it. Sailor has found what his setup needs but your number may not be as high depending on how much water you have compared to your milk.

Do you happen to have the math, again, of calculating one's needs, Farmer?

Sorry to bother you.

[FarmerJd]

I would need a little info:
1. amount of water in the boiler/outside container
2. amount of milk
3. final temp of milk desired
4. starting temp of milk


of course the answer would be an estimate since the containers both lose heat to the air.

[tananaBrian]

For you guys that are calculating the thermodynamics, e.g. rate of heat transfer from bath to milk, are you calculating the resulting rise in milk temperature per minute as well?  So you don't blow the 2-degrees per minute rule?  Here's some other questions:

- Maybe a circulation pump in the bath doesn't help heat transfer to the milk, it should help make sure your controller's heat measurement represents the overall average of the bath's water though, right?  I mean, if the water isn't being mixed, won't it have hotter/cooler areas depending on what kind of convection is occurring and where the heating element(s) is/are?

- Would a slow circulation of the milk be a good idea?  The reason would be to make sure the pot of milk is heating evenly as well, not hotter towards the outside and cooler towards the inside.  No?  Plus, if the milk is hotter to the outside, wouldn't the heat transfer slow as well since the rate of transfer depends on the temperature differential across the boundary?

Thanks,
Brian

[FarmerJd]

For you guys that are calculating the thermodynamics, e.g. rate of heat transfer from bath to milk, are you calculating the resulting rise in milk temperature per minute as well?  So you don't blow the 2-degrees per minute rule?  Here's some other questions:

- Maybe a circulation pump in the bath doesn't help heat transfer to the milk, it should help make sure your controller's heat measurement represents the overall average of the bath's water though, right?  I mean, if the water isn't being mixed, won't it have hotter/cooler areas depending on what kind of convection is occurring and where the heating element(s) is/are?

- Would a slow circulation of the milk be a good idea?  The reason would be to make sure the pot of milk is heating evenly as well, not hotter towards the outside and cooler towards the inside.  No?  Plus, if the milk is hotter to the outside, wouldn't the heat transfer slow as well since the rate of transfer depends on the temperature differential across the boundary?

Thanks,
Brian

Good questions. You think alot like I do. These are a few thoughts not meant to be authoritative in any way .


The initial heating phase does not have to follow a slow schedule. The process Sailor described was the initial heating of the milk to 88-90 degrees before adding culture/rennet. With the "cooking" phase, you have to tinker with it a little. I keep my water temp 3-5 degrees ahead of my milk and stop this many degrees ahead of my target. The math on that would be much harder (considering all the variables) than just trial and error for 1-2 makes.


In regards to circulating the water, convection currents will do the trick without a pump, provided the heating element is close to the bottom. See this link for a demo. Another factor to keep in mind is the stirring of the milk; since it is in constant motion during the heating process (at least mine is), the constant cooling of the water along the milk pot creates another dynamic in the water tank to keep the heat moving by conduction.
On a practical note, this isn't just theory. I have measured my water tank in various places and found no discernible difference even though I have no pump. I actually bought one and wound up not using it because of this.
Hope that all makes sense.