Sous vide controller accuracy

[Robert Jueneman]

I just accuracy tested the newest Sous Vide Magic ID controller, the 1500C from Fresh Meals Solutions, using my $500 Traceable 4000 data logging thermometer from Control Company, which is calibrated to NIST standards to within 0.1F.

The 1500C uses a new probe which is advertised as being more accurate at higher temperatures, and that proved to be the case. Out of the box, with no other offset calibration, the 1500C was accurate within the 1 degree F roundoff error throughout the sous vide cooking range of 100F (very rare salmon), 131F (a nice medium rare for beef and lamb, at least to my taste), 148F for the perfect egg and perhaps for pork (although 155 might be better, at least for most people who don't care for pink pork), to 185F (asparagus and other vegetables).

(The first three temperature settings are actually quite demanding as far as sous vide accuracy is concerned -- certainly 2 degrees F will make a readily observable difference. It might not spoil a dish, but it is the difference between perfection and being merely acceptable. I haven't cooked enough vegetables sous vide to know how important absolute accuracy is that the upper ranges, but I suspect it is equally important, although perhaps less observable, at least directly.)

The slightly more expensive 1500B (Celsius only) controller offers 0.1C resolution and was exactly right at 55C, 0.3C low at 38C and 0.4C high at 65C, but was off by 1.1C at 85C.

I plan to use the 1500B for cooking meat sous vide, the 1500C for vegetables and other foods requiring higher temperatures, and the somewhat less accurate (in an absolute sense, at least without calibrating it to the specific temperature range to be used) but still quite stable 1500A whenever I need a third unit for some reason.

If anyone has tested the accuracy of the PolyScience or Grant immersion circulators with a NIST-certified accurate thermometer, I'd be very interested in the results. They provide specifications regarding the tightness of their controls, but they don't cite their accuracy or repeatability.

And without knowing the accuracy of the measurements, I don't know how to interpret someone else recipe recommendations. Sous vide is all about precision, but unfortunately, out of 10 or more thermometers, most are rather inaccurate -- at least if used across a wide range of temperatures.

What kind of thermometers are professional chefs using to calibrate their equipment, and how often is that being done?

[slkinsey]

Robert, I'm not sure what exactly you tested here.

You were somehow able to test the accuracy of the temperature probe directly?

Or did you hook up the various controllers to some kind of water bath and test whether the temperature the controller gave for the the water bath matched the temperature your fancy logging thermometer gave for the water bath? If so, what setup did you use? And did you place your fancy probe exactly alongside the controller probe when taking the temperature readings?

[dougal]

...

What kind of thermometers are professional chefs using to calibrate their equipment, and how often is that being done?

←

Robert, I'm not even sure that many "professional chefs" would even have heard of the concept of calibration, let alone against traceable standards.

[Robert Jueneman]

slkinsey,

You are asking the right kinds of questions!

No, there is no way (that I know of) to test the accuracy of the probe itself. I used the probe that came with the individual SVM controllers, and read out the reading form the display.

I used the Sous Vide Magic PID controllers to bring a 10 liter rice cooker full of water to the desired temperature, then waited until everything was stabilized.

I used a submersible garden fountain pump within the rice cooker to make sure everything was circulating nicely, and there were no dead spots. Since the Traceable 4000 has a read-out resolution of 0.001F, it was easy to tell that everything was quite uniform once everything stabilized -- within 0.01F at least, throughout the cooker.

I first brought the temperature up to 131F/55C, and then adjusted the offset of the probe/SVM controller combination in accordance with the instructions. This was to establish a basic accuracy point, so that I could then measure the linearity (deviation from the Traceable 4000's reading) across the temperature range.

The reason for selecting 131F was that it is the lowest you should go and still assume meat is being adequately pasteurized. A degree or two error at that temperature could potentially end up making people sick.

However, now that I think about it , establishing a basic accuracy point of 100F might be a better practice, because 100F is within the range of the most accurate yet inexpensive thermometer most people are likely to have -- a fever thermometer, or better yet, a basal or ovulation thermometer, which should be accurate to within 0.1F.

The only other calibration point that might be reasonable available for people would be the triple-point of water (32F/0C), as measured in a slurry of distilled water and ice made from distilled water. When using this technique, it is important that you keep pressing down on the ice to eliminate any water-only at the bottom of the thermos or other vessel, but at the same time, the mix must not be just ice.

Unfortunately, the SVM controllers don't seem to like such cold temperatures, so you will have to calibrate another thermometer, and then calibrate the SVM at say 40F.

One of the best, affordable thermometers I have found is the All-Clad T201. It reads to 0.1F, and has a Self-Set calibration button for use with an ice bath. if you calibrate that unit at 32F, and also at 100F and find the 100F setting to be reasonably accurate, you are probably OK within the meat-cooking range. However, even the All-Clad was off by 2.2F at 150F, and 1.5F at 185F.

Dougal, although I haven't read any of the HAACP standards, I certainly HOPE that they call for periodic calibration of thermometers and/or circulators using NIST-certified reference thermometers. In my experience testing over 10 digital thermometers, I wouldn't trust ANY of them without calibration, as I have seen as much as 14F errors!

BTW, the cost of the Traceable 4000 without the data logging software and cable is only $305. I realize that seems like a lot, but not when you compare it to a PolyScience circulator or a chamber vacuum!

Please, everybody, if you are going to cook sous vide, and post recipes as to what you did, at least spend the $17.98 to buy a basal thermometer, and perhaps a couple of bucks for some distilled water for an ice bath.

[Mallet]

I use a Lauda analog circulator for SV, and always check the final temp with a Thermapen. Unfortunately the thermapen only reads to the nearest 1F, but they always match up within that range.

[Robert Jueneman]

I recently had a problem with the All-Clad unit. it seemed reasonably accurate above room temperature, but when I tried to re-calibrate it in an ice-bath, it wouldn't read below 60F!

I exchanged it at Williams-Sonoma, and the new unit seems to work reasonably well.

My unconfirmed suspicion is that that you shouldn't submerge the entire probe in water, but only the metal-clad tip. The instructions are far from clear -- it says not to put the entire unit in the oven(!), or in the dishwasher, but it does not say whether it is safe to immerse the entire probe. In any case, henceforth I am not going to.

In the past, I have tested a number of thermometers at 100F, using a basal thermometer. I still think that is a good check point, but I was moderately encouraged by the accuracy that I got with five different thermometers at 130F or 55C, when compared to my Traceable 4000. All were within 1 degree F, and most were within 0.6F, after having weeded out the ones that were grossly wrong.

The most recent All-Clad T201 was off by -0.6F at 131F, whereas the cheaper Sur la Table unit was off by -1.3F at 32F, but only -0.2F at 131F.

All in all, if you are serious about sous vide, or if you are running a business where people's health might be at risk, spending the money for a NIST-traceable reference-grade thermometer to periodically calibrate your water baths and other thermometers would be a wise investment.

[tomdarch]

I'd add two points to this. My soon-to-be wife is a paralegal and has worked on food-borne pathogen cases, so I've heard some grizzly details, and frustratingly nonexistent paper trails on critical items like fish for sushi. It's tough to defend your restaurateur-client when there's no evidence or records to point to.

1. Get the reference thermometer, and use it, but don't forget to log the results. For science-types, logging calibration data is just assumed, but it's pretty uncommon outside of labs or ISO 9000-type operations. Maybe just keep a small notebook near where you have the ICs, and note the date and the result for each setup. Someone here can suggest how often it should be done - weekly? monthly? If you do get dragged into some bogus suit ("They didn't cook my food, they just held it in the Danger Zone!"), you will be able to demonstrate that when you say you cooked at 134F, it really was 134F, and not 128F. Plus, it's a good way to demonstrate that you know what you are doing and take more than standard care regarding food safety in this "avant garde" field.

2. Don't assume that the circulation function is really keeping the bath at a consistent temp throughout. I've seen some photos of pretty crowded looking tubs in restaurant kitchens. Just (carefully) poke around in different parts of the bath with the thermometer to confirm that you don't have any "dead" spots. If you do, rearrange, use racks, whatever, and keep checking periodically. Yes, you are a professional, no you haven't made anyone sick, but don't get complacent, particularly if you are doing lower temp, shorter cook items.

That's another point - if you are using an internal probe thermometer to "cook until the internal temperature is X", then you should use the calibrated thermometer to confirm that your everyday thermometers are correct in the temperature range that you are using them. (and, yes, you should log those checks and repeat the checks periodically. wheeee! Isn't this exactly why you got into "the culinary arts"?) That's probably true for non-sous vide kitchens, too.

(On that dead spot point - I was pleasantly surprised that my home setup was pretty consistent (+/- 2C around 60C) even without a bubbler/pump. (I'm using a SVM to control a large coffee "urn".) I was using an untrustworthy thermometer, but I'm guessing that it's good for showing relative changes, even if it is inaccurate in terms of absolute temp. I was worried because I had two good sized bags in the tank, so I used another thermometer to check for "dead" spots. They were there, but only by a degree or two. I was doing a long pork belly cook, so I wasn't worried. Even at 58C, they were well pasteurized, given the cook time. Nonetheless, I'm still going to get a bubbler, and keep checking.)

[Robert Jueneman]

Tomdarch, your points are very well taken. I'm neither a chef nor an attorney (although I sometimes play one or the other on the Internet), but I hang around with enough attorneys to vociferously agree with you.

In our litigious society, if someone gets sick, they are going to want to sue someone for making them miserable, and if their last or at least most memorable recent meal was cooked sous vide, that will be an obvious target.

The plaintiff's attorney will undoubtedly have a field day with sous vide in any case, as this will be seen as new, experimental, and "risky" to a carefully selected jury that prefers well-done beef. They will bring in "expert witnesses" who will recite the USDA guidelines, even though their witnesses probably have no training in either microbiology or food science.

Having an approved HAACP plan will go a ways towards diffusing that issue, but you will have to be prepared to PROVE that you actually follow the procedure.

At a minimum, I would recommend at least a weekly check of your working thermometers and any circulator baths against a reference thermometer -- not because I think that thermometers will vary all that much in such a short period of time, but just to show due diligence at an exacting level of perfection and precision.

All such records should be prepared in ink, recorded in a permanently bound record book, signed by the person taking the measurement, and perhaps witnessed if you want to be ultra careful. A signed printout from an automatic data logger would be even better.

Periodically, at least once an year and preferably twice a year, the reference thermometers themselves should be sent back for calibration. The calibration lab will record what the accuracy of the device was as received, and they will then adjust the device (if possible) to make it even more accurate.

For the casual home sous vide enthusiast, $300 for an occasional-use reference thermometer may seem like an outlandish extravagance, and perhaps it is. After all, if your spouse or kids gets sick, they aren't likely to sue you. And if you get sick, you certainly aren't going to sue yourself, right?

But if you would like to be sure that you know what you are doing, they you aren't endangering your loved ones with a low-cost thermometer that may or may not be accurate, and that this cutting edge stuff is in fact safe, then this may be money that is extremely well spent.

Edited April 23, 2009 by Robert Jueneman (log)

[dougal]

... although I haven't read any of the HAACP standards, I certainly HOPE that they call for periodic calibration of thermometers and/or circulators using NIST-certified reference thermometers.  ...

←

But you have to admit that, over the last ten days or so, the rather 'limited' response from pro kitchens boasting about their use of traceable calibration standards would seem to indicate some basis for my earlier contention!

Perhaps there is (or should be) an opening for a new service industry?

However, for home kitchens, I don't think that many question the accuracy of ANY measuring instrument - not even their fridge, freezer or oven thermostats. So, I wouldn't expect to see much of an uptick in reference thermometer sales, based on domestic buyers.

[nickrey]

... although I haven't read any of the HAACP standards, I certainly HOPE that they call for periodic calibration of thermometers and/or circulators using NIST-certified reference thermometers.  ...

←

But you have to admit that, over the last ten days or so, the rather 'limited' response from pro kitchens boasting about their use of traceable calibration standards would seem to indicate some basis for my earlier contention!

Perhaps there is (or should be) an opening for a new service industry?

However, for home kitchens, I don't think that many question the accuracy of ANY measuring instrument - not even their fridge, freezer or oven thermostats. So, I wouldn't expect to see much of an uptick in reference thermometer sales, based on domestic buyers.

←

Mind you, home cooks who do sous vide currently tend to be slightly different (in a good way of course).

I calibrate my sous vide using my thermapen and am sure most of us discussing this topic in eGullet do something similar, particularly given the number who are scientists of some type of another.

[Robert Jueneman]

Quis custodiet custodiens? Who shall watch the watchers?

How do you know your Thermapen is accurate, short of sending it in to a professional calibration facility?

One relatively simple way is to check it at the triple-point of water (ice bath) (32F/0C) using distilled water for the water and the ice.

Then, to check linearity, use a $17 basal thermometer at 100 degrees. It should be accurate to 0.1F.

But what it does at 131F, or 185F, much less at 400F, may be a completely different issue.

IMHO, the Thermapen is good enough for home sous vide use, but I'd like to see something better for professional use.

[dougal]

...

How do you know your Thermapen is accurate, short of sending it in to a professional calibration facility?

...

IMHO, the Thermapen is good enough for home sous vide use, but I'd like to see something better for professional use.

←

Those that are concerned about such things might be interested in the "high accuracy" variant of the high speed Thermapen - which comes with a calibration certificate (against traceable standards).

Its also on 'April special' at $74 ...

http://www.thermoworks.com/products/therma...hite_hiacc.html

Note the spec BTW - reading to a precision of 0.1°, it is said to be accurate to ±0.8°F (or ±0.4°C for that version) over its entire range. Recognising that precision and accuracy are not the same thing is an important foundation of understanding.

Incidentally, I learn that calibration adjustment of the regular thermapens can be performed reasonably straightforwardly - the principal difficulty being the accuracy of the temperature standards...

http://www.kamado.com%2Fdiscus%2Fmessages%...ation-14348.doc

While mention is made of altitude, no mention is made of weather!

The reason boiling point varies with altitude is because of the drop in air pressure with increasing height above sea level. But air pressure varies constantly even in the exact same place - and therefore, so too does the boiling point of water.

What needs to be done is to correct for your actual current atmospheric pressure - not merely your altitude!

This should be useful http://www.csgnetwork.com/prescorh2oboilcalc.html