Temperature Control in Sous Vide Cooking

[PedroG]

I am fascinated by the sous vide process, but one thing that makes me hesitate, is that I have heard you have to keep the temperature accurate within a 1/10 of a degree for safe and positive results. Without buying a mega-dollar lab cooker, how is this accomplished in the home, or is it even necessary to be that accurate?

Bil

I wanted to put this in with the other sous vide section, but it ended here. don't know how to move it. First time post (Newbie  )l

←

To begin with, for short-time-cooking (less than 4 hours) temperature stability and accuracy are not as crucial, ±0.5°C or even ±1°C may do and can be achieved with the oven-waterpot-method, see http://sousvide.wikia.com/wiki/Give_Sous-v...nsive_equipment .

For all necessary information on sous vide see Douglas Baldwin's Practical Guide http://amath.colorado.edu/~baldwind/sous-vide.html .

When proceeding to long-time-cooking, temperature accuracy (thermometer calibration!) and stability (PID-Controller) will matter, as you want to navigate between the Scylla of safety (min. 54.4°C to prevent toxin production by clostridia) and the Charybdis of overdoneness (in my experience a piece of beef may lose its juicyness and pink color when cooked for substantially more than the minimum time according to Douglas Baldwin’s tables). The sum of thermometer accuracy and water bath stability probably should not exceed ±0.2°C for long-time-cooking.

I hope this will answer your initial question.

Pedro

[dougal]

I have bought this (Combo 4):

http://freshmealssolutions.com/index.php?p...emart&Itemid=26

works great, a friend of mine that is a chef has bought one for his home as well, results are as good as in his restaurant. ...

←

It looks very nice and I would love to have it, however, by the time I got it up to Canada, with shipping and taxes and exchange, ...

←

Ummmm.

Fresh Meal Solutions are based in Ontario, Canada, aren't they?

http://freshmealssolutions.com/index.php?o..._id=1&Itemid=34

[dougal]

Regarding the improvement of temperature control, circulation is important.

Obviously, its important to the uniformity of temperature in the bath.

But beyond that, circulation can facilitate control.

Consider a setup where the water is pumped past a heater, then, a little way along the tube, past the temperature sensor. The controller just has to keep the sensor temperature within a fairly narrow range - and because of the short delay between adjusting the heater and seeing the result at the sensor, the control problem is pretty damn simple.

Sure, the bath temperature (and hence the inlet temperature) will vary as food is added to the bath, but it still stays simple - adjust the heating to keep the circulator exit temperature damn close to the target, and the bath will steadily come back to temperature.

I'm not even convinced that some of the commercial circulators actually have a PID controller.

An accurate temperature measurement (in the outflow), and adequately fine control of the heater power, could potentially obviate the need for the complexity of a PID. And it might explain the reticence of vendors to publicly describe the actual controller!

Its massively harder for a (PID) controller to deal with the significant (and potentially variable) time delays in a system that relies exclusively on natural convection.

Interestingly Fresh Meals Solutions sells a cheap accessory to mount the sensor quite close to the heater -- it seems to be a steamer plate! That should certainly help.

http://freshmealssolutions.com/index.php?p...emart&Itemid=31

Putting pump, heater and sensor in a fairly close-coupled arrangement has to be the way to improve control -- and specifically to reduce overshoot (leading to 'overcooking').

The question remains as to how that can be best achieved, for amateur/domestic use, effectively, easily and economically.

Using an aquarium bubbler is one simple possibility. Especially if the bubbler stone was used in conjunction with the steamer plate sensor locator, and the bubbler was located just above the sensor, to encourage circulation over the sensor.

But what of pumps?

Does anyone know of anything readily available, that is smaller and cheaper (and less potent) than a central heating circulation pump, but which is nevertheless designed to be reasonably durable at temperatures above 50C/125F?

Edited August 6, 2009 by dougal (log)

[Robert Jueneman]

OK, I've got to jump in here.

First of all, with regard to the absolute accuracy of the controller, this is of primary importance when you get close to the point of not quite killing all of the bacteria when you are cooking something for a long time, like 48 hours. If you cook brisket for 48 hours, as I do at 131.5F, the minimum FDA recommended temperature, you want to be pretty sure you are dead on. (You might be able to get by at 127.5F, depending on who you read and who you trust, but at that point you'd better hope that both your thermometer and the original researcher's setup were very accurately calibrated!)

NSF requires 2 degree F accuracy. That might be fine if you are cooking hamburgers to 165F, but I don't think it is good enough for sous vide. I've gone through over a dozen thermometers of various types in order to find three or four accurate ones, and my standard for acceptance is +/- 0.5F. And if I buy one that advertises 0.1F but is off by over a degree, and I have, I make a hell of a stink about it, up to the President of the company that manufactured it, and the buyer of the retail chain that sold it.

Short of buying a calibrated NIST certified thermometer that is guaranteed to be accurate to less than 0.1F across the entire scale (as I have done -- the Traceable 4000), your best bet is to calibrate your working thermometer with a basal (ovulation) thermometer, which should be accurate to 0.1F or less at 105F.

If you find that your controller isn't quite accurate, you can either adjust the offset, or simply make a mental adjustment.

The next question deals with the precision of the controller. In my experience with the Sous Vide Magic controller from Fresh Meals Solutions, the controller is perfectly capable of holding the temperature of a water bath such as a rice cooker, or a CrockPot, etc., to within less than 0.5F, once it has reached steady state. (And by the way, I own a 1500A, a 1500B, a 1500C, and three rice pots of varying sizes, and I custom calibrated another dozen 1500Cs for a class, thinking that I might sell some. I still have some of those units left, so if anyone is interested, PM me.)

Now, I'll grant that when you first start up the system, you may experience some overshoot,and conceivably even some long term oscillation. Generally, those problems can be alleviated by fussing with the PID parameters, but that would be an entire thread all by itself. But you can avoid all of that hassle by simply turning on the rice cooker, etc., a hour earlier, and letting it stabilize. And using a larger container of water, such as my 10 liter rice cooker, also helps

Generally speaking, you would like to avoid any overshoot of more than 2 degrees F or 1 degree C for more an a few minutes. However, if the water overshoots a bit while the meat is still warming up, this does no harm, and may actually help.

Now, as to circulation. One of the advantages of using a rice cooker is that the units heat from the bottom (unlike CrockPots, which heat from the sides), and the natural convection is such that very little temperature gradations exist -- in fact, I've never seen any that would exceed a few tenths of a degree F, at least while the rice cooker is empty.

Now, depending on how big your pot or cooker is, how much frozen or cold food you dump in all at one time, etc., you might get some variations. In all probability these will be local cold spots rather than hot spots, and if you are cooking something for several hours, much less several days, everything will be evened out rather quickly.

However, if I am cooking multiple pieces of meat, or calibrating thermometers, I do tend to use a circulator.

For the lower temperatures, I use the cheapest submersible garden or "fountain" pump I can find -- about $20 at Home Depot, in the garden or outside section. They work great, and do the job well -- you can angle them in almost any direction, and they have little suction feet that will stick to the bottom or the side of a rice cooker. However, although I have used them very successfully at 165F, I can testify to the fact that at 190F, they melt and deform, and stop working. (If someone would like to know more accurately when they stop working, feel free to send me $20!)

For higher temperatures, such as for cooking vegetables at 185 or so, I use a small external aquarium pump with a check valve, together with a circular air stone. I use that rather infrequently, because it is a little noisy, because it tends to cool the water bath too much, and because I seldom cook vegetables sous vide. In any case, I very much doubt that precise temperature control is required for a small bag of asparagus or some apples, even in a small 1.5 liter rice cooker.

Hope this helps.

[jk1002]

I was sort if inspired by the video that Grant Achatz and his sidekick Nick did. They made a sous vide turkey at home without vacuum pump or temp control.

I followed the instructions in Thomas Keller Book. He used 59.5 celsius for about 45 minutes on a 400 gr piece. I had a little less, 320 gr piece. I spend some time figuring out that the induction burner would keep the water at 60 celsius when set to low. So I heated the water on high to 64, switched back to low, dropped the meat bag in and cooled it with ice to 60 stirring every now and then.

I don't have a vacuum pump so i used a normal ziploc bag. I put the steak in, dropped it into water to push the air out and then aeh, I sucked the remaining air out with my mouth - I cooked for myself so that was OK. I am getting the ziploc bags that attach to a manual pump though .....

I finished it off with a pan sear on all sides to kill off the bacteria. Tasted awesome, texture was really nice even though the piece of meat was a 5$ choice Sirloin filet from Wholefoods.

I am a bit of a steak snob - I haven't eaten a choice sirloin in ages so I can't really compare but I was really happy with the results.

You do need a digital thermometer though.

JK

Result:

[Brasshopper]

I have read the topic through, and I am trying to figure out what the answer to the original question is, how closely does a sous vide controller need to control the bath temperature? I think I understand the answer. I will summarize, and anyone who disagrees should, we will all learn something. Let me see if I understand:

NSF requires that a sous vide setup holds temperature within 2F. Since NSF requires this, we won't even discuss tolerances less tight than this.

The most critical thing that you might cook is a boiled egg - because of the fact that when you cook an egg, a series of proteins denature at fairly close temperatures. Thus, in order to get repeatable results, you want to have that .5C control (about 1F, loosely).

Control within two degrees is most important at the lowest end of the scale, where 2F from 131 puts you at 129, and that is on the very edge of the danger zone. This is also probably why most temperatures call for 131F rather than 130F. If the sous vide cooking range is defined as from room temperature, for tempering chocolate, up to boiling, (since water not under pressure won't go any higher) we can say that, in general, control at the low end is more important than control at the higher end. Vegetables other than starches cook at about 85C/185F and that is not such a critical temperature - the higher end temperatures have, in general, more tolerance than the lower end temperatures. Even for things like cooking meat, there is more difference in reactions at the lower end of the scale over, say, 2 degrees, than there is at the high end. Meat cooked to 133 may be noticeably different than meat cooked to 135, but a vegetable cooked to 185 is probably not distinguishable from one cooked to 187. There has been discussion, elsewhere, of using a PID to control the temperature of a fryer. (I have a metal K thermostat that will probably handle frying temps and a fryer. I should try it.)

Many home cookers, specifically crock pots and tabletop roasters, heat from the side. Those can have stratification in temperature that is greater than the error allowed (I've personally measured 4F top to bottom in a 12 quart roaster). Those cookers need to be stirred. Cheap fountain pumps work up to 160F, then they melt. Above that, an aquarium air pump with silicone tubing (It can be had in food grade) can be used to stir (I plug the end by folding it and stuffing it into a stainless steel nut and clipping holes into the side of the tube so that the bubbles are spread out. The nut also weights the end.) There are two problems with air pumps, one is noise, and the other is that the air flow can cool the bath. I have used an air pump in a roaster for 72 hours and gotten no more than a half inch of evaporation, if that, using silicone baking mats as lids.

(I have bought an immersible pump that is rated for 105C, but not rated at that temperature when immersed. They don't give you the immersed temperature. The pump does not self prime as far as I can tell, but filled with water, it will continue pumping. It pumps 1L/Min, not nearly as much as other circulators. It works on 6 volts DC, and I found a cheap adjustable brick that produces 5.9 volts when it is set to the 3v setting. I use it to stir my rice cooker.)

You are best putting your thermostat nearer the heat source, not away from it, it will make your end result more accurate.

A small initial overshoot is not a problem and might be a good thing.

So, what I get from all this is that tolerance tighter than 0.5cish is meaningless other than for bragging. And if you are not cooking eggs, tolerance tighter than 1C (1.8F, 2Fish) is for bragging rights.

Is that basically it, the summary answer? 2F for everything but eggs, 1F for eggs?

[PedroG]

Temperature tolerance tighter than ±0.5°C may make no difference in organoleptic outcome. But I strongly disagree allowing a ±1°C tolerance for LTLT cooking at 55°C and for pasteurizing. 54.5°C instead of 55.0°C increases pasteurizing time from 89min to 110min, see my article "Importance of temperature control on pasteurizing times" in wikiGullet.

Forced circulation is necessary for even temperature distribution in whatever cooker (bottom heater or side heater), as natural circulation at steady state (with power as low as 30-50W) is nearly nul, see my post in the SV topic on 22 June 2011.

I agree that claiming ±0.1°C stability and accuracy or better is bragging, or fun for geeks.

So my summary is: ±0.5°C or better (±0.25°C desirable) for eggs, for pasteurizing, and for LTLT cooking at 55°C, ±1°C for everything else.

[nickrey]

I agree that, apart from eggs, the main reason for having high levels of accuracy is if you are playing at the edge of safe cooking temperatures and if you are working with minimum times for achieving outcomes such as pasteurisation. If you cook over 55C and add a safety margin onto the pasteurisation times, a reasonable level of accuracy (+/- 1C) is more than sufficient. Also, if your error is higher than the target temperature (overshoots with no undershoots), you can even get away with larger variations.

[Brasshopper]

So, how about this: "The person cooking sous vide must be aware of their equipment and the tolerance that their equipment has relative to the actual temperature. When calculating pasteurization times, the lowest possible temperature should be used - in other words, if the equipment might be +-2C, pasteurization times should be calculated as if the temperature was 2C below the indicated temperature. The same margin should be used when approaching 55C/131F. When cooking for a long time (days or any time longer than 4 hours at above Refrigerator temps) the minimum temp that one sets on their machine must be 55C plus the same tolerance that is used when the pasteurization time is calculated."

For safety reasons, when setting the offset on a PID controller that might be in a range, set the offset so that the temperature reading is at the low edge of the range.

The comment that excursions on the high side can be tolerated is interesting. I just found a cheap controller that was being advertised as suitable for sous vide, it was not PID based, it was thermostatically based with no proportional capability. You set a target temp and a tolerance which could be as low as 1F. When the temperature, as read by the thermocouple, dropped below the target - tolerance (in other words, if you set 131 and 1 degree, when it hit 129) the heater would turn on full bore until the target temp was reached. For at least one of my appliances, that would mean a 4 degree swing on the high side. So if I was insuring that it stayed at 130+, I would have to set the machine at 132 so that it would turn on when it hit 130, and it would probably wobble between 130-135.

I think that is bad relative to a tuned PID, but would it be OK for sous vide?

[PedroG]

It might be OK as long as you won't cook on the edge of safety.

Temperature swings inside the food are attenuated by the slow heat transfer, see http://egullet.org/p1756373 and http://egullet.org/p1755780; these experiments were done in an oven, i.e. in air, but heat transfer from water to food is much faster than from air to food or from air to a pot of water. So someone with a four-channel-data-logger (mine is only two channels) could do the following experiment: Vacuum-seal a 5cm pile of wet rags (see www.wiki.egullet.org/index.php?title=Core_temperature_development_in_frozen_versus_refrigerated_meat_cooked_sous_vide) to simulate meat, place logger probes at the surface of the pile, at the center of the pile, half way between surface and center, and one probe in the water bath near the sensor of the PID controller. With a SVM, set P=0 (i.e. bang-bang control), set HY (hysteresis) to different values (e.g. 1°C, 2°C, 3°C), and take heating curves from 5°C to 55°C for every value of HY and for a stable water bath with optimal PID settings.