More on incubator design

I started thinking more about the freshman design project to design an incubator (see Temperature-control project for freshman design seminar  and PWM for incubator).

The box they’ll be using has a capacity of about 15l, which at 25°C is about 18g of air, and dry air has a heat capacity of about 1J/g/°C, so to heat the air to 37°C would take about 12°C * 1J/g/°C * 18g = 216J.  A 10W heater would get there in about 22 seconds, so the students are unlikely to need anything beefier than 10W. Too powerful a heater would be difficult for them to control, as the delay from changing the power input to the temperature of the air changing could be a long time, and a lot of heat would get stored in the resistor and heat sink, continuing to heat the air well past the set point.

I’ve been thinking of using a 9V supply, because I happen to already have a 9V/6A supply, but I could use 12V supply also (I have a 12V/2A supply, and beefy 12V supplies are readily available).  Power resistors rated at 50W are fairly cheap (particularly from Mouser), but ones rated at 100W start getting pricey. The advantage of the higher power resistors is that they can be used without a heatsink at the lower power will be using them, and their surface temperatures don’t get as high (reducing the hazard of the students setting things on fire).

resistance	rated power	power @9V	power @12V	cost
20Ω	50W	4.05W	7.2W	$3.78
18Ω	50W	4.5W	8W	$2.43
15Ω	50W	5.4W	9.6W	$2.38 or $3.78
12Ω	50W	6.75W	12W	$2.31
10Ω	50W	8.1W	14.4W	$2.53 or $3.64
10Ω	100W	8.1W	14.4W	$9.78
8.2Ω	50W	9.9W	17.56W	$2.25
5Ω	50W	16.2W	28.8W	$3.64
3.3Ω	50W	24.55W	43.64W	$2.53
3.3Ω	100W	24.55W	43.64W	$10.86
3Ω	50W	27W	48W	$3.64
3Ω	100W	27W	48W	$10.76 or $11.74
2Ω	50W	40.5W	72.2W	$2.31 or $3.64
2Ω	100W	40.5W	72.2W	($7.15 + $2.98 bracket)
1.8Ω	50W	45W	80W	$2.53 or $4.61
1.5Ω	100W	54W	96W	$10.76

Even a fan in the box is likely to add significant heat (1—2W), so it may be necessary to add a (controllable?) cooling port to let in cooler air, otherwise the box would heat up at about 7°C/minute, until the temperature was high enough for there to be significant heat loss through the thick styrofoam walls (which is probably past the point where the fan overheats or the inner surface of the styrofoam melts).

If they use a fan with a tachometer (usually 2 or 4 pulses per revolution), then they can’t use an nFET to switch the ground, as the tachometer is referenced to ground.  But even a high-side switch for PWM control of the fan causes problems, because the tachometer uses a Hall-effect device that is powered by the same power as the fan.  Although there are tricks you can use (like pulse stretching—temporarily using 100% duty cycle on the PWM to read a few pulses from the tachometer), the programming gets beyond what I’d want to be handing to the freshmen.  Changing the control input to make measurements in a servo loop is also rather inelegant—changing the frequency with which you make measurements changes the behavior of the system because of the change in the average control value.

So it looks like I need to use 4-wire fan that has a separate PWM input.  I could have them use something like AUB0812L-9X41, which is a 12V motor with open-collector tachometer output and PWM input.  The threshold on the PWM input (V_IH > 2.8V) is low enough that one can control the fan from a 3.3V processor (though we’ll need a 5V processor for low-side switching of the heater).  One problem is that the motor does not run below about 10.8V, so I can’t run it off my 9V power supply.  All the PWM control fans I’ve seen have a minimum voltage over 10V, which would mean not running them on my 9V supply.

Also the fans with PWM inputs are fairly heavy-duty (1.08W or more), but this application really calls for a very low-power fan to keep the the air stirred up for uniform temperatures without warming the box.  The lowest-power PWM fans I’ve found are the AUB0812L-9X41,from Digikey@1.08W and 9GA0812P6M001 from Mouser @700mW (but that’s a $20 fan).  Both require over 10V (10.8V for the Delta Electronics fan from Digikey, 10.2V for the Sanyo Denki fan from Mouser). The Sanyo Denki fans are a bit unusual, in that 0% duty cycle on the PWM control does not correspond to the fan being off, but running at a low speed.  Most of the other fans will stall below about 30% duty cycle, but can be turned off completely at 0%.

Fans with tachometer but not PWM go as low as 0.1W for 3V fans, 0.25W for 5V fans, or 0.36W for 12V fans).  Many of them have a high minimum voltage also, but the very cheap ME40101V1-000U-G99 by Sunon is rated for 4.5V–13.8V and starts at 4.5V.  It even has a locked-rotor feature with automatic restart.

I have several choices: a beefy 12V supply for the heater and fan,  running a fan continuously without speed control, using separate power supplies for the fan and the heater, doing pulse stretching with an ordinary 12V fan + tachometer (to read the tachometer speed), or switch to a 5V PWM fan.  Hmm, 4-wire 5V fans don’t seem to be available, so that option is out. I was hoping to use the 9V/6A supply that I already have to test stuff out this summer, to see how hot the resistors get and how fast temperature rises with different power levels (partly through using different size resistors, but mainly through PWM). I want to test out the power supply also, as I want it for another project, and need to see how hot it gets under heavy load.

I could handle the extra programming needed to do pulse stretching for a more ordinary 12V fan and tachometer run at 9V, though I wouldn’t ask it of the students. I also have a somewhat wimpier 12V/2A supply that might be enough for the incubator, so maybe I should try doing everything with a 12V supply and a 12V PWM fan—perhaps the small, cheap Sanyo Denki 109P0412P3H013, though I can’t find a datasheet that tells me how much power it takes. If I’m guessing correctly about their part-numbering system, it is probably the same as 109P0412H3013, but with PWM circuitry added.  It has the right flow rate, but the noise rating is higher than Mouser claims for the 109P0412P3H013, which is unbelievably low.  (Mouser has a problem with sloppy data entry in their tables, so I never trust their numbers unless confirmed by the manufacturer’s data sheet.) It is strange that the official Sanyo Denki site does not admit the existence of the 109P0412P3H013 part.    The underlying fan runs down to 7V, so it might be worth seeing if the PWM version also runs at that low a voltage.

Even if the PWM fan doesn’t run at 9V, I can still use the power resistors to test whether the 9V supply really delivers what it claims, which is my other reason for wanting the power resistors.

Of course, I’ll need a big heat sink for the resistors—their power rating is based on their being attached to a big hunk of metal, and the 50W resistors can only go up to 20W without a heatsink.  I’m thinking of getting a piece of  6″ × 12″ sheet aluminum (maybe 1/16″ or 1.6mm thick) to mount the resistor on, to spread out the heat over the bottom of the incubator box.  I could mount the fan and baffle on the same piece of metal, if I add some angle brackets.

For the lab, the students will have access initially to an adjustable bench supply, which could deliver 0—25V @1A and could be set to 12V (so a 12Ω resistor would dissipate 10W and a 20Ω resistor would dissipate 7.2W), but they wouldn’t be able to deliver more than 12W at 12V with that power supply.  If it turns out that more power is needed, they’ll have to buy a power supply.