Fan / Cooling questions - I'm going to try to challenge you guys!

[alank2]

Here is what I am trying to do.  I've got a rubidium RFTG unit (see attachment for a picture) and its case is basically a large heat sink.  Inside it in the rear rightward side is a LPRO-101 rubidium module 5" wide x 3.7" from front to back.

My goal is to keep the temperature of the Rb stable by varying the speed of a PWM controllable fan.  I have a temperature sensor on the inside of the RFTG that gives me close to the baseplate temperature of the LPRO and I am feeding that into an AVR running a PID loop to decide how much PWM to produce for the fan.  My testing so far has been very positive, although I've only put the fan beneath the RFTG unit about 2-3" below it and blown air directly at the LPRO area from the underside.  My goal is to keep the LPRO at 50 deg C and my room ambient temperature runs 24 deg C +/- 3 C depending on summer/winter and fluctuations.

The fan I am using can go from 1000 rpm to 4250 rpm depending on the PWM signal sent to it (0-100%).  It is a 60mm square fan and pushes 23.4 CFM at full speed, so I'm assuming at 1000 rpm it is pushing 5.5 CFM so my control range is 5.5 to 23.4 CFM.  So far having the fan under the unit the PID keeps it stable by varying the fan from PWM 10-35% or so.

Here is where I am hoping for some advice.  I am putting this in an enclosure finally and the enclosure is going to be designed to sit directly underneath the RFTG unit.  The fan will be mounted in the bottom of the enclosure pulling fresh room temp air into it.  I am going to drill holes in the top of the unit so it can blow air out onto the bottom of the RFTG (sitting on top of it) to cool it.

I have a CNC so I can cut a nice grid of holes in the enclosure.  In the bottom I'm going to cut a grid that matches exactly the 60mm square foot print of the fan so it can pull air directly into the case without any air leaking around the fan.  The pattern will be 46 holes at 1/4" each or 0.049 in^2 * 46 = 2.258 in^2 for the fan air intake.  My first plan is to cut a grid completely under the LPRO area of 5" x 3.7", 187 holes * 0.049 in^2 = 9.163 in^2 for the air out to cool the RFTG.  So, if I'm thinking right air will be traveling 4 times faster on the input grid (4x less holes) than the output grid.

My big concern is that my PWM range will not give me a wide temperature control range, for example 10% PWM and 100% PWM are not much different.  Ideally I'd want the fan to run at 50% pwm and have some flexibility in cooling capacity if shut down to 0% or up to 100%.

I don't know anything about air cooling so I'm taking a guess at this.  If the cooling is too much even at low RPM speeds I could cut additional holes in the side of the enclosure to vent off some air in a direction that doesn't cool the RFTG, but I'd rather not.  I could cut a more even set of holes that cools the entire RFTG bottom surface and not just the LPRO area as well.

What advice do you guys have?

[AndyC_772]

Sounds like a physical prototype is in order. It's quite difficult to predict air flow without using (costly) 3d flow simulation software, especially if you don't already have the experience to be able to assess the likely outcome by comparison with existing products.

Many fans are available in a variety of speeds, so you could always substitute for a different version if you find yourself lacking airflow, or at the other end of the range, if the fan is tending to stall.

If I were you I'd make a model out of a cardboard box, and machine your intended pattern of holes into a sheet of something that's about the right thickness. Duct tape is your friend!

[Skimask]

1) Tack on a couple of TEC modules just after the fans.
2) Whip up a circuit to run them both ways (heat and cool).
3)
4) Profit...

[alank2]

If I were you I'd make a model out of a cardboard box, and machine your intended pattern of holes into a sheet of something that's about the right thickness. Duct tape is your friend!

That there is mighty fine idea!

[dr_p]

How about constantly cooling it with a fixed speed fan and controlling the speed by PWM heating.

[calexanian]

If I am not mistaken that unit has internal temperature stabilization and compensation already. But my brain squeaks when I walk so I am probably wrong.

[alank2]

How about constantly cooling it with a fixed speed fan and controlling the speed by PWM heating.

It already uses 20W and produces enough heat that I don't want to add more.

If I am not mistaken that unit has internal temperature stabilization and compensation already. But my brain squeaks when I walk so I am probably wrong.

I agree that I think it does, but the datasheet also shows a frequency difference vs. baseplate temperature.  If I've got to cool it anyway I figured why not try to stabilize it.

My first method with the grids posted above failed miserably.  Too many holes meant barely any airflow even at fan over 4000rpm.

Last night I tried a new approach, intake area completely open (I'll get a metal grill), output a small grid of 9x5 about 1 inch apart covering the entire bottom of the RFTG.  Sort of like an air hockey table.  This works much better and is much closer to the fan 4" below the unit I was doing.  The attached graph shows it being pretty stable until my furnace kicked on an hour ago, I probably need to retune the PID loop (the D is disabled on it right now).