Armory Sponsor
Posted: 4/5/2013 8:55:25 PM EDT
|
I have a Sturdy safe bolted down out in the garage that gets cold when it's cold outside. I consider that completely unacceptable, so I'm going to solve the problem in the most over-wrought and unnecessary way possible:
Home-brew closed-loop temperature control!  (On a tiny budget)
Why do this? Why not? But seriously, 1. I bring stuff from the safe → house all the time. You'll get immediate condensation on items if it's cold out there when you do this. 2. It will lower the relative humidity in the safe. I keep some documents and stuff in there that I'd prefer didn't get mildewy. 3. I don't like a cold carry gun. 4. It's awesome and way more fun than my research topic. The plan is: Measure air temperature with a sensor, and apply a feedback algorithm to drive a shielded 100W lightbulb inside the safe so that we keep the internal temperature at 75 deg (in the winter). I am using a custom digital controller based on the Arduino platform, and will be controlling current through the lightbulb with a MOSFET controller. The model I've put together of the safe/guns/garage thermal system is rough as all hell, but in general, we say that: 1. Heat q_bulb enters the safe from a lightbulb (which we throttle up and down). The air inside is well-mixed by a small fan. 2. There is a “load” mass of guns/ammo etc which takes time to change temperature. 3. Heat q_wall leaves the safe via convection to and from the (insulated) walls. 
The lightbulb doesn't operate oven-style (pure on-off) because that's no fun, and leads to very quick burnout of the bulb. By continuously varying the output current, once the safe contents have reached their desired temperature the bulb can throttle back to a small fraction of its maximum output, which will extend its life. The bulb should have no problem bringing the safe up to temp, even when it's very cold outside. Based on a simulation of the model, if we start from 32 degrees inside and outside the safe, then running the bulb full-blast can bring the safe up to near 160 deg F: 
I will be updating this periodically as I get stuff done. If anybody has any interest, I'm estimating total cost at <$60, and I will be posting all code/circuits/parts lists separately. And yes, I have nothing better to do on my Friday night than drink Fat Tire ale and write code. 
|
|
TL:DR
Goldenrod EDITED TO ADD... I had the gun safe in a walk in basement. It was located in a cinderblock side room about 3 inches lower than the basement slab. About twice a year I would get 3-5 inches of rain and the side room would have at least 2 inches of standing water. The gun safe was on pavers to keep it above the standing water. I had a single goldenrod in the safe and there was NEVER any rust problem in the gun safe! |
|
Quoted:
What are you using for your q-wall value? I would be surprised if a 100w light bulb would hear a gun safe to 160 deg. I like the idea of using a bulb for a Hester since it kills 2 birds with one stone. Changing the bulb would be a little annoying though. The effective R-value of the safe walls, including convective coefficients on the inner and outer walls is computed to be around 3.3. That is an extremely rough estimate, however, and could be off by as much as a factor of 3. That much error will still put my final temps in the butter zone, though. The safe does have 3" of fiberglass and ceramic fire insulation in the walls, so I wouldn't expect R=3 to be out of the question. |
Here is the safe in question:
I'm going to be drawing air across the lightbulb with a small fan so as to avoid creating any local high temps that could burn or melt stuff. A metal box will contain the lightbulb and fan and help make the air circulation more effective. I've tacked a plate into the box to hold the lightbulb, and the wall of the safe will create the 4th wall of the box. 
With the fan and bulb installed, it will be something like this: 
I will be attaching it to the back wall of the safe in such a way as to circulate air throughout the interior without getting in the way too much: 
|
|
A few questions:
1. is the safe fire lined? [1] 2. what convection coefficients did you use for air/metal metal/air transfer? 3. have you run through the psychrometrics in your garage to see how far the RH is going to fall with your safe at 75ºF, and your garage at <40ºF? The absolute humidity should be nearly constant between the garage and the safe, so I think you might have an issue with the RH dropping too low inside the safe . You might be in danger of cracking a wood stock. No wooden stocks, no worry. Also, how do you plan to data log the Temp vs Time? I'd be interested in seeing the transient response, and comparing this back to the original model. [1] judging by the knee in the curve at 1 hr, it looks to be fire lined. [great thread, btw] |
|
Quoted:
A few questions: 1. is the safe fire lined? [1] 2. what convection coefficients did you use for air/metal metal/air transfer? 3. have you run through the psychrometrics in your garage to see how far the RH is going to fall with your safe at 75ºF, and your garage at <40ºF? The absolute humidity should be nearly constant between the garage and the safe, so I think you might have an issue with the RH dropping too low inside the safe . You might be in danger of cracking a wood stock. No wooden stocks, no worry. Also, how do you plan to data log the Temp vs Time? I'd be interested in seeing the transient response, and comparing this back to the original model. [1] judging by the knee in the curve at 1 hr, it looks to be fire lined. [great thread, btw] Safe is indeed fire-lined with 3" ceramic wool. Inside the safe, the convective coefficients are based on a forced-convection model with an average free-stream velocity of 0.1 m/s. Outside the safe, they are based on a free-convection model. Total thermal resistance through the exterior wall is the serial connection of convection->steel->rockwool->steel->convection. For the load mass, I'm using a really dirty lumped-capacitance model (assuming the dimensional temperature differences within the load are negligible) with forced convection. The load contributes a good portion of that knee at the start of the graph. No wood stocks, so I'm not bothering with a humidity model. |
|
Quoted:
Why not use a goldenrod or similar device? Just to be complicated? When the filament in your bulb goes out, safe drops temp and condensate forms. K.I.S.S. and go with of the shelf technology designed to run indefinitely. I don't want to use a Goldenrod because it won't keep the safe at the same temperature as my house. The system I'm talking about will maintain a constant temperature inside the safe regardless of outside temps. The controller is designed to throw an alert when the bulb blows, so I will know right away that it needs changing. Is it needlessly complicated? Absolutely! But just like Ed Hubel and his "12 Gauge From Hell" thread, sometimes I like doing stuff people haven't tried before just for the hell of it. |
|
Quoted:
Quoted:
Why not use a goldenrod or similar device? Just to be complicated? When the filament in your bulb goes out, safe drops temp and condensate forms. K.I.S.S. and go with of the shelf technology designed to run indefinitely. I don't want to use a Goldenrod because it won't keep the safe at the same temperature as my house. The system I'm talking about will maintain a constant temperature inside the safe regardless of outside temps. The controller is designed to throw an alert when the bulb blows, so I will know right away that it needs changing. Is it needlessly complicated? Absolutely! But just like Ed Hubel and his "12 Gauge From Hell" thread, sometimes I like doing stuff people haven't tried before just for the hell of it. I get all that, But why not use a purpose built heating element that is robust, rather than a light bulb? I'm sure you could accomplish the same goal and have a much more durable solution. The golden rods get up to 150 degrees or so. |
|
Quoted:
Quoted:
Quoted:
Why not use a goldenrod or similar device? Just to be complicated? When the filament in your bulb goes out, safe drops temp and condensate forms. K.I.S.S. and go with of the shelf technology designed to run indefinitely. I don't want to use a Goldenrod because it won't keep the safe at the same temperature as my house. The system I'm talking about will maintain a constant temperature inside the safe regardless of outside temps. The controller is designed to throw an alert when the bulb blows, so I will know right away that it needs changing. Is it needlessly complicated? Absolutely! But just like Ed Hubel and his "12 Gauge From Hell" thread, sometimes I like doing stuff people haven't tried before just for the hell of it. I get all that, But why not use a purpose built heating element that is robust, rather than a light bulb? I'm sure you could accomplish the same goal and have a much more durable solution. The golden rods get up to 150 degrees or so. Gotcha. While the Goldenrods get hot, they are all such a low wattage that I wouldn't be able to get the safe to the proper temperature under all conditions. I could just fill the safe with Goldenrods, but that would cost an absurd amount of money relative to the power output. There are specialty wire-wound resistors that are often used for this application, but they are expensive. A bulb is a very cheap heating element of a known wattage that I can replace easily. Since the bulb will spend most of its life throttled down to very low levels, I should get years of life out a single bulb. If this solution works well, I can always replace the bulb with a more robust element somewhere down the road. For now, it will get the job done. |
I was hoping to have the controller figured out tonight, but that won't be the case. I linearized the system so that I could apply an LQG regulator with Kalman filter to estimate the unknown states, but unfortunately some very important system characteristics were lost in the process, and I don't trust the resulting controller.
While this appears to work, the temp rises too fast to be real and the steady-state input required is much higher than that predicted by the nonlinear model. I will switch to a brute-force nonlinear control instead. |
The nonlinear control appears to work well. While the steady-state power output required by the bulb is higher than I thought, it should still allow effective heating down to -20 F.
You can see the slow thermal-soak into the load mass at the end, which is what I would expect. I've also gotten the electronic control board mostly complete. This board contains the logic-level components, such as: Microcontroller Fan relay Power supply jack Data logging connectors Thermometer connector 
The high-voltage components, such as the MOSFETs and light relay will sit on a separate board behind an opto-isolator so I don't fry anything. |
| would this work for you? because it seems like you are making something really complicated that doesn't have to be... |
| Typical gun safe products won't do what the OP is looking for; the wattage just isn't high enough. He needs to keep the inside of his safe within 10-15ºF of the house to avoid condensation when he brings firearms from the safe into the house during winter months. |
Armory Sponsor