Need Help: How to Make a DIY Convection Fan for a Cheap SMD Toaster-Oven

[SuzyC]

My imagination is unlimited, my money is low,
So I chose to make my own oven for SMD Relfow!

My best ideas  so far for my DIY SMD Reflow Oven:

I got me a nice toaster-oven 1000-watts, small enough to store, big enough to heat PCB's like they were pizza slices.

I don't need to make a real convection oven, I just got to stir up the hot air a little..or so I think.

I want to do this using my big pile of electronic salvaged parts from printers, etc, no buying fancy stuff like real-world shaft extenders.
I don't have a machine shop, but I do have lotsa stuff, like  bare PCB stock, simple common work tools, a sabre-saw, files, etc. but no lathe or even a drill-press!
I am getting quite good with coding and electronic circuit design and making PCB's, but mechanical stuff design is not on the top lines of my resume.
No need for a spin-doctor, have no problems getting the motors to dance to my tunes.

My DIY solution:

Cut a round or square piece of a salvaged electronic shield cover from a discarded router PCB. Cut this square of thin metal  into 4 slices and bend each up from the center at an angle  to make fan blades..easy 'nuff. I have a nice metal shears.

Now my best idea was to drill a hole in the center of this blade asm. and somehow connect a shaft through the wall of the oven to a small motor salvaged from a CDROM door-opening 5V DC-Motor.

My best idea to connect to the blades was to buy a length of all-thread and with two nuts and lock-washers I can easily attach the fan blade asm. to the inside-the-oven end of the all-thread.

Now the problem is how to support this "drive shaft" through the thin wall of the interior oven and attach it to the small motor (inside the controls area of the cabinet) and mounted far enough away from the hot hot's of the oven wall.

I have one salvaged motor that has a round shaft  about 3/32 diam and another with a 3/8-in plastic gear attached to the motor and I even have the tiny screws that originally mounted each of these motors.

Uhmm..now what?

Any ideas??

[Mecanix]

Hey Suzzy, I'll see what I can do to help.

You wouldn't have a few pics of your existing hardware by chance? The dimension of the oven interior and position of the 1k heater(s) would help so to call on whether you really need that neutral pressure circulation, or not. Secondly, if really you need that circulation (I doubt at 1000w mind you) I'm happy to look at your motor and shaft assembly and suggest a practical mounting system.

Just hard for us to imagine all that hardware and size it, that's why I ask. If that's a gigantic oven interior then you might be right, it may require that circulation to avoid hotspots.

Let's see what you got and take it from there.

[SuzyC]

Wow, a user name that sounds like someone who likes to do mechanical design!  Thanks for your offer to help.

Making a test on this oven, just a few minutes ago, in a unmodified  oven, no thermocouple, just using the crude theromostat control,a timer and my skilled twisting of the knobs, what i got with just a  couple of SMD resistors was that one resistor was tombstoned and the other was perfect.

Ok, not too much research to support my project yet!

I assume that once I add my own coded MCU control with a K-Type thermocouple with a Maxim chip with 5% or better temperature accuracy and if I do the programming for the recommended temperature/timings to fit the reflow graph for the euctectic solder paste I was using, I should expect much better results. I think I bought the wrong  type of thermocouple, it is a probe type, where a small terminal hole design or even a screw-in type would have been a better choice. This probe-type was the cheapest online,  it has a diam of 1/8-in.

Now I also have to figure out how and where to mount the 9.5-cm long x 1/8-in diam metal probe sticking out of a 4.5-cm plastic handle.

Do I need a fan? I've read that uneven temperatures can cause bad results and even a small makeshift fan would spread the heat.

Here are the specs and attached is a picture of this hotbox.

Capacity: 10 L
     Material: Stainless steel
     Power: 1000 W
     Timer: 60 '
     Max temperature: 230ºC
     Size: 36 x 25 x 20 cm (exterior)

It has two identical heating elements, one below the solid crumb tray and one at the top.

Coming soon..pictures of the motors.


I haven't bought the all-thread yet, having not finalized my elegant design, but all-thread is allthread, the smallest diam. I can get at the local hardware, likely 3/16 to 1/4-in diam size, 12-in long and likely a medium, but not fine thread pitch.

[Mecanix]

haha. It's all about aliases lol Thought yours was genuinely in need of help (no discrimination meant I assure, jjk).

Reason I was asking for visuals is because I plan on modeling a kit for an accurate tell-tell. As you can see from the first visual I'm clueless about the oven's air domain (interior size), how many heaters, and where they are positioned relative to the pcbs, just using an estimated 32 x 22 x 18 cm with 2 randomly positioned heaters. Not that it matters (yet) as you can already see the air moving a whole lot with two heaters (1000w/2) suggesting that you might be spared of that extra fan requirement. However it would be quite nice to run a accurate transient on this infamous 'pizza-oven' reflow thingy (I'm also curious to know whether they actually work).

Let me know if you're keen of providing some visuals and dims of that particular oven's interior

[SuzyC]

Wow! How did you make those thermal images?  I attached a picture with dimensions of the motor I am using with the other  pictures above.

(RF-310T-11400) Short Shaft DC Motor 5.9V Electric Speed Reduce Motor For CD Player


What in the world is a "tell-tell"?

Each of these images suggest a single heater element placement, one image with heaters on the bottom, the other with a heating element on the top..at least to me???

But the really hot need I have is how to make a Micky Mouse Motor Mount  for the motor I have inside the cabinet and couple the short motor shaft  to the interior fan blades.

[spicyjack]

My imagination is unlimited, my money is low,
So I chose to make my own oven for SMD Relfow!

My best ideas  so far for my DIY SMD Reflow Oven:

I got me a nice toaster-oven 1000-watts, small enough to store, big enough to heat PCB's like they were pizza slices.

I don't need to make a real convection oven, I just got to stir up the hot air a little..or so I think.

Question: are you dead set on adding a fan to your existing oven?

The reason why I ask is that most of the SMD toaster oven conversions that I've seen only changed the temperature circuitry (to be more accurate); at the size of your average toaster oven, there's not of places inside the oven where the heat doesn't reach. 

[Mecanix]

tell-tell = "reveal". I made this up, don't look it up the dictionary.

Yeah those dimensions, n of heaters, and their position relative to pcbs would add to the accuracy! Can't confirm whether that thing needs air circulation or not without a bit of help from ya. Might actually do need circulation, who knows you might be correct. Lets place two pcbs on a tray in that model and fire up the heaters transient for let's say 15min and let's see.

I hope you don't need that fan... just so you don't have to bother with all that tacky mounting motor.shaft stuff (yurk). Looking forward your visuals (I'm off for now but I'll rig this for you tomorrow). tc.

[SuzyC]

Hello Spicy Jack,

I saw a Youtube video of a guy who was making his own SMD Reflow Oven and ordered a mini-oven online that was described as a convection oven and then had to discard the item when received because it had no fan.

His video, along with researching the causes of "Tombstoning" suggested that an important cause was uneven heating of the PCB.

OTOH, I noticed most of the Youtube vids show people building their oven without any fan.

I thought adding a small approximation of a real fan might more often make the difference between success and oops results in this hot game.

[Mecanix]

Hey Suzy, assumptions here are a sealed-tight fluid domain(air), two heaters both fired at 300Watts (30%), and a 2mm thick plate of aluminum with two PCBs of 1.6mm sitting on it.

The third visual represent a steady state simulation (timeless, just meant to test whether the model works/converge). From that sim alone it appears that you are correct saying that the temp is not uniform all across (you win?). I'd still need to see the transient sim to confirm that though (tomorrow).

Few questions before I send this 900 seconds/100 steps model over a capable solving cpu; am I close to your exiting hardware and set-up? and are you (others) using both top & bottom heaters or just one when reflowing pcbs? Using a solid or perforated plate? Grill?!

[SuzyC]

Mecanix, your first picture, showing the top and bottom heating elements and a tray on the bottom, is the exact configuration.

I am certain that the most likely best idea here is to make the whole hot party going in my small near-hell simulator to be as much as possible an isothermal experience. 

Another important consideration is the varying thermal masses of the SMD components. Wouldn't a little air circulation with a fan help to ensure even heating of both large thermal mass SMD components(Like a 7805 regulator on a large copper pour) and nearby 0602 size resistors?

After all, professional industrial SMD machines surely use convection fans, why shouldn't I?

Has anyone reading this topic made their own SMD cooker that works excellent every time with/without a fan?

I have and will likely near-always use both the top and bottom heating elements.

The bottom stainless steel (crumb) tray is removable, has slots on the oven's side  to insert it in.

Its a solid, thin, dinner-plate shape, no holes for crumbs to fall through and burn and smoke upon the lower heating element, but there is an air gap in front and back of this crumb tray that allows hot air to circulate.  The PCB itself would be placed centered upon the grill above.

I see a possible instance where if I were to make a double-sided PCB where the bottom layer had already been reflowed, that it would be smart  to turn off the bottom element during the during the last stage of the heating sequence.


But that's not the purpose of my post here. I would assume I am a fan of fans.

I wanna know how someone would make a MacGyver mount for my instance of a MacGyver Fan.

[Nominal Animal]

I would bodge together an all-metal low-RPM fan, with a metal groove pulley on the "cold" side; then use a large silicone O-ring to the motor with the pulley, maybe with an extra fan to ensure the motor completely on the cold side of the oven is kept cool.  Silicone O-rings can be used in up to 450°F/230°C, Kalrez > 600°F/300°C.

Essentially, while the axle and fan pulley are on the cold side, the O-ring would act as the heat break.

[fourfathom]

Has anyone reading this topic made their own SMD cooker that works excellent every time with/without a fan?

I use a toaster oven (one that has two heating elements, one above and one below the tray).  There are no fans in the oven.  The oven itself is unmodified, but I have a thermocouple temp sensor attached to the aluminum tray and use a homebrew external controller to turn the heat on or off.  I get a good thermal profile: ramp/soak/peak/cool, but unless I open the oven door after the reflow is done it takes longer to cool than recommended.  I've built about a dozen small to medium sized boards and have had zero problems.  Well, zero problems when using a solder stencil to apply the paste paste, and occasional rework needed when I apply the paste by hand.  I'm using leaded solder.
The attached photo show a typical board (early prototype with wires and a few hand-changed components).  This one has a mix, mostly 0805.  The smallest footprints are the 32QFN FPGA (0.5mm pitch) and the 0402 TVS diodes.  No problems with these.

[I see that I didn't install the TVS diodes on this board, except for one hand-soldered one.  I didn't have the correct parts when I loaded the board.  Since then I've placed and reflowed the diodes with no problem.]

So for my prototype needs the oven works fine, but hand-placing the parts is horribly tedious.  For volume work I'm going to use an assy house.

[fourfathom]

The bottom stainless steel (crumb) tray is removable, has slots on the oven's side  to insert it in.

Its a solid, thin, dinner-plate shape, no holes for crumbs to fall through and burn and smoke upon the lower heating element, but there is an air gap in front and back of this crumb tray that allows hot air to circulate.  The PCB itself would be placed centered upon the grill above.

For what it's worth, I removed the grill (rack) and slid the aluminum crumb tray into the rack position.  The PCB and thermocouple are laying on the crumb tray.  I haven't tried putting components on both sides of my boards,  My thermocouple is attached to a scrap PCB, which is secured to the tray with a screw.

I tried using a thick aluminum plate, and a steel tray, but there was too much thermal mass and I couldn't get the thermal profile I wanted.  The thin aluminum crumb tray works well.  I've seen no evidence of uneven heating. but my boards are fairly small: 3" x 3.5"

Both oven heating elements are connected together.  I haven't seen the need to control them independently.

[Mecanix]

The ultimate dream reflow oven and way to go is forced convection. Suzy was correct. However it is not as simple as just fitting a fan, flowing an unknown cfm all over and hoping it all works out. It may in fact decrease the effectiveness (positive/negative pressures comes to mind). Our industry experts appears to harvest the hot air (where it is hot) and force that onto what needs heating, if that optimum performance is required mind you, and the reason why professional reflow ovens are fitted with similar. See linked research paper where you can visualize such forced conv. fan intake & flow design of a baking oven (pdf page 9 of 11).

https://pdfs.semanticscholar.org/ca0a/5a5f95fc20853f580a91bd88fb7c98de1ba7.pdf

All that to say; if you know exactly what you are doing then go ahead and install a force convection fan mechanism, and if you don't then probably best to remain on the natural convection system and position your thermocouples (several and scattered being better) exactly where the heating needs to take place and control the heaters with that data. As per "fourfathom's" findings and recommendation.

Last option remains of course trials & errors. Quite a lot of work, potentially significant energy waste, components, pcb... you get the picture.

[jmelson]

I've been doing this for 14 years, using a "GE" toaster oven.  Mine has a tiny convection fan, I have tried it, it doesn't seem to do much.  The fan might be too small.  Anyway, my system is to use a tiny thermocouple poked into a through-hole in the board.  This way, the controller is sensing actual BOARD temperature, not air temp.  The IR from the heating elements heats the board a LOT more than the air.  I first tried just putting the thermocouple in the air, and got burned boards.  One difference is my oven has TWO elements below the tray and two more above the tray.

I do mostly lead-free with GC-10 paste, and have had GREAT results with this setup, I've done over 2000 boards with it.

One other note, I use a "ramp and soak" temp controller that allows you to program timed temperature ramps and holds.

Jon

[Mecanix]

One of those thing where you get to know your tool, find the sweet spot, stick to it and improve upon that. 14 years in I bet you could teach us a trick or two on that subject! Good insights, fourfathom also had valid points.

New to me. May look into procuring my own reflow kit eventually and as Suzy did I was thinking about getting a mid sized oven/mcu/thermocouple/relay system. Sounds so fun and look so easy but sadly it seems pretty complex in practice. For instance changing the thickness of the plate those pcbs sits onto and/or simply changing that plate material from AL to SS gave completely different results in the natural convection set-up, one heater -vs- two, etc. It all over the place. And hence why I've concluded (like Suzy, and other experts); just force air on it and be done with it.

Challenge is, what sort of air flow and pressure is required for this thing to perform. Could run CFD sims all day all night to find that perfect equilibrium, and it will, easy said, but that require an incredible amount of processing power to solve. Not being into the business of baking chocolate cookies at the first place might waste me a week or two (more?) of trial and error before it even starts to make sense (potential for that $ouch$).

Anyway. Subscribed. Let's see what Suzy decide on doing... 

[SuzyC]

Wow! Lotta good feedback here.
Many additional thanks to all!

How much for and where  do I find/buy a teeny-weensy thermocouple that can fit in a via?

Jmelson: the toaster reaches a certain temperature in the reflow sequence..but then the IR heat lamps turn off for a soak..so wouldn't stirring up the air help?
 
My idea is not to make a true convection system, just stir the air while the board is soaking. A powerful fan might also move some parts around. I don't want to do that. The idea is: prevent temperature gradients developing.

Mechanix, the idea of me finding and mounting a grooved pully to what kinda shaft and then finding the right high temp o-ring..ha ha ha ha!
I don't even know how mfgs manage to put gears on motor shafts or even how then to take one off.

I am trying to just get a length of allthread and somehow find, with my treasure pile of salvaged scrap materials to make a way to mount the small motor and probably then  try to use shrink tubing and epoxy to couple the allthread to the small motor shaft on the cool side of the oven's control compartment. Where the allthread fits through the oven side I can file down the allthread to almost smooth and its no longer screwing with the oven wall, but not quite working like a sleeve bearing.  That's my best idea so far.

I don't even care if the fan I make is very noisy.  I am happy that I  have a Nuke in the kitchen that has a very noisy squirrel-cage motor. I only have to hear it grumble and complain so long as it takes to make my dinner.



I don't want to spend a lotta bucks and a few months waiting for expensive mechanical parts I would have no skill or tools to work with or even know how/where to order.

I can buy a much larger convection toaster-oven for about two(has a small fan)  to ten times (true convection oven) the price I paid for one I got. I like the one I got because it is small, cheap and maybe even work well with a few mods.

Remember, I am talking Micky Mouse/MacGyver technology, not industrial professional technology. I just want to get the job done somehow and now, having just enough reliability that it will work a reasonable number of SMD cooking sessions before repair.

[beanflying]

Worth watching for background. Also as these have been used for Coffee Roasting (Rotisserie models) lots of benefits to adding some 'appropriate' insulation for much better ramp control.

Both Coffee Roasting and Reflow share some common attributes with Ramp and even airflow control and finally cooling.

[Mecanix]

For your existing fan and motor, shall you decide to go on with this idea, the part that you need to mate both the fan + motor shafts together is called a "coupling". An example of the part below although quite overkill in price and mat so see if you can find an equivalent in aluminum for the shaft ID dims you have. Assuming your fan is in The Very Tiny family of fans (weight-wise) here, Suzy. If weighty then a dual bearing mounting system will be required.

For Shaft Size 3/32", Max. Torque: 50 Oz/In
https://www.amazon.com/Stainless-Mini-Joint-Coupling-Length-Torque/dp/B000PKOEDE

Keep in mind that the motor you have is not designed for high temp. That heat will most certainly reach the motor through its shaft (thermal expansion) and have it fail sooner or later. See if you have the patience to find something that half-resemble to what you're trying to achieve within your constraints and budget in there:

Convection Oven Fans for sale | eBay
https://www.ebay.com/b/Convection-Oven-Fans/43566/bn_7023534666

[SuzyC]

Mecanix, many thanks, but you have shown the throw money and time upon the problem professional way to do shaft to shaft coupling.
Your convection oven link doesn't work. The shaft coupler item you show that was offered on Amazon is listed as oh oh no-more-of-dem.

I see China industrial mini (steppermotor) shaftcouplers sold on ebay that I could get for a few bucks and free shipping and I even could have it to my door by sometime like the beginning of summer, if the US hasn't declared war on China.

https://www.ebay.com/itm/Stable-Shaft-Coupler-Plum-Shaft-Coupling-Reliable-Small-Insulation-Stepper-Motor

But a real metal (brass) shaft coupling would have quite good mechanical but also quite good but unwanted good thermal coupling.


How would you MacGyver it?

My idea so far. handcraft a shaft coupler that  has poor thermal coupling, good-enough mechanical coupling, cool everything in the controls area of my hotbox with a 2-in  mini-fan salvaged from a small backup diskdrive enclosure that I already have on hand.

(My Proposed Solution)
Make a sandwich coupling. The Cola-Can (patent bending) MacGyer shaft coupler device.

Kinda like a makeshift u-joint/coupling  made from something like  small rectangular piece/pieces of cola can metal. Use 180-deg folding to bind shafts to coupling to create a cola-metal sandwich with a few small nuts/bolts to tighten/bind the folded metal coupler to both of the shafts.   The shape: imagine one piece which is two folded squares of cola-can metal to bind to each shaft and a small not wide rectangular linkage area of the ColaMetal that connects the two shaft-binding folded rectangles together but has poor thermal coupling.

What's nice about the art of ColaCan fabrication is that no fancy lathes, drillpress are required. Only a scissors and an electric drill.
(and its low-cost, available everywhere(except Antarctica) and you can drink the soda too!)

I wouldn't be concerned about exact mass balance and  even a little vibration would be music to my ears.

Remember, the entire ride of a PCB in the hotbox only lasts for a few minutes, and this means it doesn't take enough time to heat everything up to meltdown in the controls area, especially with a small fan to breeze the heat away.

Anyone else have an idea for a better MacGyver shaft coupler?

[Nominal Animal]

Anyone else have an idea for a better MacGyver shaft coupler?

Silicon tube?  Inner diameter a little bit smaller than the shaft diameter, so that it's a tight fit.  Yes, it will lose some torque, but for a fan/air stirrer, you don't need much anyway.

Downside is you need the fan shaft and motor shaft to have pretty much the same diameter, except for how much you can get the silicon tube to stretch.

[jmelson]

New to me. May look into procuring my own reflow kit eventually and as Suzy did I was thinking about getting a mid sized oven/mcu/thermocouple/relay system. Sounds so fun and look so easy but sadly it seems pretty complex in practice.

Well, it actually looks like black magic!  I did hand assembly for YEARS, with soldering iron and tweezers.  When I got my first board to reflow, it was just stunned at how simple and easy it was!  There's actually a LOT of latitude in the process.  The biggest learning curve was the stencil aperture design!  Using the right stencil thickness and aperture size is really important.  I reduce all apertures to be a bit smaller than the pads on the boards, and the aperture reduction is more for smaller pads and especially for finer lead pitch ICs.

Also, good solder paste is essential.  I have used some "big name" solder pastes that worked very poorly, and am now using Loctite GC10, which is just AMAZING!  I have much less opens and bridging with the GC10 than most other solders.

For instance changing the thickness of the plate those pcbs sits onto and/or simply changing that plate material from AL to SS gave completely different results in the natural convection set-up, one heater -vs- two, etc. It all over the place. And hence why I've concluded (like Suzy, and other experts); just force air on it and be done with it.

I do not use a plate, my oven has a wire "rack" that slides out.  Since I often do two-sided board assembly, I need to be careful to not disturb the parts already soldered on the first side when reflowing the 2nd side.  I have some 1/8" stainless rod that I bend up with vise and pliers to make a custom support for various boards.  It generally supports the board by the edges, so nothing touches those parts on the bottom.  I do the bottom side of the board first, Usually has less parts than the top, and then place and reflow the top side components.  I do as many small boards as can fit in the oven at a time, often 6 or 8.

Challenge is, what sort of air flow and pressure is required for this thing to perform. Could run CFD sims all day all night to find that perfect equilibrium, and it will, easy said, but that require an incredible amount of processing power to solve. Not being into the business of baking chocolate cookies at the first place might waste me a week or two (more?) of trial and error before it even starts to make sense (potential for that $ouch$).

I think you probably need to use a pretty hefty fan for convection to really work.  I've heard people discuss parts being blown off the board!  I think these ovens are too small for convection to really work.  You don't want the fan to suck in cool outside air, that will cause a huge temperature gradient across the oven.

The 4-element toaster oven just works so well in the small PC board shop, I just don't see the need to go building something super complex.  If the 2-element ovens have the element far enough from the rack, that may also work quite well.

Jon

[jmelson]

How much for and where  do I find/buy a teeny-weensy thermocouple that can fit in a via?

Well, I bought a 1000 foot roll of micro type-E PTFE-insulated thermocouple extension wire on eBay for a song.  I just strip and twist the ends together, I don't have a thermocouple welder here.  But, type-E is an unusual type.  You won't find a controller that comes set for that.  BUT, almost all modern controllers have all the conversion polynomials built into them, you just have to find out how to get into the secret 3-levels-down menu to change calibration type.  If you want to go that route, I could give you some of my wire.
I will obviously NEVER use it all!

Jmelson: the toaster reaches a certain temperature in the reflow sequence..but then the IR heat lamps turn off for a soak..so wouldn't stirring up the air help?
 

Well, it is NOT the air that heats the boards, it is the IR straight from the element.  When I get up to 230 C or whatever, (peak lead-free temp) the elements are glowing visibly red.  That's like 900 C.  I think for a convection oven to really work, they have wide copper plates just above and below the board that are heated to 300+ C, and the convection air blows in a circular pattern between the boards and the plates.  A thermal profiler rides on the conveyor to find out what temperature-time profile is given on a sample board with thermocouples attached to it.

My idea is not to make a true convection system, just stir the air while the board is soaking. A powerful fan might also move some parts around. I don't want to do that. The idea is: prevent temperature gradients developing.
/quote]
Well, the fan is small on my oven, I have tried it, it DOES NOTHING, as far as I can tell.  If anything, because it is on one side of the oven, it may make the temperature gradient WORSE.

I can buy a much larger convection toaster-oven for about two(has a small fan)  to ten times (true convection oven) the price I paid for one I got. I like the one I got because it is small, cheap and maybe even work well with a few mods.

Yes, get the largest oven you can find.  Mine will take a sheet of US letter paper-sized board (9 x 11") and has 4 elements.  Maybe this is important, the small ovens might only have 2 elements.  There is some temperature gradient, but it is certainly tolerable.
I do have a ramp and soak controller, this means I can duplicate the ramp-hold at 180C-ramp-hold at 235 C- and done profile pretty accurately.  So, I can start the cycle and walk away while it is running.

As for reliability, this thing has been in use for 14 years.  I have a wire soldered onto the built-in thermostat, and that has a bimetal element that gets hot.  So, I had to get inventive to keep that wire from de-soldering itself.  That is the TOTAL maintenance issue I have had with it.  You can watch the PV and SV displays on the controller to see how accurately the board temp is following the programmed profile.  The board lags the setpoint by 10 - 15 C while ramping up, then overshoots by 5C when it goes into the hold, and then lags hugely when the cooldown is startted.  I open the door a crack at first, then more after the board is below 170C.

Jon

[Mecanix]

Only a scissors and an electric drill.

I sincerely doubt a simple electric drill will work, needs to be cordless, with the DeWALT yellow trims preferably.

We use PEEK material in the engineering composites for high temp. Assuming you won't ramp that thing above 275~300dC it should then live and limit that thermal coupling property you've highlighted.

[Mecanix]

Additionally, I'm surprised nobody recommended the Great Fire Extinguisher. When CE approved household hardware gets significant modifications yadayada...
So I'll recommend one of those too, in the correct class, and in the fire dept's RED kinda trims.

Have fun, stay safe Suzy.