09/17/2022 at 09:55 •
With all the virus frenzy behind us now, finally got back into making some progress.
A lot has happened since the last update.
1. Lined the inside of the kiln with 1400C KAO wool and learned a couple of things:
- It is not easy to mount a hot element on KAO wool.
- You have to glue KAO wool to the fire bricks using mortar and the density of high temp wool makes it not want to stick to the sides and top. Eventually used some pins made from Nichrome to pretty much staple it to the bricks.
- Used 25mm wool, reducing the inside dimensions by 50mm on all axis.
- Reached 900 C in about 10 minutes and shortly thereafter reached room temperature as the element burned right where it enters the kiln. Suspect the mounting/"hanging" of the element put too much strain on the single strand entry into the hot face and gravity pretty much did the rest.
- Even though the KAO wool product I used only had a non-carcinogenic irritation risk during handling, after it was heated to over 700 C, apparently it forms structures that are pretty much back to carcinogenic when inhaled again. This prompted me to seal the hotface to prevent those small needles from circulating and venting. Used some water glass; it worked okay but its like glue and sticks to everything. For some reason a yellow deposit formed just behind where the element touched the wool/glass sealed surface after firing. Hoping it was sulfur and not chromium.
- To avoid stuff sticking to the water glass and thus sealing the door shut once fired, had to use some kiln wash on the door to create a barrier.
All up, the KAO wool lining made the internal go hotter faster using less electricity, but the lack of structural integrity on the KAO wool walls caused a element failure and the super fast temperature climb and fall caused the sealed glass structure to crack and decompose even after the second firing. This made the kiln a pretty unsafe tool as it was and decided to abandon the KAO wool hot surface idea.
2. Having removed the KAO wool, the overall structure was back to the original design. Difference now was that the bricks all had mortar between them, sealing the inside more than it was. Cut deeper groves to take the element a bit better and replaced it with some off the shelve 1200W elements (5mm x 480mm).
- Testing revealed similar performance to the original design, suggesting that the sealing did not make that big a difference and convection was not that big an energy drain.
- Cheap hacksaw blades work like a charm to cut the grooves and a lot less dust than power tools. They work for cutting the bricks in half as well but I used 5 blades in total for 4 brick splits and grooves on 3 bricks. Probably because the blades are just carbon steel and the stuff the bricks are made of is the same stuff they put on emery paper which in turn tends to be used to sharpen metal blades (doing the blunting role in this case).
At this point, the only way to get to the ramp rates I was after at higher temperatures was to either go to a 15A power point or start looking at 3 phase/multiple feed points to power the coils. High voltage is scary and adding so much complexity is not something I'd like to troubleshoot and maintain. So major turning point for the kiln design was in order.
3. Pretty much ditched the original brick layout and reduced the overall design to use 7 bricks. Cut all but the base 2 bricks to 50mm thickness and created a 127x127x280 internal size oven. Wrapped the lot in some of that leftover KAO wool from point 1 above and only fitted one coil. Some notes and current performance:
- Front door is a stainless box wrapping some of that 25mm KAO wool.
- Brick structure is now snugly wrapped in a KAO wool blanked, 25mm thick. As a result, the biggest heat leak is the 25mm thick door.
- With less brick mass to heat up and less surface area to conduct heat to the outside, the kiln now goes from room temperature (20 C) to 1000 C in under an hour and up to 700 C in about 20 minutes. Sounds slow, but to speed it up is simple, just throw more watts at it. Currently it pulls a mere 1200W under about 400C and because the coil resistance increases, it looks like it settles at about 1000W over 400/500C.
- Calibrated the PID up to 1100 C and the element did not melt, guess that's a nice bonus and gives me the full range I was aiming for. Just have to be patient with the > 700 C sessions.
- Outside of the kiln was constructed from a mix of galvanized sheet and stainless. The door gets to about 260 C, but the rest of the structure stays under 200 C easily. The bottom half of the structure can be touched and handled; this excludes the floor though as it is in contact with the base bricks and not KAO wool as the rest of the inside.
- Even with the venting hole open, the temperature does not drop by more than a couple of degrees and the PID does a good job to add some more juice when needed.
That wraps up some notes on the enclosure construction and the overall design changes to where it is today. On the software front, some changes include:
- Wifi connectivity to control the system using REST endpoints. Updating the firmware to test small changes to code gets old very quickly and UI is very complex needing lots of changes to tweak and test things. So writing a website on my local PC with the flexibility to change things easily with the aim to use it as a final UI website hosted on the ESP. The touch screen will still contain basic controls and feedback but things like custom schedules etc. will have to be done via the website UI.
- PID now supports calibration with 12 or so preset points. One reader asked how the system would cope with the temperature accuracy across such a wide range; this is how. In general the overshoot is less than 5 degrees on the low end and at the high end (>500C) it seems to be less than 2 degrees. These numbers are not very indicative as it is the temperature measured at a single point in mid air by the thermocouple but I suppose I'm not using it for super accurate lab work and in most use cases some sort of soak is needed to make sure the work piece is a temperature anyway. Suppose in use I'll get to know what the on screen temperature suggests a certain point in the kiln is actually at (like colder at the door and hotter at the top).
- PID also supports ramp rate limiting now. This was implemented using a very basic technique and does the job given the limited resources available on the MCU. So slowing down the heating process on the way to a hotter target temperature is working as well as slowing down the cooling process on the way down. The brick insulation does a pretty good job of slowing down cooling without this feature, but I suspect for glass work it will still be too fast.
Managed to fuse some PMC silver at 900 C for 2 hours. It did a good job as the pieces flex without breaking. Also melted down some silver directly into a graphite ingot mould at 930 C. Guess it was sterling silver. As for PID accuracy, polished up some mild steel and set the target to 290 C. Came out an even, dark blue as the color charts suggest.
That's a lot of writing, but there was even more making to have it all happen.
Next steps include finishing that UI and hunting down some bugs. I think the PID is pretty much done and can be used to control any oven matching the control circuitry it uses. If anyone is interested in building a clone PID component and giving some feedback on notes and code I can get it hosted for so even more people can make their own working builds.
In the meantime, my wife will be using it to fire some of the PMC as the touch screen UI as it stands supports that. I'll use Postman and the REST endpoints to fire some other projects needing custom schedules and temperatures.
In your face CovId.
05/26/2020 at 23:33 •
After the last test, decided to add some fiber padding to the kiln inside. Faster ramp times and better sealing keeping more heat in (is the theory). The blanket is 25mm thick and I'm trying to work out how to mount the element now as it used to be stapled to the brick walls. The fiber is not rigid enough to support the staple approach and it will be coated with water glass which in theory will be molten at the element operating temperatures..
As a result I created some water glass as a side project to seal the fibers and keep them out of lungs.
Currently the frame is pickling in vinegar to get rid of the mill scale. That stuff is evil :)
Once the pickling is done, I'll finish up some final welding needed to get some handles and feet of the oven attached.
Painting then comes next. I managed to get 1093C resistant paint from the auto store. It needs to bake at 200C before it is ready and I don't have anything large enough to bake it in. Thinking I'll just blast it with a flame for a bit and hope the oven at full speed will do the rest on first launch.
05/03/2020 at 10:38 •
Tweaked PID controller to stabilize target swings. Still needs further tweaking, maybe even adding Auto Tune as the oven behaves differently at different temperatures. Increased the PWM from 25% max to 100% max to do a full load test.
Wrapped most of the oven in foil, shiny side inward. Made a big difference to perceived heat emission and it completely blinded the laser temp sensor (was reading room temp all over the surface)
Ran up to 800C to check the thermal stability of everything hooked up. Some findings:
- Max external oven temperature measured was 350C and seems to mostly range between 77C and 123C.
- Between 1400W and 1600W needed to keep oven at 800C. Calculations for ideal oven suggest 817W.
- Controller box was cold to touch with the 80mm fan on. The SSR heat sink reading was max 29C at ambient 24C.
Overall the construction seems sound enough and electronics/electrics can handle the design well.
Think there is a massive loss to convection caused by the gaps between bricks. Also noticed the max power the coils are willing to draw peaked at about 1650W suggesting a 3rd coil might be needed to reduce the resistance. Trying to keep the max current consumption under 10A to avoid needing a special wall socket.
Some other random notes:
- Mains fluctuated between 245V and 226V during the day. Suppose as people start cooking in the complex towards the evening the grid takes a bit of a knock.
- Guesstimating it took about 3 hours effort to get to 800C from 24C room. Was tweaking the PID controller a lot during the day, pausing the temperature on its way to 800C for extended periods. According to my calculations it can take about 72 minutes min.
- Aluminum foil has a little dance on its surface when exposed to convection at ~200C. Suppose its a bit of contraction/expansion competition across the surface as the hot air flows under it trying to escape to room temperature.
- Power meter for the day suggested 6kWh spent on the experiment.
Next potential steps:
- Investigate a layer of Kao wool and brick interleaving to reduce convection and thermal conductivity.
- Reflect on the coil layout to see if a 3rd might fit.
- Start uploading some code to see if the community can identify any glaring improvements.
05/01/2020 at 14:34 •
Great news! First integration test went okay. Hooked one set of coils up to the assembled controller box and fired it up to 108 deg C. Held it for about 1H.
Only two things got hot: The inside of the oven and the 9V PSU in the controller box. Didn't have the 80mm fan on so hoping it will keep the PSU temperature in check. It was hovering around 50 degrees with ambient at 18.
Had some teething issues with noise in the I2C bus and the GPU controller ran out of RAM causing random things to show on the display. Don't think it is all fixed but getting close.
Before running the test I limited the PWM to 25% of its max capacity. This translates to the coil only ever getting power 25% of the time max.
Some measurements while cruising at 108C (18C ambient):
- Coil at 65 Ohms
- Max Temp 110.1C/Min Temp 106.7C
- Max Power 500W / Min Power 270W
- Power meter power factor of 50
- Power meter 4A max
- Multi meter AC Voltage over coil Min 40V Max 80V
- Clamp meter Amps at 0.3A - 0.45A
To the untrained eye these numbers don't add up. The trained eye doesn't care to do the math to recon why not :)
Not sure what's next. Really want to check the controller box can handle full on state for 8 hours. That needs a couple of small problems solved before it can be executed (like the stuck terminal block screw for the second coil, lugs and insulation for the remaining coil wires and more schedules to ramp the temperature to various intermediate levels).
The PID needs some serious tuning but 2.5 degrees swing around the target is not bad for an initial guess.
Wonder how one can calculate the efficiency of the insulation and oven design using the results from above?
04/27/2020 at 13:24 •
Thought I'd add some of the journey towards the final product.
Where is it up to?
- Main kiln body manufactured and fired up to 700C.
- PID Controller Software Developed (Alpha)
- LCD Touch Screen Controller Developed (Alpha)
- Master Controller Software Developed (Alpha)
What's still left?
- Controller Box Manufacturing
- Electronics mounting
- Final Component Integration
- Tweaking body design
Kiln body design in links. Will add code to GitHub as soon as initial integration testing is successful.