Building a 1000°C Furnace for Semiconductors

Comments · 33

• MR

  @Mr.Leeroy · 1 year ago

  crimp the wires together using a ferrule

  6

• HY

  @hype_synergy · 1 year ago

  Great description and walk through the process. I would suggest to coil the Kanthal wire then wrap around the tubes. This allows you to use a longer wire with larger gage (16 G) and it allows the wire to expand and contract without issues. A thicker wire lasts longer since the wire oxidizes as it heats and at some point breaks,

  1

• PV

  @pvc988 · 1 year ago

  When you are making a connection with heating wire you fold a couple of centimeters of the ends on themselves and twist them. This way the ends will have less resistance per unit length and are going to heat up less than the rest of the wire. With that you can use regular ceramic terminal blocks to make connections with supply wires. You can also double fold if temperatures are still too high.

  14

• UL

  @UlrichHarms-ci1ov · 9 months ago

  Safety is an important point. Make sure the metal case has a good ground connection and that the outlett used has GFI protection. The glass fiber insulation may want an extra layer on top for mechanical protection where it goes through the metal. In the US with a defined neutral and life wire, just 1 fuse in the life side should be good enough.<br><br>For the rather high load the rules usually limit the use of PWM speed. To be within the limits one may need a relatively long cycle and maybe synchronization with mains to use full cycles and avoid a DC component in the current.<br><br>Much of the heat transfer is via radiation. This works well at high temperatures, but no so good at lower temperatures. This can mean that rather low temperaures may need a different set of PID parameters with a slower response.

  1

• XZ

  @xzddakfdmiug5832 · 1 year ago

  I can&apos;t wait to see the next video in this series to make a semiconductor.

  3

• MO

  @mostafagaberahmed6657 · 1 year ago

  writing code with C is fantastic job , you are the one, and every one this days use Arduino&apos;s library

  1

• VE

  @Veldrade · 1 year ago

  This was really interesting ! Can&apos;t wait to see you build a homemade ion implanter though

• JA

  @JaenFoster · 1 year ago

  It&apos;s been a while but if I remember correctly it&apos;s better to take the input of the derivative directly from the output so it&apos;s looking at the rate of change of the output, not the rate of change of the error.

  4

• PV

  @pvc988 · 1 year ago

  When you&apos;re drilling thin sheet metal, put sacrificial piece of wood on both sides and drill through that. This way you are going to get much less tearing and much nicer edges.

  3

• HA

  @hardywoodlandcypress · 1 year ago

  Great work, with fairly simple tools. &nbsp;Having recently completed an IR reflow oven, here are a few ideas that worked for me. &nbsp;You&apos;ve already got the PID loop down pat, and the second PID loop was genius, so you probably won&apos;t have a need for this, but here it is. &nbsp;I guessed at a thermal model for my oven: heater power (known), heat capacity of heater (unknown), thermal resistance to chamber (unknown), heat capacity of chamber (unknown) and thermal resistance chamber to ambient (unknown). &nbsp;That&apos;s 4 unknowns, in a 2nd order system. &nbsp;To find them, log the input power and temperature over a full excursion from ambient to max and then back down to ambient. &nbsp;On a PC, use scipy etc. to find the best parameter fit. &nbsp;Once you have a reasonably good fit, write code to simulate the model response in the controller (easy). &nbsp;Then, the controller simply models each case at every point in time (e.g. every 100ms): heater on, or heater off; then decides which decision best approaches the setpoint. &nbsp;No PID loops required. &nbsp;Response speed is faster, and by modeling far enough in the future, you can avoid any overshoot. &nbsp;It&apos;s like a bang bang controller except for 2nd order systems!<br><br>For sheet metal: I found that McMaster sells thin (10mil) stainless sheet - this is very useful for furnace work. &nbsp;It&apos;s a bit tough to work. &nbsp;I avoid machining, and use tin snips and spot welding. &nbsp;You can make terminals out of it since it spot welds nicely to Kanthal A1. &nbsp;Always drill and saw between plywood sheets.<br><br>You should not be able to break quartz by heating/cooling too quickly. &nbsp;You can heat it to red heat and dunk in water, and it won&apos;t crack. &nbsp;Heater wire is very gentle in comparison. &nbsp;Most stories about tubes breaking are from when the heater wire is bonded to it. &nbsp;Since you&apos;re not doing that, then there is no extra mechanical stress from the bonding.<br><br>Finally, be kind to the heater wire. &nbsp;It is running in a 1000 degree environment, so it only has about 200K headroom. &nbsp;So while you can put 1000W into it when starting at ambient, your controller should probably lower that limit once the temp rises. &nbsp;I designed a circuit to measure the wire&apos;s own temperature by periodically sampling its resistance compared to a fixed reference, so an additional SSR or logic could be used to prevent it burning out. &nbsp;You can do this because Kanthal A1 has a normal positive temp coefficient of resistance (about +5% at 1200 deg C, IIRC).

  3

• GU

  @guillaume8437 · 5 months ago

  Wooow! The explanation of the PID was even better (clearer) than what we can have at school. 😮<br>How much further could you push the temperature? Can you reach 1100°C, 1200°C?

• DM

  @dmitry-diehl · 1 year ago

  Wow, very useful! Thank you!

  2