Dankdryer improvements

(7 intermediate revisions by the same user not shown)

Line 1:

In an [[Dankdryer|earlier article]], I designed and constructed a high-temperature [[Filaments|filament]] dryer. Before I was even done putting together the first design, I was thinking of improvements for reliability, efficiency, cost, and ease of assembly. I've put some of them into effect, and the results are most pleasing.

[[File:Dankdryer.png|thumb|right|The Dankdryer PCB]]

In an [[Dankdryer|earlier article]], I designed and constructed a high-temperature [[Filaments|filament]] dryer. Before I was even done putting together the first design, I was thinking of improvements for reliability, efficiency, cost, and ease of assembly. I've put some of them into effect, and the results are most pleasing. Most importantly, I designed a PCB to replace the perfboard and through-hole mess of the original. Like the rest of the project, the KiCad schematics and PCB design are open source.

==Problems with v1==

Line 8:

Line 9:

==Models==

* I raised the air shield to cover the entirety of the load cell.

* Cut the bottom chamber down mercilessly to save mass, aiming to get the sum of both chambers below 500g using my recommended settings. Haven't reached it yet, but we're close.

* Cut most of the midsection of the bottom chamber wire channel away. It's easier to work with if you can see/touch the wires in the middle of the channel.

* Completely changed the motor mount. It's now vertical, and sits atop the load cell. I didn't want its weight in the load cell's measurement, but it wasn't at all worth the added complexity of an orthogonal motor (worm gear, etc.). Things work **much** better now -- you can put a level atop the spool, and watch it rotate serenely for hours. Eliminated the air shield; it didn't prove necessary, and significantly increased print time.

* Cut the bottom chamber down mercilessly to save mass, aiming to get the sum of both chambers below 500g using my recommended settings. Haven't reached it yet, but we're close. ~~Most noticeably, I~~ changed the motor mount ~~from a rectangular to a trapezoidal prism~~, and ~~then cut~~ an ~~inverted trapezoidal prism out from its center~~.

* Added a hole for an RP-SMA antenna, and a hole for a NEMA plug + rocker switch.

* ~~I separated~~ the ~~bottom of~~ the ~~control chamber from its walls~~, and ~~added corners~~ to the ~~bottom of the latter so the two can be fastened with screws. This ought (1) eliminate warping when printing with polycarbonate~~ and ~~(2) allow more rapid iteration on~~ the ~~control~~ chamber, ~~since the bottom represented significant printing time, yet stayed pretty much constant throughout~~.

* Added helical forms to the top and the hot chamber, so they screw together.

==Electronics==

* Let's toss the TB6612FNG motor controller. We only need one direction of rotation, so we control the motor with an ~~[https://www.sparkfun.com/datasheets/Components/General/RFP30N06LE.pdf RFP30N06LE]~~ N-channel MOSFET, a [https://www.vishay.com/docs/88525/1n5817.pdf 1N5817] Schottky diode~~, and a 10K resistor~~. The n-FET goes on the ground side of the motor, and the diode runs in parallel with the load (i.e. is connected to the motor's two pins). The resistor pulls down the gate lead. This eliminates four net wires (AIN1, AIN2, STBY, AO2, APWM go away; we add GATE) while adding ~~one resistor, a net reduction of seven joints. I'm not certain that it's actually any cheaper, though; it might actually be more expensive.~~

* I added a Hall sensor to the roof, and with a magnet on the spool platform, we can now detect spool rotations. More importantly, we can detect a failure to rotate, probably indicative of obstruction or even melting. This is important, because the spool will otherwise heat asymmetrically.

** It's not easy finding a good throughhole logic level MOSFET for 3.3v (surface mount are readily available)! The RFP30N06LE will only pass about 20% of its rated amperage at 3.3v...but that's more than enough for our needs (the TB6612FNG could only sustain 1.2A, after all). I already had the necessary MOSFET thanks to this [https://www.amazon.com/EEEEE-Transistor-Assortment-Channel-RFP30N06LE cute little set].

* Let's toss the TB6612FNG motor controller. We only need one direction of rotation, so we control the motor with an N-channel MOSFET, a 330 and 1.62K resistor, and a [https://www.vishay.com/docs/88525/1n5817.pdf 1N5817] Schottky diode. The n-FET goes on the ground side of the motor, and the diode runs in parallel with the load (i.e. is connected to the motor's two pins). The resistor pulls down the gate lead. This eliminates four net wires (AIN1, AIN2, STBY, AO2, APWM go away; we add GATE) while adding two resistors.

* The LM35 is operating in a nasty thermal environment, and on about ten centimeters of poorly-shielded AWG22. Let's give it some bigger boxing gloves. Put a 0.01µF bypass capacitor across its power and ground leads. Put a 75Ω resistor and a 0.22µF capacitor in series between the signal and the ground. These recommendations come directly from the [https://www.ti.com/lit/ds/symlink/lm35.pdf datasheet]:

[[File:LM35-circuit.png]]

* We should use the AVDD analog power input of the HX711 at 5V.

===PCB considerations===

* We're ditching the Sparkfun NAU7802 breakout board, so we'll need add 2.2kΩ pullups on both I2C lines, 0.1µF and 1µF capacitors on VDDA, and 47Ω resistors on VNN1N and VNN1P with a 0.1µF bypass capacitor across them. See the [https://cdn.sparkfun.com/assets/6/a/5/9/d/Qwiic_Scale.pdf schematic]:

Line 29:

* I experimented with adding a kinda inverted eggholder (actually just a bunch of inverted frustra) to the underside of the bottom chamber, in the hope of eliminating warping. It didn't really pan out (a good number detached from the build plate), and you need defeat warping for the top chamber anyway, so I removed them.

* I'd love to get rid of the internal AC, but the heater maxes out at 4.5A of 120VAC. That's 45A of 12VDC, and now you're talking a (fairly large) power supply running close to $30.

* Maybe we should replace the hot chamber thermometer with a K-type thermocouple and the necessary PCB electronics/connector? This looks quite expensive; if we don't need it, let's not.

Dankdryer improvements: Difference between revisions

Dank (talk | contribs)

@@ Line 1: / Line 1: @@
-In an [[Dankdryer|earlier article]], I designed and constructed a high-temperature [[Filaments|filament]] dryer. Before I was even done putting together the first design, I was thinking of improvements for reliability, efficiency, cost, and ease of assembly. I've put some of them into effect, and the results are most pleasing.
+[[File:Dankdryer.png|thumb|right|The Dankdryer PCB]]
+In an [[Dankdryer|earlier article]], I designed and constructed a high-temperature [[Filaments|filament]] dryer. Before I was even done putting together the first design, I was thinking of improvements for reliability, efficiency, cost, and ease of assembly. I've put some of them into effect, and the results are most pleasing. Most importantly, I designed a PCB to replace the perfboard and through-hole mess of the original. Like the rest of the project, the KiCad schematics and PCB design are open source.
 ==Problems with v1==
@@ Line 8: / Line 9: @@
 ==Models==
-* I raised the air shield to cover the entirety of the load cell.
+* Cut the bottom chamber down mercilessly to save mass, aiming to get the sum of both chambers below 500g using my recommended settings. Haven't reached it yet, but we're close.
-* Cut most of the midsection of the bottom chamber wire channel away. It's easier to work with if you can see/touch the wires in the middle of the channel.
+* Completely changed the motor mount. It's now vertical, and sits atop the load cell. I didn't want its weight in the load cell's measurement, but it wasn't at all worth the added complexity of an orthogonal motor (worm gear, etc.). Things work **much** better now -- you can put a level atop the spool, and watch it rotate serenely for hours. Eliminated the air shield; it didn't prove necessary, and significantly increased print time.
-* Cut the bottom chamber down mercilessly to save mass, aiming to get the sum of both chambers below 500g using my recommended settings. Haven't reached it yet, but we're close. Most noticeably, I changed the motor mount from a rectangular to a trapezoidal prism, and then cut an inverted trapezoidal prism out from its center.
+* Added a hole for an RP-SMA antenna, and a hole for a NEMA plug + rocker switch.
-* I separated the bottom of the control chamber from its walls, and added corners to the bottom of the latter so the two can be fastened with screws. This ought (1) eliminate warping when printing with polycarbonate and (2) allow more rapid iteration on the control chamber, since the bottom represented significant printing time, yet stayed pretty much constant throughout.
+* Added helical forms to the top and the hot chamber, so they screw together.
 ==Electronics==
-* Let's toss the TB6612FNG motor controller. We only need one direction of rotation, so we control the motor with an [https://www.sparkfun.com/datasheets/Components/General/RFP30N06LE.pdf RFP30N06LE] N-channel MOSFET, a [https://www.vishay.com/docs/88525/1n5817.pdf 1N5817] Schottky diode, and a 10K resistor. The n-FET goes on the ground side of the motor, and the diode runs in parallel with the load (i.e. is connected to the motor's two pins). The resistor pulls down the gate lead. This eliminates four net wires (AIN1, AIN2, STBY, AO2, APWM go away; we add GATE) while adding one resistor, a net reduction of seven joints. I'm not certain that it's actually any cheaper, though; it might actually be more expensive.
+* I added a Hall sensor to the roof, and with a magnet on the spool platform, we can now detect spool rotations. More importantly, we can detect a failure to rotate, probably indicative of obstruction or even melting. This is important, because the spool will otherwise heat asymmetrically.
-** It's not easy finding a good throughhole logic level MOSFET for 3.3v (surface mount are readily available)! The RFP30N06LE will only pass about 20% of its rated amperage at 3.3v...but that's more than enough for our needs (the TB6612FNG could only sustain 1.2A, after all). I already had the necessary MOSFET thanks to this [https://www.amazon.com/EEEEE-Transistor-Assortment-Channel-RFP30N06LE cute little set].
+* Let's toss the TB6612FNG motor controller. We only need one direction of rotation, so we control the motor with an N-channel MOSFET, a 330 and 1.62K resistor, and a [https://www.vishay.com/docs/88525/1n5817.pdf 1N5817] Schottky diode. The n-FET goes on the ground side of the motor, and the diode runs in parallel with the load (i.e. is connected to the motor's two pins). The resistor pulls down the gate lead. This eliminates four net wires (AIN1, AIN2, STBY, AO2, APWM go away; we add GATE) while adding two resistors.
 * The LM35 is operating in a nasty thermal environment, and on about ten centimeters of poorly-shielded AWG22. Let's give it some bigger boxing gloves. Put a 0.01µF bypass capacitor across its power and ground leads. Put a 75Ω resistor and a 0.22µF capacitor in series between the signal and the ground. These recommendations come directly from the [https://www.ti.com/lit/ds/symlink/lm35.pdf datasheet]:
 [[File:LM35-circuit.png]]
-* We should use the AVDD analog power input of the HX711 at 5V.
 ===PCB considerations===
 * We're ditching the Sparkfun NAU7802 breakout board, so we'll need add 2.2kΩ pullups on both I2C lines, 0.1µF and 1µF capacitors on VDDA, and 47Ω resistors on VNN1N and VNN1P with a 0.1µF bypass capacitor across them. See the [https://cdn.sparkfun.com/assets/6/a/5/9/d/Qwiic_Scale.pdf schematic]:
@@ Line 29: / Line 29: @@
 * I experimented with adding a kinda inverted eggholder (actually just a bunch of inverted frustra) to the underside of the bottom chamber, in the hope of eliminating warping. It didn't really pan out (a good number detached from the build plate), and you need defeat warping for the top chamber anyway, so I removed them.
 * I'd love to get rid of the internal AC, but the heater maxes out at 4.5A of 120VAC. That's 45A of 12VDC, and now you're talking a (fairly large) power supply running close to $30.
+* Maybe we should replace the hot chamber thermometer with a K-type thermocouple and the necessary PCB electronics/connector? This looks quite expensive; if we don't need it, let's not.