Thanks @Jon_Charnas for the recommendation for another camera. If the camera I have now dies, I now have a better option available. It’s also higher resolution. It’s available in both fixed and auto focus. And it’s not expensive.
The writing comes off the build plate if you print on it, so if you are going to print something large where the exposed side is the bottom layer, keep this in mind:
This piece actually shows two problems. The writing is the most obvious. But the other is that you can see some gaps in the lines. This is because it’s at the edge of the bed and the bed mesh probing can’t reach, the bed isn’t as flat as it could be, and so it’s just not as accurate a map of the bed height as it might be. It’s close enough that the prints aren’t failing, but there are print artifacts near the edges.
a pinch more extrusion on the first layer might fix that.
No, the extrusion is right except in a few places. This was one of 29 identical pieces on the bed, and most of the are just right. This is a bed flatness mapping failure from the sensor not being able to reach the whole bed.
1 Like
I just had two adhesion failures in a row trying to print on the unprinted side of the bed plate to avoid transferring the image to the printed objects. The same filament is fine on the printed side of the bed plate. I’ve asked Qidi about this.
I will note that Qidi support have continued to be pretty good at prompt replies. Of course we’re on opposite sides of the world, more or less, so there is usually a day of turnaround.
The main problem has been answering the wrong question. There’s been some “lost in translation” at times. For example, I told them I’d successfully set up a web cam, but wanted to switch to using HLS, which is not showing up as an option in the fluidd that they ship. I asked about updates that would add HLS video streaming (MJPEG Adaptive is giving me only 2-3 fps and HLS video streaming is likely to work better), and the help they tried to offer was for web cam not working at all and was not related to my question. But that’s been the exception, mostly they have been directly helpful.
I see that a new firmware version has been released. I had noticed that shutting down from the attached screen didn’t seem to do anything. From diff -uNrw QIDI_MAX3-Max3_V4.3.6 QIDI_MAX3-Max3_V4.3.8 I see this:
diff -uNrw QIDI_MAX3-Max3_V4.3.6/src/mks_gpio.cpp QIDI_MAX3-Max3_V4.3.8/src/mks_gpio.cpp
--- QIDI_MAX3-Max3_V4.3.6/src/mks_gpio.cpp 2023-08-04 21:50:22.000000000 -0400
+++ QIDI_MAX3-Max3_V4.3.8/src/mks_gpio.cpp 2023-09-02 04:44:07.000000000 -0400
@@ -200,10 +200,11 @@
if ((val - '0') == 0) {
// 检测到低电平之后要执行的
// set_GPIO1_C5_low();
+ //4.3.7 CLL 关机时自动保存
//4.3.5 CLL 屏蔽关机页面
- //system("echo TEST > /root/TESTGPIO; sync");
+ system("echo TEST > /root/TESTGPIO; sync");
//go_to_page_power_off();
- //system("sync; shutdown -h now;");
+ system("sync; shutdown -h now;");
}
}
usleep(110000); // 检测电平减少一点
Looks like a test leaked into production, but I should be able to shut it down from the screen now without pulling out my computer or my phone to run the shutdown from fluidd. 
3 Likes
I’m glad that they are doing frequent firmware updates with fixes and new features. The most recent one, 4.3.8, released after I got the printer — so I’ve already applied two updates in the brief time I’ve had the printer — has two new features (shows data including the thumbnail properly on network upload and print; factory reset) and a bug fix (shutdown from the front panel should work now; see previous post showing that it was commented out before).
They do overwrite the WiFi information on every update. So I’ve had to enter my wifi information three times now. (It’s almost enough for me to switch to wired ethernet! I do have I think four ethernet jacks in this room, so it’s not like I couldn’t do that. )
I logged in, ran cp /etc/wpa_supplicant/wpa_supplicant-wlan0.conf /root/wpa_supplicant-wlan0.conf, and created /etc/systemd/system/restorewifi.service with the contents:
[Unit]
Description=Restore WiFi configuration if removed by update
Before=network-pre.target
[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/bin/grep -q 'psk=' /etc/wpa_supplicant/wpa_supplicant-wlan0.conf || test -f /root/wpa_supplicant-wlan0.conf && cp /root/wpa_supplicant-wlan0.conf /etc/wpa_supplicant/
[Install]
WantedBy=multi-user.target
Then systemctl daemon-reload && systemctl enable --now restorewifi and we’ll see whether the wifi information survives the next firmware update. 
1 Like
I have also determined that updates overwrite my fluidd configuration and my webcam configuration.
I’m going to have to do something like back up files to a directory on another computer on which I run git so that I can see what changed from version to version, or something like that… 
2 Likes
With the web camera in the back, the LEDs shine harshly and make it hard to see the part.
In that picture, I’m printing a diffuser to clip over the LED strip to try to avoid the problem that the image shows. And it’s much worse if I turn off the room light. But the reason I care about the LEDs is specifically to turn off the room light and still be able to monitor a print.
I did some rough measurements and printed out a sample of the shape I came up with:
The thick part at the left holds onto the bar that the lights are mounted to, and the thin part diffuses the LEDs. The thin part pulled away from the bed, so for the whole thing I’m using a brim.
Even with the brim, it’s trying to pull away a little bit. There’s no good angle to take this picture, but it is kind of visible how there is a lighter area around the tip where it is starting to pull away.
Also, the filament tube brushed the outside of the clip. I also added as many rounded corners as I could get away with to improve print speed.
I couldn’t print all 334mm of it in one go in this printer since it is limited to 315mm in Z (first time I’ve run into the vertical limitation in this printer), so I’m printing it in two halves, back to back:
Even though the printer ships with a 0.4mm nozzle, the slicer profiles don’t offer layers thicker than 0.25. I had to modify that to get 0.3mm and speed up the print.
It fits very securely. In fact, it was hard to remove one of the pieces to reposition it while I was installing it. But also, diffusing the light doesn’t have the impact I was hoping for on the webcam.
Maybe I should have made the thin part thicker to attenuate the light a little bit more, or maybe it’s hopeless and I should take the printer apart a bit to find out how to switch to LEDs in the back instead. (Not doing that for now, anyway.)
QidiLightDiffuser.FCStd (19.0 KB)
1 Like
The temperature sensor, like on almost all printers, is on the bottom, heated side of the bed, not on the top surface that you print on. When the bed first hits the target temperature at the point where it measures the temperature, the top of the bed is colder and still soaking up heat, and not yet ready to print.
To know when the heated bed is heated through, I can watch the power going to the “Heater bed” in the fluidd interface. It will read about 85% power when it reaches the target temperature, then power will go down as the bed heats through.
When the power being used goes below about 50%, then the bed temperature is relatively stable and I can start the print cycle.
When I’m heating the bed to 100°C in a 50°C heated chamber to print ABS, it seems to hit equilibrium at about 45% power, typically fluctuating mostly between 43% and 47% that I’ve seen (I wish I had a graph of power though).
1 Like
This printer is accurate enough to print fine threads.
I printed an object in ABS with vertically-oriented M3 threaded holes using the finest default setting (0.12mm layers), and the print was accurate enough that I was easily able to drive a M3x12 screws in to full engagement. They are not loose in the holes, but they didn’t bind and they weren’t hard to drive. I didn’t have to add any clearance to the thread form, and I didn’t have to chase the threads with a tap. I don’t think I’ve ever done that before with M3 threads.
When I start a screw into a formed thread, I practically always turn it backwards until it clicks, then drive it forwards, in order to avoid cross-threading. I did that with these screws and could feel the click as they crossed the beginning of the thread form. It wasn’t as positive a click as a tapped hole in metal, but it was there.
3 Likes
The LED light is either on or off, and on is too bright for the webcam. On my other printer running klipper, I have hardware PWM set and can dim the LED. I found the section in printer.cfg:
[output_pin caselight]
## Chamber Lighting - In 5V-RGB Position
pin: PC7
pwm: false
shutdown_value:0
value:1
#cycle_time:0.01
Support responded promptly to my question with:
Maybe after you modify it, it will conflict with the startup lights.
You can try modifying it.
Given that encouragement, I tried using hardware_pwm. I used fluidd to edit the printer.cfg file and restart klipper:
[output_pin caselight]
## Chamber Lighting - In 5V-RGB Position
pin: PC7
pwm: True
shutdown_value:0
value:0.3
cycle_time:0.01
hardware_pwm: True
That didn’t turn on at all. I’m not sure why; I found the datasheet for the mcu CPU, and if I interpret it right, hardware PWM ought to be available using timer 3, channel 2, and I don’t think that is shared with any other ports. (The adjacent PC8 controls the bed and is on timer 3, channel 3, and is PWM-controlled.)
stm32f401rc.pdf (2.1 MB)
But for now I’m trying out software PWM, which seems to work so far, though I haven’t tested whether it causes problems for printing yet.
[output_pin caselight]
## Chamber Lighting - In 5V-RGB Position
pin: PC7
pwm: True
shutdown_value:0
value:0.3
cycle_time:0.01
Putting that in place gives me a slider to set the brightness in fluidd. This way I can set it so that it doesn’t overwhelm the webcam’s auto exposure adjustment.
2 Likes
It is worth noting that for some reason, the nozzles on the Qidi aren’t quite the same length as a Volcano nozzle. The Qidi nozzles are an unusual 19mm long, whereas Volcano nozzles are 21mm long. They appear otherwise the same.
I think there is enough room in the bltouch that you could use a standard volcano nozzle in the Qidi if you adjust the Z offset, though I would suggest changing the offset to give at least 2mm distance before changing the nozzle to avoid gouging the bed.
I bought a bunch of extra Qidi nozzles instead, though for most nozzles (maybe not hardened steel) I could choose to shorten them 2mm on my lathe; they both use the standard M6x1 thread form.
1 Like
Qidi support kindly provided me with the mainboard pin mappings along with responding to my request for help with PWM dimming of the case light.
X-6 V1.0_001 PIN.pdf (727.6 KB)
This means I could just disconnect that strip and put my own strip at the back. I might do that and make a new webcam holder from scratch that moves the webcam further up; if I do I’ll share the design.
So far, using soft pwm with a cycle time of 0.01 has not shown up in print problems as far as I have notice, nor in the MPU load graph in fluidd.
I wonder whether engineers being concerned about conflict was due to 0.01 cycle time and 50Hz electricity? Not seeing obvious flickering with 60Hz power here, anyway. And anyone on 50Hz could choose a different cycle time. 
3 Likes
Qidi are finally rectifying the puzzling omission of an integrated webcam by at least offering it as an add-on. Not sure why it wasn’t integrated in the first place given the price, but should be more convenient than the separate camera I bought and spent hours getting working…
4 Likes
I previously doubted that they would raise the preorder $999 price to the $1399 list price; indeed, I wouldn’t recommend it at $1399. I didn’t expect that they would drop the price immediately after general release. But it’s now listed at $949, and the camera for $40, so now it’s under $1K including the camera.
The camera alone didn’t qualify for “free” shipping, but I had already been thinking about buying some of their 0.6mm copper nozzles for even faster printing for at least some prints and that bumped me over the line. (This is amusing since the camera is a pre-order and will still ship separately.) I’ve already bought an extra hot end assembly that I can repurpose to be my 0.6mm assembly.
3 Likes
If nothing else comes out of the current state of the printer market, I hope that Prusa start innovating again on both features and community. I realize that it’s hard, frustrating work but I fear that they’ll start a death spiral if they aren’t re-energized soon.
2 Likes
Competition is the best motivation.
3 Likes