It’s immediately ambiguous as to whether that spec means 10 megabits per second or 10 megabytes per second.
For the first two weeks after receiving my H2D printer, I didn’t connect a flashdrive to my H2D. My prints were fine. I wasn’t aware of the minimum requirement, and later I installed a usb 2.0 drive that, it turns out, was below the minimum. The result of my error was that my PETG prints became mysteriously stringy with boogers. After I tracked down my mistake, I inserted a USB 3.2 drive, and the stringy boogers vanished. Back to good prints again.
The Bambu wicki didn’t recommend any particular usb flash drives by name. Be aware that performance can vary by a lot, and not only due to difference in IEEE standards. For example, some also have a narrower operating temperature range, which means they may throttle the data rate below the minimum if they get too hot. For that reason, I think I’ll probably connect mine using a short USB cable, so that it’s not in direct contact with potentially 65C metal (if, for example, the H2D’s chamber heater were on maximum). Maybe it was the potential for undesirable thermal throttling that Bambu opted for an external usb flash drive rather than a microsd card, like the X1C uses. Who knows? Bambu doesn’t usually explain its rationale for anything in much depth, so maybe it’s just coincidence.
I have one in each machine and can confirm no issues. Appreciate OP for raising awareness on this item; I didn’t realize there was a minimum. I just surfed Bezo’s best and ordered the first item that emcame next day
I used a PNY Elite Turbo Attache 4 USB 3.2 Flash Drive to diagnose and troubleshoot the problem, but only because it was laying around. For a more permanent solution, I would look for something known to run cool, and preferably with a metal case (to facilitate heat dissipation) and a lot of surface area (again, to help with radiating heat) that I can connect with a short USB cable. For those reasons, maybe something like:
You might be tempted to go for something compact and stubby so that you don’t accidently snap it off by careless impact, but in my experience those tend to run very hot, even when not plugged into a 65C printer.
Alternatively, vibration might also be a factor, so possibly something rated for that. Maybe something like:
I don’t yet have a handle on how much storage space would be a comfortable number. Perhaps a false impression, but my initial impression, though, is that the H2D is consuming a lot more flash storage than the X1C does. Maybe becaue the camera is better, so the timelapse adds up faster?
You could go for something industrial, like swissbit, which has a wide operating temperature range and is noted for high reliability, but, meh, expensive and probably overkill.
Those are my top-of-mind thoughts at this moment. I welcome any further ideas or suggestions.
I don’t want to argue if this is what you experienced but I still wonder, the key is only to store logs and timelapses. You can also store the models you send for printing but for the actual printing, the gcode is read from the internal memory of the printer not from the stick.
As far as stringing is concerned, I am currently playing a lot with the filament profiles in an attempt to replicate the ironing quality of what I can achieve on my P1S to no avail… but in the course of my testings I found out that I needed to lower the hotend temperature to at least approach my goal and for example hotend temperature of my PLA Matte filament had to be reduced from 220C to 200C and as a side effect of course, no more stringing whatsoever which I had before. It sounds like the bed temp was not accurate (too low) and the hotend temperature may also be inaccurate (too high) but unfortunately I have no way of measuring it accurately.
Actually, credit goes to @krellboy 's comment on a different thread entirely about the side effects on printing from slow microsd cards:
that gave me the idea to look into whether slow USB sticks on the H2D might result in something similar and thereby explain the mysterious print defects I was having at the time.
I understand but P1 does not have internal memory, the gcode is read from the SD card. If you remove the SD card while printing on a P1, print will stop and printer will scream SD card is removed. I did not check but I’m fairly sure you can remove the USB key while printinhg on the H2D and nothing happens (may be a notification that USB is removed and timelapse cannot be recorded).
Also, I’d be very careful with connecting any high-capacity and high-power devices, like portable SSDs or USB3.2 Gen2x2 devices - majority of USB devices are not really capable of providing enough power, and to be honest, printer manufacturer is probably the last manufacturer in line who would be capable of a proper BC1.2 or PD spec implementation.
The USB drive needs a minimum spec for constant writing speed as its saving the video/time-lapse on it, 10MB/s is the write speed required by bambu. You need to be careful as a lot of USB flash drives will drop in speed significantly as they warm up. Good idea to stick to brand name fast drives like Samsung or Sandisk. I picked up the Samsung FIT as its a small drive with fast speeds from a company that manufactures its own NAND chips.
It does cache the print files to it if you choose that option, but the read speed is pretty irrelevant for that unless you have some huge complex g-code.
Even in this case, I think it transfers to internal memory prior to re-print. If you access files in internal memory from the screen menu you’ll notice that the last print is always stored there and gets overwritten everytime you send a new file to print.
This is interesting, though wondering why the USB would affect print quality.
I have not done repeated trials but i noticed this localized layer shifting on the left side of my helmet print, but this shift doesn’t appear in the supports nor the right side of the helmet.
If it were a lead screw issue id see it across all those layers
I was wondering why my usb flash drive was filling up so quickly, so I ejected it and had a look. It turns out that the timelapse video and the cached models barely amount to anything. Instead, by comparison, the ip camera files are numerous, and huge:
You can see from the second screen clipping that it is saving 1080 video at 30 frames a second, and this comes out to just under 8mbps. So, this explains that 8mbps minimum write-speed requirement.
The cool thing is that you can play back these files, which on some occasions may be far more useful in pinpointing exactly what happened to cause a failure than a mere timelapse video. I had thought that capability would be a nice feature enhancement, but it turns out we already have that capability!
Matter of fact, it would be kinda cool to record video continuously, but it seems to time-out “due to inactivity” after awhile, forcing me to reconnect. Maybe there’s some way around that limitation for those of us who might want this capability in a non-interruptible manner?
Maybe there’s some way around that limitation for those of us who might want this capability in a non-interruptible manner? 