Wednesday, September 25, 2013

Friends don't let friends RUMBA

I know that I have been slow about updating this blog but I ran into several issues with printing that I have just recently been able to solve. I'll detail the problem here and my solution and hopefully it can help someone else.

Problem 1: Randomly crashing X axis
I was getting really bizarre motions where after starting a print that was going well, the X axis would just randomly start moving in much bigger motions than it should. For example, I notice that when printing a circle that instead of printing a 0.5 inch circle suddenly it will start printing a 6" circle. This frequently happened when I would start a print and so when it tries to find the center of the bed it will go off the +X axis and start grinding on the gears. 

Solution 1: Unshielded LCD cables were causing a problem
This seemed like an electronics issue so I replaced my USB cable and switched to printing from a SD card and it did not solve the problem. I was running a new RUMBA board so it was hard to find any support for this issue even in general much less for this specific board. I eventually found that removing the reprapdiscount LCD screen seemed to fix the issue. It is likely that it was interference with the unshielded cables even though mine were <30 cm.

Problem 2: Extruder randomly extruding 4x as much plastic
I thought that I was out of the woods with the X axis problem but then I ran into a new issue that I didn't have before. I would turn my printer on and start a print which would go fine. Often I'd want to tune a slic3r setting so I would stop the print and restart it several times. I noticed that sometimes my extruder would put out way more plastic than it should. At first I thought it was different slic3r settings but after doing some controls where I took the same gcode and printed it several times I discovered that the printer would just apparently squirt out different amounts of plastic with the same gcode. Here is an example where you can see one parameter on the left when it is thick and one parameter on the right which is what it should be doing.

This made me think that it was a steps/mm setting that was getting reset but I noticed that after unplugging the board for an hour or so that I could get the settings to reset and behave normally. This led me to believe that it was a flaw in my Rumba electronics themselves. 

I paid more attention to the extruder motion which gave me a clue. Here is the normal extruder extrusion rate:
video
After printing for a while I can do the same experiment and see that the extruder is moving 4x the speed as before:

video
This evidence makes it seems like there is some sort of switch such as on the microsteppers that is defective and is randomly switching between microstepping and normal stepping during the print. The hitch with that hypothesis is that I get the same behavior when I hook up the extruder to a different stepper on the board which has different switches. 

Solution: Purchase non defective RAMBO Electronics
After reburning the bootloader to no avail, and getting no where with reprapdiscount to actually replace the board, I gave up and purchased a set of RAMBO electronics instead. I wanted a way to directly test whether it was the RUMBA board or some other firmware issue that I couldn't think of. Boy was I impressed when I hooked it up. 

After printing a test cube that came out beautifully, check out my second print with the RAMBO electronics, the curve fan holder, with all the stock settings:
There was not a layer or boundary out of place on this print. I am just so impressed with the quality of the Mendel Max Setup now that I have the electronics working properly!


4 comments:

  1. This comment has been removed by the author.

    ReplyDelete
  2. (argh, browser dropped comments, retyping... and I forgot to check notify me...)
    I have no affiliation with rumba, but I switched to it and am pretty happy with mine. I do have a few nits, but they are different than your problems.

    I like your theory on the full/micro- steps, but when you say different stepper did you move positions on the board, *and* new pololu, or same pololu? I'm just guessing but I'd look for current problems on that pololu and/or thermal ones. heatsink? fan? (I have both). When you went to the RAMBO did you move the pololu's (same one for the E)? I did tune each pololu (clone) on my setup by listening to the steppers.

    I am curious to know the root cause. As I said I like mine, but do have a few nits.

    ReplyDelete
  3. I forgot to mention that I agree with your LCD comments. I have issolated mine (routed steppers, heaters etc away from LCD) and it has gotten better. I have also seen this problem with RAMPS. I'm not happy with the level shifter on that LCD, It drops to 3.3v, but there is nothing to go back to 5v. They rely on the CMOS threshold (2.5v being below 3.3v). This doesn't give much room for noise immunity and i suspect that is the real "short cables" issue, and that it only relates to the SD.

    I'm also bothered by the single ground connection between the smart LCD and the RUMBA. It is both the power return, and signal ground reference. I've measured +/- 500mV ground noise on the LCD board. it seems lots of these LCD's do this? You might also note that the reset/stop pin on the RUMBA is swapped. This means the LCD "stop" button is a "reset". Not what was intended, but I might like it better.

    I considered adding R/C termination to the 2 clock pins with a dedicated ground return, but decided in the end that that wasn't my dominant problem... Someday I may redo all that in an attempt to reduce system noise.

    ReplyDelete
  4. I know these posts are quite old in the grand scheme of things, but thought I'd post my experiences with Rumba electronics. I have been using a Sanguino setup on my Prusa Mendel for a couple of years now, with no real issues at all, until I plugged a new LCD into it that had a bad adapter board, and it fried my 1284P chip. I took the opportunity to look at new (possibly better) electronics, as I was in the process of building an oversized Kossel. The Rumba looked great on paper.... plenty of expansion for future (I wanted to try out a dual extruder setup). I have pretty much always run Repetier firmware without any issues, so I loaded it onto the Rumba and tried a few prints.
    Random Y-axis skipping steps. I tried EVERYTHING to fix it to no avail. My Y-axis is VERY smooth, if I lift the front of the Prusa up at a 30 degree angle, it would roll nicely to the rear. Changed printing speeds, right down to a crawl, and up as fast as the extruder could handle. Tested/adjusted driver currents. Rolled back to older versions of Repetier firmware.... nothing helped. Every 10th or so (not repeatable) layer it would skip, without me even hearing it skip steps, which was very weird.

    I gave up and bought a Ramps, and everything worked fine.

    Not wanting to let it get the better of me, I recently did some upgrades on extruder/x-axis setup, so decided to attack it again.

    I put some DRV8825 steppers in to see if that helped.... nope.

    But, something else appeared. The power to the heaters would now shut off without warning, with Repetier host saying that a heater had decoupled from the thermister (which wasn't happening). After some searching, I found a single post on a forum that lead me to believe it was a new feature in the firmware that did not work. I changed back to the previous version, which fixed that problem. In all of this, I hacked my power supply voltage/currant protection, bought a brand new power supply, current clamp meter, fuses to replace the thermo fuses on the Rumba etc... wasting a lot of time and money, but that's neither here nor there.

    Finally, I downloaded the latest Merlin firmware and the mysterious skipping disappeared completely.

    BUT... now I have another issue... with Marlin. My bed will easily stay at 115 degrees all day. When I start a print, with the bed at 115, it starts to drop and eventually levels out at about 98 after 10mins. It then gradually starts to rise until usually getting back to 115 near the end of the print (about 35-40mins). My power supply is more than adequate at 50A and I wasn't having this problem with Repetier firmware.

    I'm going to swap back to the Ramps because it is just so frustrating, but I REALLY want to get the Rumba to do what it's supposed to do.

    So, yeah, from my experience, don't let your friends Rumba.

    ReplyDelete