May 10, 2011
I’ve spent the past few days trying to optimise the Prusa Mendel printer with respect to speed and temperature, having already calibrated steps/mm etc. The printer however, was having none of it and very kindly gave me wildly varying prints where the top surface rarely bridged properly. I noticed that the temp was varying quite a bit so I decided to change out the thermistor which had probaby come away from the heater block. I replaced it with a different model and decided to tape it to the block with kapton tape. However the fluctuations were worse (+/- 15 degrees C around setpoint!). The new thermistor, too, was struggling to stay in contact with the heater.
I had reinforced the previous thermistor/fire cement setup with a flat brass piece so i bent this to try to pin the new thermistor in place in the (now widened) hole. This helped but didn’t solve it completely. So then i stripped some light wire and wrapped it around the brass and the heater block to hold the brass tight against the thermistor and the heater block. So the temperature is now nice and stable! However I need to set it to 300C to get the flow i used to get at 210C….. must be a temptable problem… but at least its extruding repeatibly again!
Printing from SD using sprinter firmware works well – its nice not to have blobs everytime you do something that forces the netbook to work hard 😉
April 22, 2011
After almost 3 weeks of printing the parts using my Repstrap I’ve finally got a RepRap! 13 mm/s was the top speed of the repstrap, limited by the leadscrews…. the Prusa will be limited by the extruders ability to keep up but I’ve been printing at 30 mm/s for now as 50 mm/s was a little sparse and any faster and its difficult to see whats happening (I’m used to slow prints remember!) .
I’ve also printed out an Accessible Wade’s Extruder by GregFrost to replace the plywood extruder.
Here is the printer…
and the extruder…
So its back to some optimising…. happy days.
Recently I bought a micro SD card reader from Kliment (hes always in #reprap irc channel), so when I’ve the skeinforge settings under control I’ll upload his firmware and print from SD which will make life easier for my rusty old pc.
March 30, 2011
Just as I seemed to be getting good prints I started to get a lot of failed prints. Basically the extruder wasn’t able to take in filament if there was any tension holding it back so I had to babysit quite a lot and ensure there was slack filament. It seems that the springs I attached a few weeks back weren’t quite up to the job (I had robbed them from pens!). So I’ve gotten hold of some meatier ones. With the help of varying sized washers I’ve put them in place and compressed them a little…
This means that the filament is forced further onto the hobbed bolt which bites in more and has better grip. Hopefully this has solved the issue for once and for all as I’ve started printing out the Prusa parts.
The frame vertex on the left was done @ 7 mm/s, 0.25 layer height (with babysitting!). The other parts (with the spring mods) @ 13 mm/s, 0.25 layer height. 13 mm/s is a nice step forward but is the limit on the leadscrew stages and is still very time consuming!
March 18, 2011
I’ve ‘installed’ a heated print bed – its 4x 2.2 ohm resistors wired as [(2 in parallel) in series with (2 in parallel)] giving 2.2 ohms. This gives ~5.46 Amps and ~65 Watts @ 12 Volts which heats the bed to 55C (dibond underside) in a few minutes (I haven’t timed it!). Its driven from an old atx pc power supply with a max of 6 A @ 12 V and I’m using RAMPS to drive a relay to control it (my main atx gives 15 A @ 12 V but the printbed trips it out). Hopefully the relay will survive….
The first tests, some shamrock i designed for my children to paint for St. Patrick’s Day, came out pretty good:
Here is an image of some 20x20x10mm ‘cubes’:
The one on the left is 100% fill @ 6 mm/s, WoT of 1.8 and layer height of 0.25 mm. It seems to be well filled, possibly bordering on overdone. The one on the right is the same at 20% fill. The top surface is pretty good but the light scatter from layers below makes it look bit worse. The next thing to do is print a few different sized cubes to check the dimensions in case there’s backlash present. Then I an hopefully move onto printing Prusa parts….
Here’s the latest image of the repstrap contraption:
March 14, 2011
While trying to get things optimised I discovered a more annoying problem – lack of repeatability! Nice prints sometimes…. horrible at other times with same parameters. Tension in the spool seemed to have an effect – lots of slack gave nice prints but slight tension and there were lots of voids. The extruder was stripping the filament at these times also. Some of the folks on IRC recommended trying springs as per a standard Wade’s so I’ve modified the extruder to allow this at there seems to be a lot of variation in the filament I’m using. I’ve also re-hobbed the bolt while I had the chance.
Here’s an extreme example of the voids:
Normally in this Sliced Wade’s Geared Extruder there are two pieces that push on the idler but are then locked in position. I’ve spring loaded these instead of fixing them:
Back to optimisation, hopefully….
March 7, 2011
I printed my first test object a little over a week ago. I used skeinforge 39 to slice and repsnapper to send the code to the RAMPS setup. The print quality was poor so I started to figure out how to optimise it. It turned out that skeinforge 40 was released recently and is a little different to 39 – see RepRap forum here – so I decided that I may as well get used to it instead of ver 39.
First thing was to figure out E_steps_per_mm: to do this I nicked the filament at 50 mm intervals using a blade, took the extruder off, ran the extruder as far as required to go the 50 mm using Repsnapper and plugged the 50 mm value, repsnapper value and old firmware E_steps_per_mm into Josef Prusa’s
calculator calculator to get the new steps per mm. I followed all as per the forum link above for the setup (for material type, diameter etc).
To optimise I then looked at Dave Durant’s blog and started printing off cubes. For the moment things are running ok at a really slow 5mm/s. I’m going to increase the temp and see if that allows me to extrude faster and hence up the feed/flow rate and do things a bit quicker as I suspect the higher temp may allow me to introduce material into the nozzle and melt it faster.
Something that puzzled me for a while was the Perimeter Width over Thickness ratio (PWoT) and layer thickness (mostly the former!). My current understanding is that its basically the resolution at which you build your object.
So Layer Thickness (T) is the vertical resolution: the lower the thickness value the more layers you will have in your object. For example imagine 0.25 mm v.s. 10 mm layer thicknesses – the object with 10 mm thick layer will be much coarser than the fine detailed 0.25 mm one – like a 791×791 pixel image vs a fine 10M pixel photo (3162×3162 pixels).
So far so good. What about the PWoT? This is the same horizontally – how closely lines will be spaced across the surface. PWoT * T = PW or width. Increase PWoT to spread the lines further apart and thus lower the resolution, decrease for the opposite effect. Maybe they should be called vertical resolution and horizontal resolution and let the software calculate the ratio if it needs it but there are probably good reasons for being as it is. According to Dave Durant you need to set/fix these values (choose/decide your print resolution) and find a feed/flow rate to achieve good quality at the chosen level of detail.
February 24, 2011
In the past few days I’ve tested the hotend (pla is melting nicely from 200 degrees C upwards) and I’m happy enough with that. I’ve also installed and leveled the dibond bed – this post has good pointers to help. I’ve setup the opto-endstops and I have a build envelope of approx 195x135x100. I’ll post photos in a few days. At the end of all that today, I managed to print a small 10 mm cube (designed in 3 seconds in openscad):
This was printed onto kapton tape (heated bed not yet setup) and obviously theres plenty of fun optimisation to be done…. but its nice to finally print a solid object 8)
Printer settings were mostly Araspitfire’s with a feed rate of 5, flow rate of 10 and max z speed of 2 (all mm/s I believe) as I didn’t want to push things too hard just yet (i have heatsinks to bond to the pololu drivers for doing that!).