I changed the model material adjustment to a lower value than I orginally had, and ran by modified Quickslice calbox. I stopped the program before the support layer, and measured the part. The road thickness was 0.031″ — exactly what I raised the values to achieve the other day. The height of the 10 layers was 0.103″ — pretty darn close to what it should be.
Confused, I remeasured my old calboxes, and the new. They all measured what I originally measured them.
I ran another calbox to see if I had a fluke part. The second was the same as the first.
I decided to run the small part I’ve been running and see how it did:
The part had a much improved top finish, and none of the hair/blobs on it. The part thickness measured 0.255″ to 0.257″ depending on where I measured — a lot closer to the 0.250″ thickness of the part. I could see gaps between the Y roads & fill (most noticable by the cutout) — I’ll look at it closer tomorrow.
The Z Axis DRO, still tacked into position, continued to work okay, and was helpful in setting the initial Z position.
I filed out the previous parts I made for the Z Axis DRO, as I wasn’t making any progress making new ones that were better. I fitted it in the chamber & I tacked it into place with some modeling clay. I just ran the cord out the door to the readout to see if it worked okay. I ran it up and down and everything seemed to work okay.
I thought I’d do a few quick runs to try it out. I started the program for the small QuickSlice part that I have been using as a test piece. At the first ‘pause‘ I used a 0.08 mm feeler stock between the tip & foam as a zero point for the DRO. The table went down a bit (I didn’t catch the DRO reading), the alt-tip clicked in, and the program dragged the support tip accross the foam, scoring it (maybe 0.030″ deep). The support built up in the foam, and was about level with foam when the model layer started. I was amazed the 1st model layer builds at the same Z level as the last support layer (according to the DRO). I stopped this part after the 1st model layer as I wanted to see the gap between the roads & fill, especially on the Y+. Here’s a photo, and the top Y+ gap is plainly visible.
This was really embedded into the foam! I dug it out, and started another run. This time I started 0.03″ above my previous Z Zero at the 1st ‘pause‘. The support tip didn’t drag this time, and I stopped the program after two model layers had built of the support. A photo of that part is below:
The loose strand of model material on the first layer, was getting tacked on by the second model layer. I decided to run a full part & call it a day — I started as I did previously, 0.03″ above my Z Zero. The DRO showed the table incrementing up 0.010″ each layer nicely. There were 25 model layers, and I should have a 0.250″ thick part. By the time the build got to the top, the surface was getting really ruff. The table would go down 0.020″ to move to another area, and the tip would drag accross the part. That’s probably why my Thursday part got knocked loose. Here’s a photo of the result:
The part was way over the 0.250″ thickness, and also thicker than parts I made before the Thursday Adjustments — It was 0.290″ thick!
A change in plans allowed me to spend some time with the FDM today. The small Insight part was done, and I took some quick photo’s:
There’s a spike on the top surface toward the front just to the left of the slot. This was also on the original Insight part I did last week.
I didn’t take time to measure the part. I wanted to try and get my Z part thickness correct. I started by fixing the ‘calibration box’ with the 5 extra layers. I inserted the missing “MA1007,1108; # Exit” lines. Here’s a link to the fixed program:
I ran a calibration part, and got out the calipers & manual. The calbox sides for the model material were supposed to be 0.8 mm (0.0315″) and the support 0.51 mm (0.020) according to the manual. The support was to be centered on the model material. I took notes and saved the parts, but left them on the FDM so I can’t get in to details right now. I did get the proper dimensions entering the corrections using the FDM keypad (not quite what the manual had as they left out several [enter] instructions, and the value entered wasn’t incremental). The support is close to outside of all the edges, and on the front edge it falls off the model material. I didn’t futz with that adjustment at all, as it looks like a scale problem more that a X-Y location problem. There is nothing in my FDM 1650 manual about checking the layer thickness of the calbox, or how to adjust it, so I wondered if my Z part thickness would change at all. I need to check some of the other FDM manuals as I saw it mentioned somewhere.
With these values set to aim, I sent the QuickSlice sml file for the small test piece — again, I wanted to see if the part thickness was closer to the 0.250″ part thickness. I watched the support layers, and several model layers after that. Everything running good, I went to do some other stuff while the part modelled up. When I returned:
So, no Joy or measurements today. I think I’ll take a 4 day MLK holiday, and wait till next week to start again.
I was a little worried about having the print heads up to temperature while cooking the material, support, and desiccant for an extended period of time. I unplugged and removed the print head. The machine would not power on without the print head connected. So I put it back on and started cooking.
I was surprised how much material oozed out while it was cooking with the filament completely removed from the print head.
While the material and support were cooking in the chamber, I opened up the back chamber cover, and reattached the hose support spring to a different screw on the X-Y Stage. Originally it was on the bottom terminal (with two blue wires) screw, but when I changed stages, it didn’t hold and broke loose. I recovered the screw from the bottom of the pan, but the nut is loose in the X-Y stage. I hope it doesn’t get caught up in something.
As this was cooking, I was thinking that I have six 4 ounce bags of desiccant & if I left three in the chamber bottom & three in the material box, I could swap them every once in awhile to avoid this.
After 3 1/2 hours of cooking (plus 1 1/2 hours the other day), I was anxious to see if the cooking was working.
I performed an ooze test, and it was very much improved. It looked to be ~15 to 20 mm, but I dropped the small filament before I could get an exact measurement. It’s is still above the acceptable ooze but not nearly as bad as it was when I started.
I had modified the ‘calbox’ sml program, to print 5 more layers (see http://www.3d4u.org/MyFDM_Files/cb12abs-5more.sml ). I ran that to see how it looked. It did print the 5 additional layers but the last 5 I added didn’t have the little X- flick to end the Z level path. Instead it slowly raised up, and started the next path. Apparently when I copy-and-pasted, I lost the ‘MA1007,1108; # Exit‘ line at the end of each Z level — rrr…..
I ran the same small Quickslice part that I ran several days ago. The X modelled to 0.926″ was 0.903″ X Bottom (0.006″ bigger than 1st Quickslice part), to 0.921″ X Top (0.002″ bigger than 1st Quickslice part); The Y modelled to 1.546″ was 1.517″ Y Bottom (0.007″ bigger than 1st Quickslice part), to 1.543″ Y Top (0.003″ bigger than 1st Quickslice part), The 0.250″ part thickness measured 0.272″ (0.002″ bigger than 1st Quickslice part). The slot modelled to 0.176″ in X measured 0.156 (0.001″ bigger than 1st Quickslice part), and modelled to 0.796″ in Y measured 0.764″ (0.014″ bigger than 1st Quickslice part). The outside X modelled to 0.926″ measured 0.921″ (0.001″ bigger than 1st Quickslice part), and the Y modelled to 1.546″ measured 1.543″ (0.004″ bigger than 1st Quickslice part). The front & left edges had a slight taper (angle) to it, and the right edge had a smaller taper. The rear edge had a strange shape to it — see the image below. On the bottom side, the roads in the slot at Y+ & Y- locations were not touching the fill.
Here’s some photos of the front, left, & top views:
I’d like to run this same part and stop after the first model layer is put down. Then repeat and stop after the second. Lastly, run again stopping after the third. Then compare the three partial parts. There’s something wierd going on as the part starts.
Some dimensions were improved, the part was still ruff (visible in Top View above), so I’m not doing any victory dances yet.
I’m thinking I need to get the Z dimension much closer to the modelled dimension as my next step. I’m gaining almost 0.001″ per layer as the part builds.
Before I left for the day, I started up the same part with Insight (same as I ran the other day), stuck the machine in auto shutdown mode, and I’ll see how that looks. I’ve got some other irons-in-the-fire that’s going to keep me busy for a few days though.
The small Insight part run the other day had the following measurements:
The X modelled to 0.926″ was 0.900″ X bottom, to 0.924″ X top ; The Y modelled to 1.546″ was 1.512″ Y bottom, to 1.539″ Y top; 0.250″ part thickness was 0.272″. The slot modelled to 0.176″ in X was 0.168″, and modelled to 0.796″ was 0.781″. The front part edge had a very noticable taper (angle) to it, and the left edge had a smaller taper. The other edges (right & rear) looked pretty straight. The support was harder to remove than the Quickslice part.
All-in-all, the Insight part was much better, & closer dimensionally, than the QuickSlice part. Some part photos are below:
‘HaveBlue’, in a recent comment suggested moisture in the ABS filament could be some of my problem. I fired up the machine to check it out using the ooze test. My ooze was more than 10x the acceptable number, and a photo of my printting head shows I should also do some cleaning:
I plan on using the chamber to cook the moisture out of the material. I lowered the Z down, and stacked the filament spools, & desicant on the stage.
I tried cranking the envelope temperature up to 80 Deg C, but it would only go to 75 Deg C. After the envelope got to 75, the machine alarmed with an over-temp alarm and shut down. I backed the setting off to the normal 70 Deg C setting. I cooked it about 1 1/2 hrs today, and I’m going to try get a whole day of cooking in before trying the ooze test again.
Last night, I remembered some details about the trays from the FDM 2000 and FDM 1650. The FDM 2000 tray had foam in it, and the FDM 1650 tray did not. When I started trying to run parts I simply grabbed the FDM 2000 tray and used that (thinking interchangable standardized parts). It fit, but was a little loose.
I put an indicator on it today to quantify ‘loose’. Here’s a shot of it pushed all the way in, and then a shot of it pulled back against the stop:
Just a bit shy of 0.045″. I couldn’t easily get an indicator of the X, but it didn’t seem nearly as loose as the Y.
Looking at how this is being located and held by the flip-latch on the left-front corner, there is a vertical rod in the center-back of the tray holder:
The button, on the left-front of the tray that the flip-latch catches, has a set-screw coming in from the top, so I guess you can adjust this in/out to make the Y thighter/looser:
On the back of the tray there’s a ‘V‘ to catch the rod. There’s also some spring loaded balls to hold — I couln’t feel any depressions next to the rod for the balls to snap in to:
There’s also spring loaded balls on the right and left tray edges:
I got out the empty FDM 1650 tray, and put the indicator to it:
Just over 0.016″ — quite an improvement. There was very little X play. I loaded a new sheet of foam from ‘HaveBlue’ into the FDM 1650 tray, modelled up a small part with a 0.010″ oversize scale slot in it, and processed thru QuickSlice using the Automode with FAST enabled.
48 minutes later (small parts take less time), I had the part. There were not as many hairs or goobers on it. Here’s a front view, and a left view:
The bottom layers (the 3/16″ noticed by HaveBlue’s in his comment to the last post) are still smaller, 0.897″ X Bottom, to 0.919″ X Top; 1.510″ Y Bottom, to 1.540″ Y Top. After the 3/16″, the sides seemed to straighten out, rather than appearing tapered as my previous parts. The 0.250″ part thickness again measured 0.270″. On the bottom, Y+ (rear side), the first 3 or 4 roads were only attached at the corners & to each other. Also, on the bottom, the roads in the slot at Y+ & Y- locations were not toching the fill.
The slot modelled to 0.176″ in X measured 0.155, and modelled to 0.796″ in Y measured 0.750″. The outside X modelled to 0.926″ measured 0.920″, and the Y modelled to 1.546″ measured 1.539″.
I’m going to run the same part using Insight, get it started, and set for auto-shutoff. See how it turns out later.
I’ve been juggling a lot of different things lately, so I have only had short bursts of time to work on this problem & get the Z Axis DRO in. I wanted to take some pictures and really try to figure out what’s going on with the FDM 1650. I was trying to make some parts so I could attach a Z Axis DRO to the machine to simplify the intial setup running a part.
The following pictures are as much as I got done today, and I still need to take more — that will be in the next post.
There have been a number of varions I’ve run on these parts — Fast/NotFast, QuickSlice/Insight, so it may get a bit confusing.
Initially what I noticed, (using QuickSlice in Auto Mode with FAST enabled) was a lot of hairs, and the slot was way undersize in Y.
I modelled the slot bigger, then I ran these parts, using QuickSlice in Auto mode, with the FAST Disabled. I was running three parts at a time initially, and this is photos of the first endpiece — it is a rectangular part, with a rectangular slot. In the slot, I could see a gap bigger to the Y+ side of the part between the fill and the road:
The gap was a bit smaller to the Y- side of the part:
There wasn’t any gap to the X+ side of the slot:
And also there wasn’t any gap to the X- side of the slot:
And then the four photo’s of the top part:
I also noticed one edge was completely loose (the picture doesn’t do the condition justice):
The fill hadn’t attached to the road at all except at the corners. When I had the FAST option enabled, there were layers that had three roads and the sparse fill, and that seems to hold thing together better. I ran a part and stopped & removed it after one of these layers. This is a scope photo of the slot with three roads running arround it:
It’s showing good attachment to the fill and would be binding the other layer below/above it.
Perhaps the problem is related to QuickSlice? I ran just the two endpieces using Insight, again using Auto Mode with FAST enabled. I was getting the same problem, but then I noticed the sides were not building straight in Z. All sides seem to be affected. Here’s the two front corners, then the two left corners:
The tapered sides are pretty obivous. A caliper check of dimensions (hairs & goobers removed) of the Insight part confirms:
X Dims (Aim 2.5)
Y Dims (Aim 2.5625)
Thickness Aim 0.250
I pulled the QuickSlice part and put the calipers to it:
X Dims (Aim 2.5) Bottom 2.469 Top 2.489
Y Dims (Aim 2.5625) Bottom 2.519 Top 2.538
Thickness Aim 0.250 Part 0.270
The numbers confirm the tapered sides, and just a little difference QuickSlice to Insight.
Running out of time today, and taking Rob’s suggestion to heart, I ran the cb12abs.sml file thru the FDM 1650. Here’s a photo of the front-left corner:
This is almost identical to the last time I ran it. The model thickness is 0.028 give/take on all four sides of the web. I didn’t try to peel a layer to check layer thickness. The support is nowhere near the center in X or Y. Also, with only five model layers (or 0.050) high I really can’t expect to see the problem I’m having. I may try to make one 0.250 high, but I’m not sure if a single track will go that high, or if you could measure it at all.
I’m almost thinking the stage is moving in X and Y as the part is building. I know I can jiggle it by hand as the latch doesn’t really lock it in. Something to try tomorrow….
I ran new parts for the Z Axis DRO. On the end plates, I added 5/16 Inch to the front edge to shift the scale closer to the table, & also opened up the cutout. For the small part, I didn’t change any of the model at all. I used QuickSlice, ‘Run Auto’ with the ‘FAST’ option not selected. (~9 1/2 Hrs without ‘FAST’ vrs ~8 hrs with ‘FAST’ — just slightly more material and support used without ‘FAST’)
The parts looked generally better, but when I went to fit together there was a problem with the dimensions. With only ‘single’ roads on the model edges, there is a gap between road & model on one edge (Y+ I think). The small part still had a lot of ‘hairs’ to it, and again the 0.100 slot was almost completely filled but not as bad as the first one. The supports came off better this run than they did before — perhaps I’m getting the knack of removing them.
I need to sit down with print, part, calipers, and ‘scope (tomorrow I hope). I’ll post the findings later.