Moisture in filament shipped with A350

I’ve read multiple reports here about problems with the filament shipped with the new 2.0 machines. I had not yet opened the packaging on the reel that came with my machine, so I decided I could perform an easy measurement to estimate the moisture content.

Experimental setup: I put the spool of filament into desiccator cabinet containing an open jar of desiccant. The desiccant I used, because I had a bunch of it around, is indicating Drierite from a freshly opened bottle. I weighed the spool and the packet of silica gel and sealed the cabinet. 24 hours later (a few minutes ago), I weighed again. I’ve resealed the cabinet for subsequent measurement. The scale I weighed the spool of filament on has 1 g resolution; that for the silica gel packet has 0.1 g.

I had planned on taking a few days data before posting, because silly me, I thought the first-day results wouldn’t be so obvious.

  • Filament. Day 0–1275 g, Day 1–1273 g
  • Silica gel. Day 0–9.9 g, Day 1–9.7 g

That’s a ~2 g loss in one day in a passive cabinet with no internal air circulation and a spool still wound compactly at the factory. Both of these limit the rate of water removal from the filament. I’ll need more data to see if this rate continues, but if it does for a week, it’ll mean the filament ships with 1% mass of water. Let’s just say that’s not very dry for a printing filament.

The results on the silica gel are significant as well. The fact that it lost weight means that it had saturated its capacity while sitting in the packaging, that it wasn’t able to absorb all the residual water. When you put a tiny packet of desiccant into a package with an excess of water, don’t expect it to be particularly effective. Call it “desiccation theater”.

I’ll continue to gather data and update periodically.

3 Likes

Would you mind additionally taking diameter measurements? I’m curious how the change in moisture will affect the filament diameter. This is great, thanks for doing this!

Edit for below: fair enough, thanks again.

I don’t have an accurate-enough device for measuring diameter. My calipers are utility-grade and I don’t have a proper micrometer. If I had an optical comparator of adequate resolution, that could work, but my shop isn’t that big.

Update.

I took data for ten days. I stopped because the weight seems to have equilibrated. Rather surprisingly, the spool of filament lost all of its weight within the first day; it’s still at 1273 g. It might be still losing water slowly, though. As the scale was equilibrating at the start of the measurements, it would fluctuate between 1274 and 1273. Yesterday it did the same, but between 1273 and 1272. It’s not definitive, but it is suggestive. As I said before, I didn’t rewind the spool, so gas exchange with the interior of the filament is quite slow. If I had the ability wind the spool with spacers for rapid gas exchange, I might try that, but I don’t. Instead I’m going to end this experiment and start using the filament.

The silica gel equilibrated at 8.4 g, losing net 1.5 g at a steady rate, slowing a bit at the end. This is a lower bound on the total amount of water the gel absorbed. I’d have to heat the gel to drive off any remaining water. I should point out the Drierite works by absorbing water into a crystal structure and its quite energetically bound; it requires elevated heat even to partially regenerate it. Silica gel will lose some of its adsorbed water spontaneously if conditions are right.

I had not been familiar with PLA chemistry before I launched into this. Chemically, PLA is a polyester. Polyester bonds form by expulsion of a water molecule. Bulk PLA can contain some of this water entrained within its interior; it’s amount varies with the specifics of its manufacture. Part of the quality of a PLA product is how much of the entrained water remains. Thus a lower-quality filament can be stored in a completely dry atmosphere and still have too much water in it. As a result I’m taking these results with a grain of salt. If the variation in quality of SM’s PLA product is due to water, it might have nothing to do with storage conditions or water on its surface.

Nevertheless, it’s clear from the experiment that the filament lost a couple of grams of water readily and the silica gel had absorbed at least a gram and a half. So it’s not that the shipping desiccant was useless, but it was certainly insufficient to dry off the filament.

1 Like

From my understanding the shipping desiccant will just remove airborne moisture.
I’m actually surprised you removed that much moisture without using an actual heat source (oven, dehydrator). I’d always read that silica gel wouldn’t work to dehydrate filament - you couldn’t just put it in a bag and expect it to dry.
I’m curious how much more weight would be lost with a dehydrator.
-S

It’s true that silica gel only removes moisture in the air because that’s the only thing it’s in contact with. The missing piece is that moisture can enter the air after packaging if the filament has enough adsorbed water on it.

Filament can have other water, though, if it’s got entrained water from manufacturing in it. It’s this water that you’d need to use heat for. Getting that water out at room temperature is a solid-phase diffusion process, and that’s very slow.

Were I to continue this experiment, adding heat would be one of things to try. First, though, would be to rewind the spool and see if more moisture were to come off the surface at room temperature. The way the spool is wound up tight, diffusion of any vapor from the spool is likely to be slow, so I’m not sure the cabinet is even in equilibrium yet. All I know for sure is that, if it’s not, any rate of loss is slow enough that it’s below the resolution of my scale.

Based on some offline discussions with @brent113, I’ve begun thinking that the right way to calibrate these machines is not by diameter, but by mass per unit length. Manufacturers currently do not specify their products this way, but it’s straightforward to do in the field. A machine that rewinds filament only needs a sensor added to measure its length. An at a nominal 1 kg spool size, even my cheap scale with 1 g resolution gives three significant digits. For commerce you’d still specify a nominal diameter, but for operation you’d use the linear density.

1 Like

An update: After having done some measurements on my heated bed, I found it’s more than 2 mm out of flat itself. I decided not to bother with the FDM head until after I’ve rebuilt the Y-axis, so that’ll be a while.

Thus, the filament that came with the machine has remained in the desiccator cabinet, happily diffusing away. It’s lost another 2 g in the last two weeks, doubling the loss from the first couple of days. This bolsters my thesis that moisture loss is limited by the close winding of the spool. I’ll be looking at ways of winding the spool with spacers that allow much more rapid removal of moisture.

1 Like

If it’s dehydrating from the outside in as you say, you could wind half of it onto another spool and dehydrate “the middle” of the spool. Then continue to wind it onto the other spool and dehydrate “the inside” of the spool.

1 Like

What I was thinking about doing is making some spools and spacers specifically for a drying process. They’d have to be physically larger to account for all the air space. It would have space between layers and between adjacent strands of filament, so about four times the volume, roughly. The flanges would be slotted for air flow. To fill it, wind one layer of strands and then wrap a layer of spacers, then continue with the next layer. To minimize the number of times you have to insert spacers, I’d use a somewhat large diameter (but small enough to fit in my dessicator) and a larger width for each layer.

The idea, if it’s not clear, is that almost all the surface area is simultaneously exposed to air flow. If most of the moisture is on the surface, it should only take a day or two to dry everything. If there’s entrained moisture that can come out, this provides maximal moisture gradient for that to happen.

Sounds like you’re describing transformer windings, with wood spacers for oil coolant flow:
image

1 Like

In that spirit, yes. But I’d put spacers every layer, not every fourth as shown in the photo. And in addition, I’d make the spacers corrugated so that individual strands don’t lie next to each other.

2 Likes

Just design a contraption with two spools that winds from one to the other and then reverses when it gets to the end and so on.

-S

1 Like

Lol just let it keep rolling back and forth until eventually it air dries?

I will add - that I had some filament from Fiber Force, a pantone color match, very beautiful but wet as can be from the box.

Measured a solid 1.75mm, very accurate.

Dried it out in the dehydrator, shrunk by .02mm.

1 Like

I’d put the whole contraption in a drying box with some heat too. I’ve heard you can just use a reptile heating pad.

-S

1 Like

The food dehydrator I have seems to do the job just fine, almost every spool i open seems like it prints wet and i have to dry it out.

literally going to just put them in there prior to using a filament the first time at this point.