Filament Drying preliminary results

[MZip]

Wall of text but pretty pictures?

I finally got all my parts so today was able to start testing. Already have good and meh things to report. We had rain showers this morning so ambient humidity has been high but dropping. The aquarium pumps I’ve tried are high flow at low pressure.

Turns out the original small quiet and efficient pump I started with is too underpowered. Luckily I had a bigger one on hand and that one seems appropriate now. It was taking too long to start seeing the empty filament dryer RH% drop but response is snappier now at the cost of more air through the desiccant bottles. The bigger pump is running with its throttle valve full open. The down side to increased flow is you’ll use up desiccant faster.

Part of the slow RH% drop is just volume. With such low flows, in the empty dryer the big mixing volume and fan in the S2+ slows down getting humid air out. I put some filament in that uses up a lot of the dryer volume and it seemed to pick up the pace again. It may be better to disable the fan that’s doing all that mixing so the air stratifies more and hot moist air exits out the top holes with cool dry air entering from the air dryer at the bottom. That may increase efficiency though it won’t help sweeping released moisture away from the filament surface as it’s liberated. For best drying you want flow around the surface. Something else to test.

Anyway, it seems to be working fairly well as is. Here it is a little after 3 hours in though it’s only about an hour for the filament spool to be in there. The main points are the air out of the silica gel has pegged its hygrometer low and apparently 10% is the minimum indication. With <=10% RH air entering the Drierite, it’s mostly loafing and the outlet from the Drierite is also pegged low. But this answers a question of if the silica gel beads can pull water out of the air fast enough to be effective. At least when fresh they can.

Something else interesting is I was playing with my IR camera and spotted this - I think it’s the heats of adsorption and condensation visualized. The air from the pump runs through the first hygrometer chamber and then down the center tube to the bottom of the silica gel bottle. In the infrared the bottom of the bottle is lighting up with a few degrees of heat.

The pump itself is warm and so is the tube carrying the air to the first hygrometer housing but the temperature drops low as it enters. The hygrometer is just showing 1.2F over the air exiting the silica gel. You can see the exit tube that carries the air up to the bottle top - it’s not warm and certainly not near as warm as the bottom of the silica gel bottle. That’s why I think that’s heats of adsorption and condensation. As the water vapor basically condenses onto the silica gel and sticks to the silica gel surface it is releasing energy as heat. For those unfamiliar with this it’s the same reason water condensing on a cold glass helps to warm it up.

But it may be a way to know how much capacity is used and left. It may be a “hot” band that moves through the desiccant when it’s in operation that will visually show what’s left. Kind of cool if it works that way.

Also, before people point it out, I’m not putting too much faith in those temperatures and relative humidity numbers. At least on those three hygrometers in the desiccant caddy, they ran for a few days with the little pump to flush out silicone seal fumes. During that time they were in lock step showing the same or close temperatures and humidities with maybe a tenth of a degree temperature difference or a change of humidity rippling through. The desiccant caddy has a little raised place inside the hygrometer housings to help cause turbulence and direct the air into the hygrometer sensing ports. It seems to speed up response and increase precision/repeatability. Can’t say on accuracy, though. But it looks like comparisons between them are pretty golden.

It’s still running with that filament spool still in there. Here’s the displays just now. The filament has been in about 3 hours now and showing 22% RH in the dryer with 43% RH ambient air entering.

I’ve got tare weights and weights of desiccant and filament before it all started to see about water content whenever I call this one. That will give an idea of how much desiccant this will go through but will be very subject to ambient humidity. So far I can’t see any color change in either desiccant (since not breaking through, there won’t be color change in the Drierite) but it could be channeling along the center with air maybe not flowing up through all the desiccant. That infrared image implies it’s sicking close to where it’s entering, though, at least for the silica gel. The silica gel being a lot easier to regenerate might argue for forgetting the Drierite and just going with silica gel especially with this amount of silica gel not letting humidity break through. That’s why I included the Drierite cleanup in case it did. Drierite grabs water fast and hard. I’m sure something is happening behind those 10% RH exit humidities but we can’t see it yet.

But it at least seems to be working. It doesn’t seem to be blasting through desiccant while the dryer humidity is now down to 22% RH. That the humidity isn’t breaking through especially with the bigger pump running full bore may make the Drierite unnecessary. Now I need to get an idea of water capacity in all this and see how well it agrees with the curves. Weights will tell a lot I think.

I’ve got a 2 pack of the crummy jade white PLA that I’ll be able to do comparisons with. They were stored together and hopefully close to each other in manufacture and so have similar water content. I think that’s the best I’ll be able to do but my S2+ is modified with a different base. One will get the filament dryer with the lid propped open using a clip from MW and the base shouldn’t interfere since I’ll close the side ambient air port. The other one will get dry air from this dryer. May the more dry spool win.

[MZip]

Just took these images at four and a half hours into the run. This filament came out of my AMS showing 2 on the dry value so was already in fair shape but it was handy for testing. Anyway, it started at 36% RH in the already running and getting dry air dryer chamber and is now down to 19% in four and a half hours with an ambient room humidity of 46%. If I’m interpreting the heat in the silica gel correctly, it’s still not used much desiccant and has been running total for six and a half hours though an hour was at lower flow with the original pump.

No comparison data yet and that will be a couple of days more to get, but this filament dryer issue is starting to look pretty easy to solve with just an aquarium pump, some fittings, and a bottle or two of silica gel. The ease of regenerating it makes it pretty cheap. The air pump was $21. If you get indicating silica gel (not the blue kind - I bought it before I knew it was nasty chemically) you probably don’t need hygrometers. This can be streamlined a lot.

I’ll know a lot more after the weights are in and I run the comparison between the two jade white rolls that are still in their sealed bags.

After 7 hours in the chamber the RH hadn’t dropped below 19% so I ended it and put the spool back in the AMS.

The silica gel gained 3.84g. The Drierite gained 0.35g. The filament lost 1.08g.

For whatever it’s worth. I’ve seen some use 20% RH as an endpoint and the chamber hit that at 4 hours. The two runs using the jade white will say more. 19% RH in chamber on a 46% RH day in a Sunlu S2+.

[user_3026326371]

Nice and nicely detailed

I am in the early stages of letting my Sunlu dryer grow up a bit.
Will be very similar to your approach just with an old bar fridge to replaced the original ‘container’ for the electronics.
Getting sick and tired of sealing containers or bags with filament, so something to hold a decent amount of rolls is a must now.
The bar fridge is a proof of concept and if it works and can find a nice sized fridge for free…

A few things that might help with your project:
Air pumps for fish tanks are nice, air pumps for medical air mattresses much better
Quite hard to find second hand as most are fully serviced until thrown out.
So try to find a matching service provider and contact them
Ask for some old and no longer serviceable pump for ANY air mattress as you only need the actual pump for a project.
Aged care facilities are worth to consider as well, just ask nicely if they happen to have some no longer used pump for an air mattress that you could get for donating some donuts or such to the staff
In some cases those vital bellows will be hard or cracked - there is a reason they come so cheap …
BUT: if you measure their diameter and thickness you can order replacements from China for peanuts.
Their only real point of failure are those reed valves hiding under the little rubber caps or inside the plastic housing.
Place them into and ultrasonic cleaner with soapy water that you de-gassed first and they are good as new.
The airflow of these pumps is about 4 to 5 times of what you get from a large double action aquarium pump.

You main limitation is restrictions.
It is hard work to get a lot of air through small hoses, especially under low pressure.
If you place a small size fridge compressor into the dry air output, behind the silica bottles you won’t have air flow issue at all, you probably will add a small valve to lower the airflow rate a bit.
There is good tutorials out there explaining how to turn those compressors into air or vacuum pumps.
Ignore all those that won’t include someone WORKING to collect and drain back the oil that is included in the air coming from the compressor!
In the most basic form a simple catch can like on old cars can be made up that you manually drain back into the compressor once it’s fill level indicted the compressor is running a bit low.

You FLIR images nicely show that the silica gel is working - ir heats up when taking in water but I guess in your case the hot air from the box is going in there ?
Seems to be a bit too much in temp unless you have a rather restricted airflow down there.
Did you ever check how long it takes for a full roll at room temp to get to the set temp of the dryer?
I found some filaments can take over an hour until their insides are on par…
Would be nice if you would have some data on that one one day to see how long your modded system takes.
I just love it for how simple and easy it while providing such great results out of the box.

[MZip]

Thanks for the pump info but I think this bigger aquarium pump has enough flow and pressure for this. I don’t want much flow or pressure because more flow is more water in the desiccant and wasteful. But the way it’s looking this isn’t using that much desiccant so there could be some wiggle room. The aquarium pump new is $21 so cheaper probably and definitely easier than refurbishing medical equipment.

The tubing and fittings are absolutely restrictions to flow that the pump has to overcome and the little pump still worked - just very poorly. I’d been watching the water column pressure at the outlet - easy to measure with a tall glass of water - and when I finally put the desiccant in line, I was down to just 2” of pressure and very low flow. But the bigger pump now supplies air with fair volume and pressure and with a spool in the dryer the volume gets easier to change out. I do suspect with the holes at the top for filament, the fan in the Sunlu is actually slowing down the air changover as mentioned above.

I’m not sure what you mean about hot air from the box heating the desiccant in the IR images? If you mean hot air from the dryer, that exhausts to the ambient air. The pump warms up as it runs and that does heat the air some but like I pointed out, the tubing heat glow largely fades by the AMBIENT hygrometer and the tubing from that hygrometer to the bottom of the silica gel container wasn’t showing warming. At least as measured by the first hygrometers, there was only a 1.2F increase due to the pump. I do think the heat seen in the bottom of the silica gel bottle is heat of condensation and adsorption.

There’s a lot I haven’t tested yet. I just got it finally assembled yesterday morning to finally get some testing done. A key package was lost in shipment and I had to wait then reorder. As to starting from cold, that’s coming when I test the jade white rolls. That’s a two pack where the spools are individually wrapped but packed together, likely were close in manufacture, and saw similar conditions to have similar water content. Those will start cold with whatever ambient humidity is present at the time and the plan is to compare drying with and without dry air.

What yesterday’s test showed, though, was even with “dry” filament that’s been in the AMS for over a week and reported humidity inside it of “2”, this setup took out another gram of water on a humid day as showers rolled through with windows and doors to the house were open.

The thing I built can be simplified which will cut flow restrictions and allow more air through. I built a test fixture that would allow monitoring all this stuff which is a different purpose. These results will get fed back into a new design that’s better optimized for actual use. It could use a single desiccant container and larger tubes and fittings for less restriction but that may not even be necessary. Extra flow and using up desiccant faster is a balance with fast drying. But with more flow possible others could pick their own tradeoff.

I’ll know more in a couple of days when I dry the jade white but this is already showing lots of promise as is. When I get more information worked up on how much capacity there is in the amount of desiccant I’ll have a better idea of how big the desiccant bottle(s) need to be. It looks like the Nalgene bottle size may have been a lucky guess at pretty close to a pound of desiccant in each. We’ll see.

Anyway, good luck with the fridge. There’s more than one way to do this.

[user_3026326371]

I actually prefer to go the other way around with the silica gel, I use large diameter containers with less length.
Provides far fewer restrictions for the airflow but I mainly do it because it slows the air flow big time.
With that the air has more time in contact with the silica gel.
Goes a bit like this: Have large diameter jar, can or print something - about 10 to 15cm in diameter.
Add a screen about 2cm from the bottom - no beads here only above.
Gives a really nice and even flow.

For the AMS I have about half a kilo in there and it stays on 1, once it goes to 2 and stays there for a day I refresh the beads in the microwave.

We might have been talking in opposite ways with the pump and air before, my bad.
Let me try to explain how I do my dryers so we are on the same page:
I did not bother to mess with these tiny filament dryer boxes, I have a Sunlu here for emergencies or to use with TPU, the rest is done differently.
I suck the air of my heated container and then it goes through the silica gel and BACK INSIDE the container - I don’t use ambient air unless it is a nice summer day with humidity below the 30% mark.
The beauty of this approach is that the beads only ever have to take in what the filament gives off.
I use about 400g of beads and once there only about 5cm of still fully blue beads on the top I refresh them before adding another roll.

In the past I tried the heat the outside air up, run it through the beads and then use the ‘dry’ air on the filament.
Worked about as good but meant I was using my microwave more on beads than on food and drinks.
For the fridge attempt I will do the same, just with the beads container inside and a fan forcing air through a little heating block.
BUT: The container for the beads, at least for a fresh load of filament, will stay closed.
First the fridge is heated, around 55 degrees for PLA.
Once the humidity goes above ambient humidity levels a little fan kicks in every now and then.
With ambient levels too high the fan won’t go on and instead I open the beads container.
As the fridge itself is still working (busted thermostat) I am tempted to abuse it to replace the crappy cooling plate in the freezer box with a little cold trap.
Little funnel and hose below it and condensing water just runs out.
Compressor has net to nothing to do as the temp sensor on the cold trap will only start the compressor once it is back to at least 10 degrees to allow any ice that might have formed to drip off as well.
If the math works out the bar fridge should be able to dry 8 to 10 rolls of left open rolls down to under 20% in about 4 hours.
Will make a proper write up in a few weeks once I got it all together and properly tested.

[huminlian]

This is great. Do you have a pellets 3D printer?

[user_3026326371]

No
Just a P1s and a resin printer, I retired my dual head printer when I ordered the Bambu - what a bad decision in hindsight…

I did try a pellet extruder once though.
Long story short:
Unless you want to spend fortunes to get good print speeds there is no way.
Moving such a heavy extruder around requires beefy hardware.
Not to mention the nightmare of keeping a good flow of pellets going.
SLS is nice though but still way too costly for hobby users

[MZip]

@user_3026326371

I actually prefer to go the other way around with the silica gel, I use large diameter containers with less length.
Provides far fewer restrictions for the airflow but I mainly do it because it slows the air flow big time.
With that the air has more time in contact with the silica gel.

I need to think about this one. In chemistry labs, drying columns or any kind of column tends to be long and narrow. Larger diameter will slow flows down at a given flow rate - and good point about the flow restriction. A long, narrow column will have more restriction. But your approach would recirculate so the length of the column doesn’t even matter.

I do like the part about closed looping the air through the dryer where the moisture it absorbs is the much lower amount from the filament. The way I’m doing it, it does indeed see more water by drying the air before the dryer so more bead regeneration, etc.

I think that approach will have some down sides too, though. The silica gel will be heated by the hot air and add load to the filament dryer heater. It will also trap less water as it gets warm. And circulation fans or pumps will be pumping hot air which is harder on most.

The base I made for the Sunlu S2 might be able to be modified to do something like that. My first mod to the Sunlu base does indeed circulate air into the base. That could be modded to circulate through a bead chamber. Olias’ suggestion about making the footprint bigger for stability would dovetail well with blowing out the walls to get bead volume.

Countering that, I had to make the base walls thicker and use honeycomb infill to in effect print insulation. There was too much heat loss for the Sunlu heater to be able to hit and hold 70C with my first version of the base. I need to think about this.

I like these suggestions though. They might be difficult to add to a Sunlu but in a clean sheet design it could be really effective.

[ElektroQuark]

In labs we use large and narrow containers for the solid/gas contact reactions. If wide containers are used, gases find the shortest path between input and output and not all the solid phase is used nor the gas phase is homogeneously mixed.

[MZip]

The long columns with more tortuous path are definitely better for single-pass kinds of things especially if the kinetics are slow.

What 6371 was saying I think is to recirculate the air over the silica gel where it doesn’t really matter if all the moisture isn’t removed first pass. What was missed can get trapped on subsequent passes. A short wide column will have lower resistance to flow which is another feature of that approach - less pumping losses. But flows don’t need to be low and kinetics don’t matter so much (though in this case that’s not an issue anyway). Higher flows would help move water away from the filament too. It’s actually an intriguing set of advantages.

He also suggested a plate or screen below the beads to get it to flow more equally by admitting the air below that screen/plate. I almost did that on my test fixture design but wanted to on with testing.

Some disadvantages too, but 6371 might be onto something.

[NeverDie]

@MZip

After 7 hours in the chamber the RH hadn’t dropped below 19%

I’m wondering why it didn’t drop to the 10% RH that was exiting from your drierite chamber? Was there just not enough flow?

I’m guessing you’re using the drierite straight from the bottle it came in. Maybe it picked up moisture before you got it and needs to be properly baked in an oven to bring it up to peak performance so that it can drive the emitted RH below 10%? Maybe the color indicator doesn’t show early degradation but is more attuned to saturation, which maybe wouldn’t be all that useful.

[MZip]

I do think it is the low flow made worse by the now more effective circulation fan I cobbled in. It’s also why the RH started dropping faster when I put a nearly full spool in the Sunlu which decreased the volume a lot.

With a large well-mixed air volume, as dry air goes in, it gets mixed in and the air that is expelled is slightly reduced in humidity. As time passes, dryer and dryer air is being expelled.

Ideally it would only be expelling the moist air with no mixing which is why before doing the dry air test with the jade white rolls I’m going to disable the fan. That will let the air in the dryer stratify and the cool dry air will rise from the bottom while the hot moist air exits.

I think it will dry faster because the humidity will be a lot lower low in the chamber but higher near the top so it would probably help to give the spool a half turn every now and then. The counter to that is wind and turbulence helps move boundary layer air and dries faster. Not sure which will win but it was taking a long time to drop more with the fan.

I think you are right that the RH should have kept dropping (except for the mixing). I have no reason to doubt that the air going into the Sunlu was well below 10% RH. I’d guess it was below 1% but don’t know that. With the air humidity measured at the outlet of the Drierite pegged at 10% or less (since that’s an instrument limitation), I don’t think it was wet Drierite. And I’ve used Drierite a fair amount before. The company that makes it has lot numbers printed on the jars and quality control. Plus it’s indicating Drierite and still showing it’s in the dry form. Good point about what level of saturation causes the color change though. The jar did have the hot-pack swoosh sound when I opened it so I don’t suspect a leaky seal.

Those hygrometers all agreed to the digit on humidity when I was just pumping air through them to purge fumes. I think it was Olias who said in another thread that the hygrometer sensors were pretty much all from the same company and with that agreement, if they are all the same, we can probably trust that 19% RH in the chamber and the 10% or less out of the air dryer.

There was also a water source in there in the spool of filament and the Sunlu was just below PLA drying temperature (54C vs 55C). RH might have dropped faster without that spool as long as I had something dry that took up the same volume in there.

This subject of dilution in a larger volume is a whole field of study and how reservoir volumes and flow rates can be determined - by adding some kind of tracer and measuring exit flows to see how long it takes to clear, concentration over time, etc. Our tracer is dry air and the reservoir is the filament dryer volume. I just don’t have a way to easily measure exit RH%.

[NeverDie]

Not to prejudge the outcome, but closed loop does seem like it would be faster, since you could crank up the flow rate to be a lot higher, and you’re not fighting the dilution problem. Most of my measurements have been on filaments packaged on cardboard spools, but on the few measurements I’ve done for filaments packaged on plastic spools, I think I’ve only measured at most about 3g of moisture loss. Then again, I haven’t measured nylon or tpu, which are both sponges. For everyday filaments, though, suppose I’m 3x wrong and the worst case is 10g of moisture that needs to be dried out of a filament mounted on a plastic spool. It’s apparently enough to make a difference in print quality, but not really all that much in absolute terms, especially compared to your big jar of desiccant. And, as far as circulating heated air through the drierite, the manufacturer claims:

There is little change in efficiency at temperatures up to 100° C.
Drierite - Drierite Overview

The faster dry time might not only be more convenient, but it would also minimize the heating time of the filament (we’ve probably all read the same assorted vague warnings from different sources about dry heating filaments for “too long”).

For the more exotic filaments, though, your current strategy is probably the most bullet proof and least likely to fail, as it can surely cope with drying temperatures above 100C (well, not the sunlu, but the general approach). For worst case estimates, I look to Vision Miner, which typically cites higher drying temperatures than just about anybody else I’ve read:

It may not be my daily driver, but I could for sure see myself on occasion needing to dry PC and/or PACF, where Vision Miner recommends 120C as the drying temperature. Would drierite be only minimally degraded at 120C? Maybe. I doubt performance falls off a cliff at 120C, but without knowing, it’s a potential gotcha. If it were a gotcha, then your current approach would solve for it. But if we knew and didn’t have to guess, it might turn out to be overkill in comparison to closed loop. The answer is obviously discoverable, just not sure exactly what experiment would decide the matter definitively and settle the matter once and for all.

[MZip]

@NeverDie

And, as far as circulating heated air through the drierite, the manufacturer claims:

There is little change in efficiency at temperatures up to 100° C.
Drierite - Drierite Overview

Absolutely true. I was referring to silica gel though where temperature can have a significant effect depending on saturation level.

Drierite binds the water chemically and is significantly more expensive than silica gel if you buy the name brand. It does work at warmer temperatures.

But at this point things are looking like silica gel is the way to go. If it’s dry it can pull the RH right on down - really low. It’s cheap, easy to regenerate, and the kinetics are good.

[MZip]

Agree on all points. @user_3026326371’s suggestion for recirculation is a good one and I agree that that could be a winner.

Recirculation has some issues to solve though since it would be recirculating hot air so everything needs to be made of materials that can take the heat and it needs insulation (even printed) to not load the filament dryer heater. The Sunlu S2+ is the only dryer I’m familiar with but until I printed my base for it with thick walls and honeycomb insulation infill, it was too much additional heat load and limited the Sunlu to 68C with the heater always on. Not a stopper but does make things more difficult.

But along those lines, i’m looking at pushing the walls out of my base, making them thicker, and putting in a ventilated desiccant drawer. The fan replacement I did makes it possible. It won’t be air tight though and will bring other challenges but might be an alternative. It wouldn’t need the air dryer fixture at all but the smaller volume of desiccant would probably need to be changed out each or every other spool. Or maybe longer - but more frequently than jars because it would be much less desiccant.

Thing is the reason I jumped on the air dryer approach is I got skittish at posting designs that were modifications to a heater. Drying the air at room temperature with the only thing plugged in being an unmodified pump is a lot safer not knowing people’s skill levels or how well they follow directions. I want minimum liability. I’ll have to see what things look like, though.

But that doesn’t take away from a recirculating heater. There are a number of advantages including the ones you mentioned.

[NeverDie]

Since, at least for now, you’re sticking with your Sunlu and its 68C maximum, one could maybe still argue for also sticking with silica gel, even in a closed loop hot air system.

It does look as though the efficacy of silica gel does plummet at higher temps, but eyeballing this chart it looks like it might still hold around 10gH20/100gDesiccant, even if heated to the full 68C. In it’s favor, it’s cheap; so maybe by using, say, 3-20x as much of it as you would at a cooler room temperature it might compensate? I have a hunch this graph isn’t telling the whole story, but maybe you or somebody else here knows for sure whether maybe you can overcome the temperature impairment with just lots more quantity. And, if not, probably a simple experiment will tell the tale.

[ElektroQuark]

The ultimate upgrade:

You can add a dehumidifier phase for the intake air.
You need a peltier cell. The intake air must contact the cold side of the peltier and the water will condense, go throught a water trap, and then guide the dried air to the hot side of the peltier, so its temperature raises before reach the filament chamber.

[MZip]

The printed-in insulation by using thick walls and honeycomb infill fixed the 68C limit. I was just saying the Sunlu dryer doesn’t have much headroom for heat losses and pointing to my first iteration that had gyroid infill and thin walls to use minimum plastic. I can hit 70C again and infrared images show a decrease in heat loss. In one image there’s a fair amount of heat escaping through the base and the other base has an outside temperature that is cooler and more uniform. That’s the one with printed-in insulation. The bottom image is the old design.

About silica gel in a recirculating dryer, it should/might/maybe will work but the caveat is the gel will probably needs to be kept dryer and regenerated more often. But margins look like they would be cut to the bone if it would work. Or, worst case is you just use Drierite instead since as you saw in their literature it is effective to 100C. The down side to Drierite is it takes higher temperatures and longer times to regenerate.

Welcome to the tradeoffs in design.

[MZip]

I haven’t used Peltier coolers/heaters before and what you say sounds simple in concept, but what I’m finding is they are not very efficient and use lots of electricity. Quiet and simple though, and reheating the air is a good idea on the hot side, but I’ll leave Peltier experiments to others.

I’m trying to keep this low tech and simple - and bring this to a close instead up ramping up other projects. I’m way behind now on other projects I want/need to do but if you have data that shows Peltier advantages I’d love to see it!

[NeverDie]

Well, whichever desiccant you ultimately choose, your build will still have the virtue of being unique. I’m not aware of any other filament driers on the market that are taking your approach.