I have recently, in anticipation of printing carbon-fiber-reinforced nylon filament (Polymaker PA6-CF) on my Prusa i3 MK3S, built a filament drybox. It is made out of a large aluminum equipment enclosure which I had spare, every seam, rivet, hinge mount, foot screw, sealing lip, etc of which has been sealed with copious amounts of high-quality silicone sealant (this is condensation-curing silicone, so it does release a small amount of moisture as it cures, but it has been in here several days and is about as cured as it gets now). The lip around the edge of is clamped hard against a rubber sealing ring, which I have cleaned meticulously, when the lid of the box is closed and latched. In short, it is, by 3D printer drybox standards, unusually well-sealed.

The exterior of the box Inside (yes, I know my silicone work is a mess, but let nobody say I missed a spot!)

For drying, I have poured two pounds of orange self-indicating silica gel, freshly regenerated, directly into the bottom of the box, exposing the maximum possible surface area (as opposed to having it in a container).

Humidity is monitored by a DHT22 temperature/humidity sensor, connected to an external microcontroller used for data reporting. The wire passthrough is a 1mm diameter hole in a custom 3D-printed passthrough (held in by a printed nut) with 3 pieces of magnet wire running through it, sealed with glue from both ends. I really doubt it's leaking through this either. The datasheet for this sensor indicates it should be accurate even at low humidities, and is accurate to +/- ~2.5% RH over most of its range and does not exceed +/- 5% anywhere. Per the datasheet, its accuracy should far-exceed that of the little battery-powered digital hygrometers many drybox builds seem to use. enter image description here

Despite all of this, per the DHT22, it only reaches between 16.1 and 16.5% RH. As I understand it, a well-sealed drybox with well-exposed desiccant should be capable of better than this - I've seen sub-10% reported online. It's also worryingly close to the 20% maximum storage RH of the filament, per its TDS. What's going on here? My only real theories are:

  • A lack of internal air circulation has resulted in stratification, with denser dry air close to the desiccant at the bottom and damper air at the top where the humidity sensor is located - this might be fixed with an internal circulation fan?
  • My sensor is not as accurate as its datasheet claims
  • The desiccant is somehow defective
  • My expectations are unrealistic and 16% RH is normal and acceptable performance for such a drybox

Having tried most of the obvious things already (seal the box better, add more desiccant, increase desiccant surface area, let the box settle over days, install a high-quality sensor, ensure nothing inside the box is releasing absorbed moisture, seal the box even better), I'm beginning to expect it's the latter. Is it acceptable to remove my nylon from its sealed packaging and begin storing it in the drybox with this performance? Have others had success with nylon stored at 16% RH (as opposed to the sub-10% RH I've sometimes seen people talk about dryboxes reaching)? Are reported sub-10% readings for other dryboxes on the internet even accurate, or do the cheap little battery-powered hygrometers most DIY dryboxes use read inaccurately low at low humidities?

  • $\begingroup$ Implementing a way to provide high precision monitoring of differential pressure between the inside of the box and ambient barometric pressure may shed some light on whether you have a leak, though that may be complicated by temperature changes. In a nutshell if internal pressure closely follows barometric pressure changes, i.e. dp remains close to zero, there's probably a leak. $\endgroup$
    – allardjd
    Mar 25, 2022 at 14:09
  • $\begingroup$ Install a bleed orifice (possibly an extruder nozzle) in the container and keep the container SLIGHTLY pressurized (a few inches of water) from a regulated dry air/gas source. Bear in mind that the container is not designed to be a pressure vessel so due caution is called for. A tube attached to the orifice and submerged in a liquid (light oil?) would create a bubbler to indicate the volume flowing through the container. A manometer could be rigged quite easily to provide over-pressure protection. $\endgroup$
    – allardjd
    Mar 25, 2022 at 14:33
  • 1
    $\begingroup$ I live where humdity is 50-60% normally and I've easily achieved <10% humidity with a 5 gallon bucket and a screw on lid with an o-ring. I use one or two fist fulls of the desiccant in a cloth bag, not a whole two pounds. So your expectations are not unrealistic, but perhaps your seal is not as good as you hope or there's a hidden hairline fracture somewhere. Ironically, I'd first suspect the entry point for the sensor. $\endgroup$
    – user10489
    Mar 26, 2022 at 13:12
  • 1
    $\begingroup$ @allardjd I unfortunately do not have the equipment to do this test, or to hook up a compressed air/gas source to the box. It's a great idea though. $\endgroup$ Mar 28, 2022 at 7:28
  • $\begingroup$ @user10489 I did a relatively thorough inspection of the box and was not able to find leaks. I gave the sensor feedthrough another look, but I really don't think that's much of a factor - it's sealed with copious amounts of glue at both ends, and the actual hole is only 1mm in diameter (and mostly stuffed with glue anyway). $\endgroup$ Mar 28, 2022 at 7:29

2 Answers 2


Track humidity graph over days (one measurement every 3 hours is fine). Be careful that polling the sensor too often may warm it up and cause inaccuracies, the temperature dependence is quite strong. Also, you may need to poll twice at 5 s interval and discard the first reading if you want more accurate values.

If the increase in humidity over a month is linear, the sensor is ok and you have a small leak. If the increase follows a non-linear curve, then the sensor is not reliable.

But I can tell you already: the sensor is not that much reliable. Use a Bosch BME280, which is reliable. I use a Xiaomi temperature/humidity sensor and it is reliable down to 1% humidity, which I achieve after drying the silica. Then it goes linearly up to 10-12% in three months time, when I dry the silica again. I use an IKEA Samla box with foam tape between lid and box.

Also, the silica that way does not expose the max surface: make rather some pouches using kitchen paper and put them vertical, or the upper layers will shield the lower beads.

For info, you may already know it: nylon filament must be dried in the oven upon opening, it's not dry enough straight out of the factory. Also, a dry box will not suck out humidity in any reasonable time, so it's only useful to maintain dryness.


Alright, there were some good suggestion in the comments, and I gave them a look, but I think I solved this myself.

I think, after a lot more testing, that it was some combination of stratification and a faulty sensor. I managed to find four more DHT22s with which I could test the box, and found that the sensor I had been using consistently read ~16% RH where others would read 6-9%. Additionally, I did some tests with the sensor at different heights in the box, which were somewhat inconclusive, so I attached a fan to the sensor's power to circulate the air inside the box, to avoid stratification if that was occurring.

Across my three best-agreeing sensors, with the fan (which I'm still not certain provides a major benefit, but certainly makes sense if stratification is assumed to be a problem), I'm now seeing values that seem to be around 6% to 8% RH depending on the sensor (this amount of difference between them is to be expected per the datasheet, 6% vs 16% is not), much more in line with the performance I would expect from what I have seen online. Time to finally unwrap the nylon, I guess.

I won't mark this answer as the solution until I see this performance maintained and get good nylon prints out of it, but for now, I think this is probably most of the way to an answer.

  • $\begingroup$ If you want to run a serious test: expose the sensors to a deliberate high humidity environment (e.g. water vapor over a boiling pot). If the sensor still reads 16%, it is totally shot. $\endgroup$
    – Trish
    Mar 28, 2022 at 9:32
  • 1
    $\begingroup$ It reads similarly to other sensors at higher humidities (like the ~30% RH it is in the room right now), but seems not to bottom out at around 16% in the box where other sensors read below 10% - it seems like it might only act strangely at low humidities? $\endgroup$ Mar 29, 2022 at 0:44

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .