[Enhancement] Suggest Investigating Arctic's PWM PST Fan Line

[lydia]

I'm working on basically the same project and built a prototype using one of Arctic's PWM PST fans and, of possible interest to your project, the PST stands for "PWM Sharing Technology," basically an integrated option for daisy chaining fans. Cheaper than the Noctua option, though with 25% lower advertised delivered CFM, similarly quiet, manufacturer provided bulk discount in groups of 5, and it'd knock a part off your BOM (the PWM splitter) to boot.

If you find yourself not running your fans at 100% max speed all the time it might just be a universal upgrade.

For the record I had to do a double take on the user name when I came across this repo. Was looking around to see if anyone else thought they could pull this off much cheaper than CAK. I'm assuming I can keep the keyboard.io model 01 at this point.

[jcorbin]

Hey there, what a fantastic way to reconnect!

So this is an interesting prompt, I'd not really considered using any of the various vendor proprietary solutions for fan chaining. My main concern, other than making the build even more sensitive to supply chain availability, would be needing to reduce the gap between the two fan bodies, so that their integrated chaining can work.

Having not seen Arctic's take on this before your suggestion, I started out thinking it'd be pogo pins requiring the fans to be flush with each other ( e.g. Lian Li Uni Fan ). But looks like these arctics use a short cable that comes with the fan to chain, maybe permanently integrated? Looks like Corsair's RS line also has some sorta similar deal.

Anyhow, for pchepa, or any CAK exhalaron clone really, the filter itself is 180mm, not 120mm, so it's not possible to place the fans flush with each other; tbd how long each vendor's specific daisy chain cable is.

In addition to the 180mm filter_d design parameter, there's also reason to have addition gap internally between the two filter bodies: allowing easy central placement of wire routing to the battery pack. Currently pcehpa adds 8mm radially, so an additional 16mm of diameter for this ( the base_overhang parameter ).

When it comes to speed, there's pretty much only 2 speeds that make (most) sense to run pchepa at: off or full send ; if you want it on, you probably want max CADR ( clean air delivery rate ). Even having a pwm adjustable knob could actually be considered an unnecessary design complication for some. Even at max, at least with noctua fans, noise if fairly innocuous; especially as compared to larger other hepa filters in my experience.

To that end btw, it's fully possible to over volt most pc fans, say to 15v, and run them even faster. This is actually even easier to build for a stationary filter, as a 15v USB-PD trigger will work with way more USB AC adapters, since 12v is an optional mode not commonly implemented, but 15v is quite common. In terms of CFM, at least in one spot check I did early on, the extra 3v got an extra 33% CFM per fan.

Anyhow, more than happy to review / consider / accept pull requests if you end up using pchepa as a basis for such an experiment and find something that works, perhaps with a different smaller diameter hepa filter component.

Also note: I've got some unreleased minor changes to the 12v 18650 lipo battery pack over on a https://github.com/jcorbin/pchepa/tree/testing branch ; have recently been building a few more of these for some folk, and iterating in minutae as supply chain vagaries around that 12v battery holder pcb change.

Keyboard wise, no worries, I considered that one yours a long time ago ; have been using an olkb preonic for years now, and almost bought one of the keyboard.io preonics when it came out since it's stuck so well for me so far :-)

[lydia]

My point about fan speed is that we definitely make tradeoffs between CADR and other concerns, like portability and aesthetics (your case is very cute but I'd wager your choice of mesh wall impedes airflow somewhat), so it's reasonable to examine how we might trade it off against build complexity.

My prototype uses an Arctic P14 140mm fan, the integrated daisy chain cable is about 2" long but you still have the entire length of the other fan's input cable to work with, this should be more than enough to close the gap. The connector isn't proprietary and I would bet good money you could plug another PWM fan in to that daisy chain and it would work fine, the sauce is in how they're doing the split without demolishing the hard edges of the square wave, there's not some other protocol under the hood.

Anyway the prototype is an absolute bodge job held together by elastic from cut up n95 masks that I shoved into the 3D printer closet to see if it could kill some of the stench from drying TPU (it worked extreme well!). I'm going to be working on a rev 2 that is going to look more like yours than the current nightmare in my closet. I'll even try to use your exact filters, the prototype uses one slightly taller and narrower, but lack the nice pre-filter.

How are you measuring change in airflow? Just looking at the RPM feedback pin on the fan?

How accurate is your filter_id measurement? My method involves printing a thin retaining cylinder that sits inside so I have more room to work with on the top and bottom plates on the outside, and it'd be nice to print some of the parts before the filters actually arrive.

[jcorbin]

... we definitely make tradeoffs between CADR and other concerns ... I'd
wager your choice of mesh wall impedes airflow somewhat

Excellent point, I've not rigorously measured airflow, let alone CADR,
difference with/sans mesh wall as yet. For pchepa, the mesh wall isn't just
aesthetic tho, it provides vertical integrity to the entire assembly in
addition to impact protection for the soft filters. While the cover and base
plates do have small grip bumps that hold the filter's plastic rim, without the
mesh walls, the portable version will disassemble itself as soon as it gets
firmly bounced firmly by the shoulder strap.

Since pchepa relies on slicer infill generation for the mesh, we can easily
vary its density to improve airflow as needed. On the other hand, the 60%
tri-hexagon infill that the project uses as it's recommended setting / baked
into the easy mode 3mf file, has proven to function as a useful dust
pre-filter.

My prototype uses an Arctic P14 140mm fan, the integrated daisy chain cable
is about 2" long but you still have the entire length of the other fan's
input cable to work with, this should be more than enough to close the gap.

Excellent, I've ordered 5-pack of Arctic P12 fans, and will give them a go Soon ™️

Anyway the prototype is an absolute bodge job held together by elastic from
cut up n95 masks that I shoved into the 3D printer closet to see if it could
kill some of the stench from drying TPU (it worked extreme well!).

Ahh yeah, when I saw an CAK Exhalaron IRL, was supremely unimpressed with the
"held together by bungee cords" vibe comparatively tbh. Fwiw I managed to avoid
handling or seeing one until after I'd designed and built a first revision of
pchepa. During the design, I only looked at online pictures, and took its
particular choice of filter model / fan size as a starting point.

How are you measuring change in airflow? Just looking at the RPM feedback pin on the fan?

For now I just have a basic anemometer to measure airspeed at the outlet. I do
have a box of electronics kit (adafruit controller, air quality sensor board)
to build a CADR testing rig, but just haven't Gotten Around To It yet ;-)

How accurate is your filter_id measurement? My method involves printing a
thin retaining cylinder that sits inside so I have more room to work with
on the top and bottom plates on the outside, and it'd be nice to print some
of the parts before the filters actually arrive.

Hmm I've not done a fit test to validate my ID measurement yet, as my filter
retention mechanism is instead a series of teardrop-shaped grips around the OD.

Hold on let me re-measure one:

• so the plastic rim is measuring a little over 31, let's call it 31.25
• so if my 180 OD is correct, that's make the ID 117.5, a bit more than the current main branch value of 114
• otoh trying to measure the ID directly is giving circa 116, maybe more like 115.5

Hope that helps, curious to see what you come up with!

[jcorbin]

Update:

• got a set of Arctic P12 pwm/pst fans ; they work Fine Actually
• mechanical fit is overall better / less bulky than the noctua + splitter deal: no cable sleeves means less bulk to manage
• from some brief testing ( with an iphone decibel meter and a simple anemometer ):
  • noise is same: 6db lift from a 38db floor up to 44db while running
  • may be slightly less capable of pulling air through this filter than the noctuas:
    a. arctic outlet speed at the grill box measured 1.3m/s when on top the filter vs 1.8m/s when lifted up into free air
    b. noctua outlet speed at the grill box measured 1.5m/s when on top the filter vs 2.0m/s when lifted up into free air

However measuring airspeed with an anemometer is both problematic to get consistent test results from, and also only indirectly related to things like cfm or cadr performance.

Overall, I think it's totally worth to update the pchepa build guide with these as a cost cutting alternate. And then also prioritize finally doing some cadr testing :-)

[lydia]

sick!

I just got the last part for my setup last night and assembled and deployed this morning, intending for a 24/7 setup for a single small room. Works like a charm so far.

Since it doesn't need to be mobile I've cut out most of the other parts and just using a wall wart to 12v DC converter which came with the fan connector already attached. For like $8. Obviously overpriced for what it is but still cheap compared to the combined controller, pd trigger, and USB wall wart, especially if you're only making one, and considering that it's nearly impossible to find a low wattage 15v or 12v wall wart... might make a good stationary kit.

Given their response to over-volting I bet you could do an inline potentiometer and just drop the voltage and it would be good enough for two of these daisy chained.