Design & Order on EasyEDA, PCB+SMT $10 Off: https://easyeda.com/editor
Get coupon & Join JLC&EDA Group: https://jlcpcb.com/EDA
Help me fund a replacement for this miner fan which got damaged in transit and exploded on first use 🙂
Interested in my new garden workshop? Follow the entire build on Julian's Shednanigans:
https://www.youtube.com/channel/UCXfDjPehpC7B7lW2JFxeS4w
Subscribed to my 2nd channel? Watch all my 3 minute videos on Julian's Shorts: https://www.youtube.com/channel/UCeewzdnwcY5Q6gcbnZKIY8g
Join me on Odysee/LBRY: https://lbry.tv/$/invite/ @julian256:d
Octopus Energy referral link: share.octopus.energy/aqua-birch-918
(We each get £50)
#ElectronicsCreators

Just got this fan on ebay, put it in my ant miner and it exploded and all the blades fell off here. They are uh, i'd written on them. Actually, what the fan is because, when this fan fits on the miner on one end, um it's the way around that you can't see the label. So what i'm doing is writing on the blades in sharpie, and i don't think the sharpie is going to affect the balance of the blade.

I should have filmed this actually because i had a feeling it was going to explode because i could see one of the blades was cracked probably happened: um in transport, but yeah interesting, 2.7 amp. This is a four wire, so it's got pwm and also taco. Of course it spins now, but uh doesn't blow any air. These are 120 millimeter outer diameter square and 38 millimeter across that dimension, they're quite expensive, the cheapest ones.

I found uh uk seller doing them just over 10 pounds each. This one came from a chinese seller. So, of course, it had a longer shipping trial and yeah. These things get broken, so i was planning to take this apart and have a look inside there and i've noticed behind this label.

There is if i can get that off um a very tiny circlip in there. I don't know if you can see it yeah, you probably see it moving around there. So if i can get the circlip off, then that should come off and we can take a look at the electronics. I don't suppose i need to worry too much about damaging this circlip, so let's just get it up in whatever way i can, even if that means breaking the circuit, it doesn't want to come off.

Oh, it's come off and it's well and truly bent, but that's should mean that this comes off yep. It does uh. Oh can't see, there's a pcb there see if i can get inside that. Well, i've got out two bearings which just sit in plastic: um yeah.

Once it's in that recess, once it's in that recess, both plastic and obviously this shaft here sits in the inside of the bearing, as i say, there's a spring there, there was a circlip, that's all bent now. This is just a very small printed circuit board. I can see on here a couple of components and there's a hall effect chip there, three pin device - and here there's an electrolytic capacitor sort of sitting up inside the coil area. I would have thought that gets quite well.

I was going to say hot, but then i suppose this thing's blowing cold air, so perhaps it doesn't, but i can't see how this board is anchored into the plastic surround. It doesn't look like there are any fixings on this side. Just some molding marks there. So i'm kind of thinking i'll just push that hard until it comes out uh.

I think this disc. This washer here is another circlip holding this piece onto this inner, because this inner plastic tube comes all the way from the back here. I believe there's a step in it. There actually um, but i think that circlip has to come out before this assembly is going to come out.

I will coax it out right. This doesn't look like it's going to be coaxed, so i think it's going to have to be cut. I'm going to cut away this center piece, which holds the bearing well that's, coming away very easily, it's very soft. Can i get that what looks like a circle about i've just started, levering it up with these needle nose, offset pliers and something moved, but it's difficult to do because i can't get them under there.
Something's slid upwards, but i can't replicate that at the moment. I think it's coming out. I can see what this in here is now, because you can just see the laminations on this. I think that is slipping off a plastic tube uh, but there's a grip between the sort of laminated surface of this.

These are the same laminations that are on the outside here, just running through to the inside uh, but that will slide off. I do believe: okay, very large, screwdriver sort of last resort stuff there we are it's off so now we can see. Oh, what's that is that some glue? Yes, that is some glue uh, oh quite a lot of electronics on there. Let's have a closer look now i can't remember, which is taco and which is pwm.

I did read the other day that pwm either yellow or blue um is 5 volt. I believe 12 volts course comes in to power. The motor that capacitor looks suspiciously close to the 12 volts coming in so could it be directly across them? Quite possibly the hall effect sensor. Is there those three pins, i'm not sure whether it's worth getting all the numbers for these chips, they're, probably highly proprietary for fan, control, um.

The other thing i read was that the taco signal is open collector, so you can externally pull that up to whatever you want to read it. I don't know whether it's positive going or negative going, but it will be presumably a frequency derived from the rotational frequency of the fan. The pwm, of course, will be a 0 to 100 or possibly a lower range than that signal to drive the um power level. To the fan and the the ratio of uh fan, frequency, uh versus level of power is going to depend on all sorts of factors like density of the air that these blades are having to whip through uh temperature, probably to some extent the freeness of the bearing In the center all that sort of stuff well, two of these chips - these two are a2031.

I don't think they are mosfets, because it you don't have the characteristic four pins on one side connected, although the four pins this side do appear to be connected, but not the same that side uh, i suppose they could be mosfets, but i can't get any um Data on those, if you put a2031 into google, you just get lots of stuff about apple airpods. This chip - here, i think, has been painted. So i can't get data for that. But yes, what we're essentially doing is taking a five volt pwm signal and using it to vary the uh drive to the motor and then the attacker is probably very simple.

Actually, it's probably just a transistor, possibly even taking a signal from the hall effect and running it back out to the taco output with this open collector connection. So that's what happens if your fan arrives with a small crack in one of the blades it breaks off, because these things do rotate very, very fast and then, of course, that snaps all the other blades off, and you end up with that - a central hub with Seven stubs on it and no blades, but that's what's inside one of these four pin speed, controllable fans is interesting. The uh magnet running around here appears to just be a piece of magnetic plastic uh like that stuff that you get um well like fridge. Magnets are made of yeah there, it is i've pulled the magnet out and it is just made of that sort of plasti magnetized plastic material.
Just pushed in to a metal former interesting.

By Julian

Youtuber, shed dweller, solar charge controller aficionado

14 thoughts on “Exploded Fan”
  1. Avataaar/Circle Created with python_avatars ԱӍβЯƐ says:

    Wow…a chinese knockoff Delta fan, no wonder why it blew up. Was not aware they made knock off Delta's this I gotta look into.

  2. Avataaar/Circle Created with python_avatars Vader Whoop says:

    Spotted you on a particularly slow crypto channel.
    Try these for a bit of substance:
    Stansberry Research
    Benjamin Cowen
    Chico Crypto

  3. Avataaar/Circle Created with python_avatars hi-tech-guy-18 says:

    Hope it didn't run to long without the fan

    Try and Get Water cooling block for Asic's – would fix the noise issue

    or say bye to Blower fans wwith 3M Novec Engineered Fluids – Immersion server cooling

  4. Avataaar/Circle Created with python_avatars The Solderman says:

    I'm not a fan of this fan….. (Pun intended). Why should so many fan blades break away? Have they been bent to often at the resonance frequency? Did someone overspeed the fan by applying compressed air or overvoltage? For cooling an antminer you would expect some airflow resistance. In case the fan was operated in an open airflow situation, without any resistance the fan might went to an overspeed.

  5. Avataaar/Circle Created with python_avatars Roland Elliott says:

    Well if you see its already cracked then of course its going to come apart at some point, makes me wonder sometimes how people can expect any other result.

  6. Avataaar/Circle Created with python_avatars Peter S says:

    It would be quite interesting to see that magnet under a sheet of paper with some iron filings or iron beach sand on top to show the lines of force.

  7. Avataaar/Circle Created with python_avatars Xfixiateher says:

    FYI;
    Those are mosfets (twin 8pin packages) however they are DUAL N & P channel fets similar to FDS8958A-F085 or IRF7309

    The 8958a are found in original Delta electronics 120mm fans (AFB1212SHE)

  8. Avataaar/Circle Created with python_avatars Helmuth Schultes says:

    I would have expected that the A2031 is a Mosfet so in typical form 4 pins Drain one side, the other side 3pins Source and one pin Gate. Am on my phone at the moment, from memory LCSC had some Mosfets of similar number to A2031 in their range, can not immediately confirm.
    Other than that A2031 as semiconductor seems abreviation of 2SA2031 PNP power transistor but not likely this device as that is a high current device in 3pin TO3 package from memory.. the three 8 pin chips are likely identical devices as three phase motor winfing drivers, but I see no devices there that take the PWM and derives a three phase drive signal. But such is needed to drive the windings with PWM signals at 120 deg phase relationship.

  9. Avataaar/Circle Created with python_avatars Ben Carrington says:

    You can get those large delta fans from server recycling sellers, large HP and sell servers had multiples of them. You will need to solder the connector as they have a hotswap one from factory

  10. Avataaar/Circle Created with python_avatars yeliab814 says:

    if you are competent with cyanoacrylate glue, you can glue those blades back on. that’s because polycarbonate breaks are brittle clean and thus line up well; and CA glue binds superbly to PC. i’ve fixed many ultra high speed blades this way – never had a failure. try it you may be surprised.

  11. Avataaar/Circle Created with python_avatars ElmerFuddGun says:

    The fan speed will NOT be affected by the air resistance, etc. The PWM input determines the max speed. The stationary motor coils are pulsed to determine the speed by the ESC (Electronic Speed Controller). Light resistance changes will only affect the current used.

  12. Avataaar/Circle Created with python_avatars PaulaXism says:

    A2031.. it's an R1230 series PWM/FVM step-down DC/DC converter with synchrnous rectifier. Output voltage 2.0V. Optional function 800kHz, fixed output voltage

  13. Avataaar/Circle Created with python_avatars David D says:

    Judging from the amps 2.7, I am guessing this was one of those super high airflow ones that runs at 10k rpm, and are scary as hell to hold in your hand when running at full tilt. I have one around the house somewhere, but never put it in anything as they sound like a vacuum clearer when running. They must explode in spectacular fasion though!

  14. Avataaar/Circle Created with python_avatars ljmike1204 says:

    I had one explode in a moving head (disco light) one of the stabelizing legs was not fixed so the fan could move in the schroud and when the moving head tilted it went BANG and plestic parts flying

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.