#garfield Logs

Jan 27 2025

02:51 PM aandrew: https://www.reddit.com/r/EngineeringPorn/comments/1i90c1q/i_made_a_cnc_to_coil_electromagnets/?ref=share&ref_source=link is actually a better link
05:14 PM aandrew: rue_mohr: do you have any links/magic for computing torroid power transformer turns?
05:16 PM aandrew: I found https://miguelvaca.github.io/vk3cpu/toroid.html which is nice but I don't think it'll work for the PC40 material
05:25 PM aandrew: I can find some data on the material but not the dimensions PC40 T118x20x55
05:37 PM aandrew: if I'm not completely wrong (big if) it seems like the relative permeability is the same as initial permeability at low frequencies, which 60Hz definitely is
05:52 PM aandrew: hm, LLMs seem to tell me for 118mm od, 80mm id, 20mm thickness PC40, operating at 100C I'd want around 250 turns at 240V
05:52 PM aandrew: that's a lot of damn turns, I'm gonna need a much bigger torroid if I want 500VA or 1kVA out of it
05:54 PM aandrew: https://ibb.co/nkdH3Bf I mean it LOOKS plausible and the flux density is correct looking at PC40 material ds and cross sectional area looks correct
06:00 PM aandrew: heh it even computed whether the coils should fit on the toroid and the length of wire for primary and secondary, but it doesn't appear to really have taken into account the diameter of the wire itself as you wrap turns on turns
06:12 PM aandrew: after correcting itself (I'm too lazy to check its math) it came up with a more reasonable answer and even suggested the number of "layers" for the primary and secondary
06:13 PM aandrew: ~70m of 14AWG and 7m of 10AWG
06:31 PM rue_mohr: aandrew, no toroid magic
06:32 PM rue_mohr: 1 volt/turn is crazy agressive
06:32 PM rue_mohr: often used to 2turns/volt
06:33 PM rue_mohr: llm's are wrong in amazing ways
06:33 PM rue_mohr: always ask, "are you sure" after its answer and see what it says
07:33 PM rue_mohr: aandrew, ok, so I got a reply from doteasy, they dont want to offer a 'please dont leave us' deal, ok..
07:39 PM aandrew: this LLM keeps making plausible but nasty errors
07:39 PM aandrew: it told me 14AWG for the primary is good since 1kVA / 240V is only like 4.2A, and then recommended 10AWG for the secondary since the current is 10x that
07:39 PM aandrew: however 18AWG is still PLENTY of margin for the primary, and then I asked it about the secondary and it is like ... oh, use 1/0
07:39 PM aandrew: wtf
07:40 PM aandrew: and then I told it "ok, if i reduce my operating temperature from 100C to 60C how many turns should I use on the primary, expecting the number to go down since the flux density can be increased... it comes back with 2x the turns
07:40 PM aandrew: so I ask it wtf and it's like "my bad"
07:41 PM aandrew: now I'm asking it how much wire I'd need and it's saying like 150m for the primary and I'm asking it how
07:41 PM aandrew: if my toroid OD is 118mm and ID is 80mm and it's 20mm thick, ONE turn should be 2* (118-80)/2 + 2* 20mm shouldn't it?
07:42 PM aandrew: put wire through toroid, that's 20mm. then bend it over the side, that's half of the difference of the OD/ID diameters. bend it down over the outside, that's another 20mm, and bend it toward the hole is another half the difference of the OD/ID diameters, no?
07:43 PM Tom_L: plus a little since the wire isn't exactly up against the toroid
07:44 PM aandrew: sure, but that's slop I'd add another meter or whatever just because nothing's that easy
07:44 PM Tom_L: and the center of the wire is larger
07:45 PM Tom_L: https://www.coilwindingmachines.eu/engineers_corner/toroidal_calculations.html
07:45 PM Tom_L: that's for an auto winder but may give you the length
07:48 PM aandrew: but to me, a single turn for that toroid would be ~116mm (20 + 38 + 20 + 38)... the LLM's saying no, it's 312mm because the wire's following the "mean circumference" of the toroid which is pi * (OD - ID)/2
07:48 PM aandrew: that's a difference of almost 100%
07:50 PM Tom_L: https://coil32.net/online-calculators/ferrite-torroid-calculator.html
07:51 PM aandrew: "theoretical outside finished diameter": 118mm
07:51 PM aandrew: fuck you, LLM. ;-)
07:51 PM Tom_L: what's the wire size?
07:51 PM aandrew: 18AWG so roughly 1mm
07:53 PM aandrew: er wait
07:53 PM aandrew: it's saying finished dia is 118mm but length per turn is 78mm
07:53 PM aandrew: I wonder what that is
07:54 PM Tom_L: sounds like you need a better calculator
07:54 PM Tom_L: but i've never wound one
08:02 PM aandrew: hm, is (OD-ID)/2 the same as (OD/2-ID/2)
08:04 PM aandrew: with easy numbers OD=100 and ID=10 that would be (100-10)/2 and (100/2-10/2) so no
08:04 PM aandrew: that's one mistake
08:05 PM aandrew: so OD=118 means OR is 59 and ID=80 means IR is 40, which makes the distance across the toroid the winding makes from outside to inside or back 20
08:05 PM aandrew: which makes sense
08:05 PM aandrew: so now a single winding would be 4x20 (the cross-sectional perimeter of the toroid) or 80mm whihc is a lot closer to that 78mm "single turn length" the calculator gave
08:07 PM aandrew: so 350 turns at 80mm/turn is 28m of 18AWG, which matches your calculator link
08:08 PM aandrew: and that is only 100 feet which is a lot more reasonable
08:08 PM Tom_L: i was watching those auto winders out of curiosity.. kinda cool
08:08 PM Tom_L: must be a fairly hefty toroid you're making
08:09 PM aandrew: 1kVA yeah
08:09 PM aandrew: and I mean a toroid isn't really awesome for this since it's just 60Hz and there'll be a lot of high frequency hash
08:10 PM aandrew: but it'll work
08:10 PM aandrew: $90 of shit on amazon would build it
08:11 PM aandrew: $25 for the ferrite, and $25 ea for 200ft of 18AWG and 15ft of 6AWG
08:12 PM Tom_L: would be nice if the wire spool fit thru the toroid hole
08:13 PM aandrew: it would be very nice
08:13 PM Tom_L: pita otherwise
08:14 PM aandrew: 175 turns, then out to a center tap, then another 175 turns
08:14 PM Tom_L: so if it doesn't, wind the wire on a 6" flat stick then wind it
08:14 PM aandrew: I was trying to fidn something on aliexpress and they have some nice transformers but unfortunately none that are 120-0-120 and either 0-24 or 12-0-12
08:14 PM aandrew: but my god cheap
08:14 PM aandrew: $50 with free shipping
08:15 PM rue_mohr: you could probably ask for a custom one in the notes
08:15 PM aandrew: I'm gonna run the calculations again just to make sure and bounce 'em off rue_mohr
08:15 PM rue_mohr: hah
08:16 PM aandrew: I did check that rue_mohr -- all the places that will do custom want $$$$ for shipping
08:16 PM rue_mohr: I'm not good with magnetics!
08:16 PM Tom_L: well don't ask me... i don't know nuthin bout it
08:16 PM rue_mohr: yea, ok, that makes sense
08:17 PM rue_mohr: custom = DHL shipping
08:17 PM rue_mohr: = shipping is 800% product cost
08:17 PM aandrew: PC40 ferrite seems to have mu-i of 2300 at 25C and saturation density of about 450mT at 60C. I don't plan on running it hot but want to design there
08:17 PM rue_mohr: uh, if 60Hz use iron
08:18 PM rue_mohr: wait... are you making a UPS?
08:18 PM aandrew: yeah I know, but I don't want to mess with E-I plates and shit, I'll just use a toroid for ease and availability and if it works I'll optimize
08:18 PM rue_mohr: ARE YOU BUILDING A UPS?
08:19 PM * Tom_L plugs his ears
08:19 PM rue_mohr: I see evasion of "yes"
08:19 PM aandrew: cross sectional area would be 20mm * 20mm (approx) so 40cm^2
08:20 PM aandrew: no evasion, thinking
08:20 PM aandrew: not necessarily a UPS but could work as such
08:20 PM rue_mohr: mhm
08:20 PM rue_mohr: I'm gonna write X
08:20 PM rue_mohr: send a bill
08:21 PM rue_mohr: $10 ea for spam accounts detected and reported
08:21 PM rue_mohr: I have 41 of them here
08:22 PM Tom_L: maybe you could ask for a ride in his rocket
08:22 PM rue_mohr: no...
08:22 PM rue_mohr: they explode
08:24 PM aandrew: N = (240 * 10^8)/(4.44 * 60 * 4500 * 40cm^2)... that comes out to 500 turns
08:24 PM Tom_L: you just said 350 so which is it?
08:24 PM aandrew: so reducing the turns to reduce the flux density... 30% margin (no idea how much I need) would be 350 turns on the primary
08:25 PM aandrew: so it seems like if you badger the LLM enough it does come out with a reasonable answer but the question becomes how much to badger
08:25 PM aandrew: not bad for getting to a ballpark and then running the numbers and doing your own smell tests
08:25 PM aandrew: that 10AWG recommendation for the secondary was kind of funny though
08:26 PM aandrew: "42A secondary current with 10AWG is not safe. sorry about that"
08:27 PM rue_mohr: 1kva
08:27 PM aandrew: rue_mohr: I'm starting out with a stupid simple 24 -> 240 inverter yes but my next step is to switch this out and instead of a stupid 60Hz inverter I want to put mosfets on both sides and see if I can't do it bidirectionally at like 20kHz
08:27 PM rue_mohr: whats the wattage on a normal microwave... 800w?
08:28 PM rue_mohr: 15.36kHz
08:28 PM aandrew: 24VDC -> 400VDC and then a 60Hz sine to get 240VAC but I'm trying to avoid the 60Hz transformer and do dual H bridge to get 120-0-120
08:29 PM rue_mohr: yea
08:29 PM aandrew: depends on the microwave but yeah 800-1500W
08:29 PM rue_mohr: two bridges
08:29 PM rue_mohr: its only 15A
08:29 PM rue_mohr: ...
08:29 PM rue_mohr: 4A...
08:29 PM rue_mohr: huh
08:30 PM rue_mohr: dont even need IGBTs for that
08:30 PM aandrew: I also wanted to experiment with a proportional resonant controller which owuld be COOL but I don't think you can do that bidirectionally
08:30 PM rue_mohr: just boost and modulate
08:30 PM rue_mohr: get a scrap VFD and re-brain it
08:30 PM aandrew: I was also toying with the idea of GaN mosfets because I can switch way faster which means my harmonic currents are way higher meaning smaller/lighter/cheaper filters
08:31 PM aandrew: yeah I spent 13y in power electronics designing them, I have some access to old stuff
08:31 PM * rue_mohr gets a bowl of popcorn...
08:31 PM aandrew: might be eating a lot of popcorn, I have to stop typing and make dinner for the kids
08:31 PM rue_mohr: you know you just put a series inductor on the input and have the synchronous rectifier short the line to generate the boost?
08:32 PM aandrew: I wanted to stop 2h ago but the goddamned LLM was toying with me
08:32 PM aandrew: the input is a battery
08:32 PM rue_mohr: same
08:32 PM aandrew: battery <switch> tx line
08:33 PM aandrew: so putting energy back into the battery has to go through the tx again
08:33 PM rue_mohr: who said the power block of a vfd had to be fondling ac? :]
08:33 PM aandrew: no I know; we've done common bus systems before
08:33 PM rue_mohr: the series inductor turns into a buck when its running backwards
08:33 PM rue_mohr: :]
08:33 PM aandrew: right -- but I hav eto go now or I risk the kids killing and eating me
08:34 PM aandrew: back later, and I can still do the mail/web if you want. lmk
08:34 PM rue_mohr: I have a rom image with the sine wave stored in pwm
08:34 PM rue_mohr: in the bits
08:34 PM rue_mohr: its seperated into H and L drive channels
08:34 PM rue_mohr: you just clock the address line and it does everythign for you
08:35 PM rue_mohr: hehe
08:35 PM rue_mohr: <projects I have not had time for yet>
08:56 PM rue_shop2: hahhahaha
08:56 PM rue_shop2: I entertained myself by designing a 4-20mA intercept/manual control thing for ddc control,
08:56 PM rue_shop2: I forgot all the stuff at work is 0-10V
08:56 PM rue_shop2: hahahah
11:03 PM aandrew: holy cow everything is freaking PC40 MnZn
11:11 PM rue_mohr: ?
11:11 PM rue_mohr: molybellium Zinc?
11:12 PM rue_mohr: magneese zink
11:12 PM rue_mohr: c
11:12 PM rue_mohr: lskjdflksjflksjfd
11:53 PM aandrew: manganese
11:53 PM aandrew: same black shit that clogs up my prefilter on the well pump