#avr Logs

Nov 14 2021

12:00 AM rue_mohr: my next challange is to build a more optimal inductor
12:08 AM rue_mohr: I know I can get to ~1V of output target
12:08 AM Thrashbarg: I'm lost... you want a separate inductor to the transformer?
12:08 AM rue_mohr: a flyback transformer is a linked inductor
12:08 AM Thrashbarg: ehh not quite
12:08 AM rue_mohr: the core is charged by one winding, and discharged by the other
12:08 AM rue_mohr: do tell
12:08 AM rue_mohr: for this project, I can ignore isolation
12:08 AM Thrashbarg: there's a terminology mismatch here. To me an inductor is a core with a single winding, and a transformer is a core with more than one winding
12:10 AM Thrashbarg: you're putting 12V into it and getting a bit less than 12V out of it?
12:10 AM rue_mohr: a lot of work got me up to 11
12:10 AM rue_mohr: I wonder if I can tune that out with the duty
12:10 AM Thrashbarg: suppose you ground the capacitor and load
12:11 AM rue_mohr: no, I have to only pull the output energy from the inductor
12:11 AM rue_mohr: the supply cant be involved
12:11 AM rue_mohr: because that winding will not be used to charge the inductor when its a flyback
12:11 AM rue_mohr: the magnetic field will be set up with another winding, via a different votlage
12:12 AM Thrashbarg: yea I see. You won't be able to reach unity with that arrangement. There will be losses in the diode and inductor
12:12 AM rue_mohr: yep
12:12 AM rue_mohr: hey, I know, I have some 19V adatpers that can output more then the current I need
12:13 AM rue_mohr: I can dial it down with the duty,
12:13 AM Thrashbarg: yea that'd help
12:13 AM rue_mohr: my main goal is to make sure that the winding can produce the right output
12:13 AM rue_mohr: I need 20w/12V
12:14 AM rue_mohr: if I can get that, I can play with the core, tune it.
12:16 AM Thrashbarg: sounds like a plan
12:16 AM Thrashbarg: what switching driver are you using btw?
12:17 AM rue_mohr: controller?
12:17 AM Thrashbarg: yeah
12:17 AM rue_mohr: at this rate I'll use a 555, I dont care
12:17 AM Thrashbarg: hehe
12:18 AM rue_mohr: I learned to not stop them, the continious start and stop really f****s with the duty
12:18 AM rue_mohr: you use pin 7 as the output and ground it to stop
12:18 AM rue_mohr: then the duty isn't thrown all out
12:28 AM rue_mohr: I'm using a signal generator for testing, once I know the duty and whatever I can use anything, I'v come to see all the controllers as prettymuch the same thing
12:28 AM rue_mohr: the LM10 is cool, I got a few and havn't had a chance to use them for anything yet
01:08 AM qu1j0t3: _bed is a good idea
01:09 AM qu1j0t3: rue_mohr: Your experiments are pretty interesting and useful
01:10 AM qu1j0t3: rue_mohr: i've just been messing with the classic ccfl circuit for high voltages, not winding my own, but baxendall self resonant is pretty temperamental
02:32 AM rue_mohr: I got to 24w out!
02:32 AM rue_mohr: yay
02:32 AM rue_mohr: at 89.8% efficient!
04:54 AM vishwin60 is now known as vishwin
05:26 AM system_error: once I know the duty and whatever I can use anything, <- beware of duty vs flyback, it only works for CCM. If load ever disconnected...
06:47 AM exp: thinking about my plan from yesterday, it won't work, which is a shame
06:47 AM * exp sighs
06:53 AM rue_mohr: yes, thats why I like pushpull converters
06:58 AM system_error: they're very different in nature...
07:08 AM rue_mohr: the downside to pushpull is getting the inductance high enough so the magnetization currents aren't stupid
08:48 AM system_error: rue_mohr> guess key downside of pushpull (esp vs flyback) it can't do stabilized out vs wide input range reasonably.
09:29 AM qu1j0t3: rue_mohr: congrats. my oscillator is only uh 30-40% and i can't figure out why ;-)
09:29 AM qu1j0t3: exp: Oh, okay, so you found the catch? :/
09:42 AM exp: qu1j0t3: the 100mV is referenced to vin explicitly, so i'd have to do the whole thing backwards and that loses most of the protection
09:42 AM qu1j0t3: ah
09:42 AM exp: best option now becomes adding another small nmos so i can reset the interface explicitly
09:43 AM exp: but quite a few caveats along with that, another watt or so of losses and the need to detect the fault and reset it autonomously
09:45 AM exp: qu1j0t3: one thing i am considering, i have some SiR626DPs in stock which have rdson of 1.4-1.7mohm which also happens to be the exact (lol) sense resistance required to trigger overcurrent comparison with the 4368
09:46 AM exp: some intelligent gate management there could get rid of the extra thermal losses and allow me to optimise thresholds a little better
12:15 PM exp: qu1j0t3: yeah looks like based on the spice model you were right to be cautious about repetitive turn-ons
12:16 PM qu1j0t3: for reverse overcurrent?
12:16 PM exp: yeah, the reset comparator takes precedence over the overcurrent comparator
12:16 PM exp: so on the spice model it will begin turning on the mosfets despite an apparent constant overcurrent
12:17 PM exp: it does signal a fault, but on the spice model it's quite a sharp edge so i'll have to check i can trigger off it in a real circuit
02:50 PM rue_bed2: whats it protecting?
03:04 PM exp: rue_bed2: a bidirectional 12v bus