#avr Logs

Aug 06 2021

07:38 AM specing_ is now known as specing
08:57 AM PublicWiFi is now known as PubicWiFi
10:07 AM PubicWiFi is now known as PublicWiFi
03:19 PM twnqx: hm. step-up and step down regulators i high current capacity are easy to find, but inverting ones seem badly handicapped
03:20 PM qu1j0t3: yeah probably because they have to use on chip capacitors?
03:20 PM twnqx: almost to the point of "just use an isolated flyback and tie its high potential to ground" bad
03:20 PM qu1j0t3: you might have to build discrete :)
03:25 PM twnqx: i only want to have +-16V from li-ion, to use low-noise LDOs to +-15V post regulation
03:55 PM exp: at what current?
03:55 PM Phantom: twnqx: why do you need dual supply?
03:55 PM twnqx: opamps
03:55 PM qu1j0t3: you can't just float a ground?
03:56 PM twnqx: exp: not sure, i think 200mA each will be enough
03:56 PM exp: twnqx: surprised to find you're having problems at that low power
03:56 PM Phantom: why not try to work with single supply?
03:56 PM qu1j0t3: ^
03:57 PM twnqx: a 3.3V microphone is involved, but all references for VCFs i find require positive & negative voltages
03:57 PM exp: it's a common technique to provide a rail at 1/2 your total dc voltage, and pretend that is your ground
03:57 PM qu1j0t3: you can possibly work around that. op amps don't care if it's actually positive to a reference they don't know about
03:57 PM twnqx: though i was contemplating the option to either shift up the mic output voltage and use single 30V, or to use something like 32V step-up and +-15V ldo to a virtual groud
03:57 PM qu1j0t3: yes what exp said
03:58 PM exp: i've not used a 3.3v microphone before, so that's harder to know
03:58 PM exp: but typically you split the AC signal from the DC bias
03:58 PM exp: and the capacitor you use to do that doesn't care what the DC voltage on the other side is
03:58 PM twnqx: mems micrphones are a bit different
03:59 PM exp: well either way, 'ground' is just a potential
03:59 PM exp: and 15v is to 0,30v as 0 is to -15,15v
03:59 PM exp: and unless your microphone needs to reference some other part of your circuit, or you have particular noise or temperature sensitivity requirements, you're probably fine rail splitting (imho)
03:59 PM exp: (not an ee)
03:59 PM twnqx: sure, but in the end i want to feed tit to a D/A and into the digital domain
04:00 PM exp: well that's just fine, you can remove a dc offset in about a bajillion ways
04:00 PM exp: and you might find that it helps with linearity
04:00 PM exp: anyway it looks like -16v sits in an uncomfortable valley
04:01 PM twnqx: yeah...
04:01 PM exp: most low power dcdcs seem to 'max' out at -15v or sit at -17 or above
04:01 PM exp: at that sort of power you can easily just do this with isolated supplies unless you need an ultra small solution
04:03 PM twnqx: MAX17577 / MAX17578 seem good enough
04:04 PM exp: i've never come across 'Himalaya' converters before, to google!
04:05 PM exp: bah just a marketing term unfortunatly
04:16 PM twnqx: actially... too high minimum input voltage, need 3.3-4.2V :(
04:30 PM twnqx: though of course i could just step up to 16V and invert from there. so many options
04:50 PM Phantom: twnqx: and why +/-15V? is there a reason to go that high?
04:50 PM Phantom: or are you going for that just for opamps?
04:51 PM twnqx: reference voltage for the lm13700
04:51 PM twnqx: (not really opamp, but close enough)
04:52 PM Phantom: and no alternative to that?
04:52 PM twnqx: well, it will work with less
04:54 PM Phantom: and that part will be used for what?
04:55 PM twnqx: voltage controlled filter
04:55 PM Phantom: so a bit hard so substitute
04:56 PM twnqx: ... the mic application note has an interesting solution for +-9V from battery
04:57 PM twnqx: looks like sepic + inverting sepic at once, regulating only the positive side which should make the negative stable at the same time
04:57 PM twnqx: (invensense AN-0284)
04:58 PM twnqx: hm or is this nonisolting flyback...
05:00 PM twnqx: ah, sepic-cuk, ok
05:03 PM Phantom: there is lots of ways
05:04 PM Phantom: a single smps would require a transformer
05:04 PM Phantom: a double smps can be a good alternative
05:08 PM twnqx: that poor regulator.. now that's an interesting design
05:09 PM twnqx: drop a transistor as a kind of LDO before, and let it regulate something completely unrelated
05:10 PM qu1j0t3: yeah, i'm doing something like this now
05:10 PM qu1j0t3: prototype uses op amp and pass transistor but i think an adj reg can drop in as well (not tested yet)
05:11 PM qu1j0t3: to be specific it's going to regulate ~1kV
07:36 PM specing_ is now known as specing