Asmyldof

Lead reacts with several common acids, among which acetic acid, very common in nature, to create decently to well soluble lead salts. Outside of that there's pressurised and/or thermal conversion of several lead oxides. The story that lead can be dumped in nature with no effects is an industry tale akin to claims early this century sponsored by the Koch brothers that CO2 has nuttin' to do with nuttin'

@NickAlexeev Not to my knowledge, we just say "I need a motor with this size"

Or mostly components, I think, since I find it outrageously easy to acquire leaded soldering materials

@NickAlexeev He's talking about the other way around. And components

Reworkable, just recoating is an issue

*cough * parylene

@PlasmaHH They comply and do the work and get to send invoices to California

Fohmfedompompom

But I can't be arsed to get into another entitled comment moaning

I really don't think it's a great idea for this person to be building his/her own SMPS...

@ThePhoton As long as you account for coupling effects and remember to double DCR there usually isn't a lot stopping you

What are you actually trying to ask affirmation for? :-P

Possibly width/length constraints

@ThePhoton Height constraints pop into my mind right away

@ThePhoton If you're talking about custom parts, the savings aren't in not testing, the savings are in being able to use more edge dies and being able to route failed dies to the custom part. Just asking them not to test isn't going to yield anything interesting, you want them to test and verify that you get the dies with the crappy timers.

If I owned a company making a chip I'd make damn sure everything was always tested, because if a customer says he won't use it today, tomorrow you'll be on the phone non-stop about that feature you didn't include or test

Or possibly not

@ThePhoton Depends on the production scale. Between 5k and 5M probably, depending on die size

@AnindoGhosh When Microchip bought Atmel I was genuinely annoyed. When AD bought LT I was a bit sad. NXP, however, could do with some chlorine

I... don't care

Byes

Bed now

**FET has not been selected. Default Circuitlab device, so likely horrendous altogether

If you'd put that in a commercial device you'd put a crowbar transistor on C1 that triggers on 50% rail voltage or non-presence of AC for 5 cycles or such

Works just as fine when the BAT is the bridge output, except that the FET response will be slower

Turns out I remembered you can just answer a question and draw something there and then not actually answer it

I'd upload a picture, but I'm lazy and tired

You do not need a resistor. At all.

Although if it's for consumer use, you may need to put in a safeguard to make sure the gate capacitor on the rush FET depletes quicker than the internal caps

Given sufficient forethought, some insight and a bit of LTSpice you can even do the whole MOSFET inrush thing with a well considered pair of two resistors and one small capacitor.

But then if you do that, you can also just insert a single HEXFET, or whatever new gizmo name they give them, with the 1:100 current sense output and use that output to create a timed servo that slowly increases the current.

Heck, nobody says you need resistors before the bridge, you can put the resistor (I still find it an ugly solution) on the DC side and just trigger a parallel MOSFET from the internal rail

You can also use an inrush NTC, though that is getting old-fashioned

If you want to go the resistance route, it seems to me making a MOSFET impedance is doable with only discretes

Either with SCRs on two spots in the bridge and fire them at the down going slope, earlier and earlier over time, snubbered with an inductance that can "catch" the 2V step you're creating, or with higher speed chopping with other discretes, post-filtered again with inductance.

Neater solution is just doing normal chopping like anyone else.

More importantly, if you're talking about 50Vrms, do not pick something that has a zero-cross hold off that can be as high as 20V, unless you like horrendously punishing caps, diodes and transformers

They may not fire, misfire, or release after trigger again.

Probably

And yes, 400kHz was supported on datasheet. No, C.S. wasn't. Ended up needing 150kHz or so in the Semtech case

I have had trouble, again Semtech is also on that list by the way, where commands just don't clock through right until you decide, hey, maybe... we need to go slower

More difficult to validate than "Not sure if I2C bus system is implemented correctly, thus not sure if commands always get processed" ?

@ThePhoton All the more reason, if I were a chip engineer, that I'd include it in every single one I did, regardless of whether it'd be needed.

Thus a slave not even considering it might just be fast, or also happens, unreliable

@ThePhoton Because clock stretching is a mechanism normally handled by well defined master blocks (i.e. those not made by ARM and TI in my experience, but otherwise most others) and is a mechanism intended to allow ensurance that a slave can internally process data correctly

... guess I forgot to label which syringe I filled with Bismuth solder...

Can get my hands on a bench top R&S RLC gizmo that measures in the kHzs

If saving money per unit is very important I know a freelancer who's happy to buy 10pcs of whatever cheaper one you have lined up from every outlet that sells small quantities (thus guaranteeing at least different reels as source) and run a couple tests