| Deutsch English Français Italiano |
|
<t7tdgjtfp104hsot4p7da4fna9l88ir86j@4ax.com> View for Bookmarking (what is this?) Look up another Usenet article |
Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail From: john larkin <jl@glen--canyon.com> Newsgroups: sci.electronics.design Subject: Re: power supply discharge Date: Wed, 09 Oct 2024 14:35:47 -0700 Organization: A noiseless patient Spider Lines: 171 Message-ID: <t7tdgjtfp104hsot4p7da4fna9l88ir86j@4ax.com> References: <qvqlfjt4ttk1qeae20tje6mblci4h4d2ku@4ax.com> <lm0a6iFg65cU2@mid.individual.net> <01lmfj52p9aurg23v6oq29j3shutja5tnb@4ax.com> <kj4ofj56re4nl4h6qses2erq6fdgl6gbj7@4ax.com> <6baofjtquh870j7ccl49q9lc1lalcgmedv@4ax.com> <p8lofj9162t9ianni8fepgbv31drc7et26@4ax.com> <bnibgj9532fp554uek2l2rnn8ls12khl89@4ax.com> <8nedgjld2g6dt2oif5q4ia483vfortc72n@4ax.com> <b2ndgjhqb7oprdg285jq4qrq360uunqpg6@4ax.com> <hdodgj5g0l6kh5r81ff1u436nmcj0kq6qo@4ax.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Injection-Date: Wed, 09 Oct 2024 23:34:34 +0200 (CEST) Injection-Info: dont-email.me; posting-host="7fa7b4ff58c2fc5d654dc97f6098c6a0"; logging-data="2953196"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX18tThf7fWVHzRtQOgWc+Vty" User-Agent: ForteAgent/8.00.32.1272 Cancel-Lock: sha1:uqb8LX5cVnvAT67tHEvQsbtAsWk= Bytes: 8475 On Wed, 09 Oct 2024 16:37:22 -0400, Joe Gwinn <joegwinn@comcast.net> wrote: >On Wed, 09 Oct 2024 12:51:32 -0700, john larkin <jl@glen--canyon.com> >wrote: > >>On Wed, 09 Oct 2024 13:40:49 -0400, Joe Gwinn <joegwinn@comcast.net> >>wrote: >> >>>On Tue, 08 Oct 2024 17:27:53 -0700, john larkin <jl@glen--canyon.com> >>>wrote: >>> >>>>On Tue, 01 Oct 2024 16:03:40 -0400, Joe Gwinn <joegwinn@comcast.net> >>>>wrote: >>>> >>>>>On Tue, 01 Oct 2024 09:59:27 -0700, john larkin <jl@glen--canyon.com> >>>>>wrote: >>>>> >>>>>>On Tue, 01 Oct 2024 11:24:34 -0400, Joe Gwinn <joegwinn@comcast.net> >>>>>>wrote: >>>>>> >>>>>>>On Mon, 30 Sep 2024 18:49:14 -0700, john larkin <JL@gct.com> wrote: >>>>>>> >>>>>>>>On Mon, 30 Sep 2024 11:49:54 -0700, Joerg <news@analogconsultants.com> >>>>>>>>wrote: >>>>>>>> >>>>>>>>>On 9/30/24 11:24 AM, john larkin wrote: >>>>>>>>>> On Mon, 30 Sep 2024 08:39:27 -0400, legg <legg@nospam.magma.ca> wrote: >>>>>>>>>> >>>>>>>>>>> On Sun, 29 Sep 2024 08:23:01 -0700, john larkin <JL@gct.com> wrote: >>>>>>>>>>> >>>>>>>>>>>> On Sat, 28 Sep 2024 22:28:07 -0700, Joerg <news@analogconsultants.com> >>>>>>>>>>>> wrote: >>>>>>>>>>>> >>>>>>>>>>>>> On 9/27/24 8:07 AM, john larkin wrote: >>>>>>>>>>>>>> >>>>>>>>>>>>>> Given a benchtop power supply, you can turn the voltage up and then >>>>>>>>>>>>>> down, and it goes down. Most have a substantial amount of output >>>>>>>>>>>>>> capacitance, and can be driving an external cap too. So something >>>>>>>>>>>>>> pulls the output down. >>>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> Often the only internal load is the resistive divider for the regulator >>>>>>>>>>>>> loop feedback. >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>>> I guess that there are no standards for this, but I've never seen a >>>>>>>>>>>>>> supply that just hangs high when it's cranked down. >>>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> I have some. They drop very slowly when there isn't much load on the output. >>>>>>>>>>>> >>>>>>>>>>>> Customers might whine if they ask for 10 volts and see 30. Amd that >>>>>>>>>>>> may be mostly held up by their capacitive load. >>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>>> I'm designing some programmable multi-channel power suplies and that >>>>>>>>>>>>>> is one of many tangled issues in the project. >>>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> A synchronous buck architecture should work quite well if you need to >>>>>>>>>>>>> slew fast. I've used that on a driver that had to modulate a hard >>>>>>>>>>>>> capacitive load at several kHz and above 100V. >>>>>>>>>>>> >>>>>>>>>>>> I'm doing some multichannel non-isolated supplies that will be sync >>>>>>>>>>>> buck, using multiple TI DRV8962 chips. >>>>>>>>>>>> >>>>>>>>>>>> One problem is that a sync buck can become a boost in the wrong >>>>>>>>>>>> direction, and start charging my +48 supply. If it hits, say, 55 >>>>>>>>>>>> volts, I'll disable the switcher chips, and the outputs can hang. I >>>>>>>>>>>> need to discharge the outputs. I'm thinking about 20 mA of depletion >>>>>>>>>>>> fet per channel. >>>>>>>>>>> >>>>>>>>>>> You might consider overvoltage protection or a (switched ?) >>>>>>>>>>> internal minimum load.There's usuaally some point in the >>>>>>>>>>> control loop that's a good indicator of a pull-down requirement. >>>>>>>>>>> A single ovp or autoload on the input looks likely to serve >>>>>>>>>>> all of your many sync-bucks. >>>>>>>>>>> >>>>>>>>>>> RL >>>>>>>>>> >>>>>>>>>> An MOV on the bulk supply could limit the reverse-pump excursion until >>>>>>>>>> the software can notice and shut things down. >>>>>>>>>> >>>>>>>>>> MOVs can gobble a lot of joules, but their clipping is very soggy. >>>>>>>>>> >>>>>>>>> >>>>>>>>>MOVs are usually cumulative. They can take a certain amount of >>>>>>>>>dissipation over their lifetime and then *PHUT* ... POOOF. Like a bank >>>>>>>>>account that runs dry. >>>>>>>> >>>>>>>>What kills MOVs? Integrated joules? Time-temperature? >>>>>>>> >>>>>>>>I don't expect a lot of joules per event. Just enough energy to keep >>>>>>>>my supply voltage down until a slowish ADC and the software can shut >>>>>>>>the buck switchers down. 15 milliseconds max, maybe. >>>>>>> >>>>>>>I think it's integrated joules per cubic centimeter of the MOV >>>>>>>material. This is discussed in the literature on MOVs for protecting >>>>>>>line-powered equipment from pulse overvoltages, such as from nearby >>>>>>>lightning strikes. <https://www.deltala.com/> >>>>>>> >>>>>>>Joe Gwinn >>>>>> >>>>>>Makes sense. It looks like most MOV appnotes assume that it's across >>>>>>an AC line, with kilo-amps available. Or lightning bolts. >>>>>> >>>>>>I'll get a few and test them at much lower loads. >>>>> >>>>>For smaller MOVs, I think that the data sheet specifies capacity in >>>>>Joules. I bet this is the max integrated dose, not the pre-event >>>>>limit. Well, the one-event limit as well. >>>>> >>>>>Joe Gwinn >>>> >>>>I'm torturing an MOV, a 470KD14. It's rated for 47 volts and 0.1 watt >>>>and 10 joules. >>>> >>>>At a constant 15 mA, it's at 58.1 volts, which is 0.86 watts. It's >>>>pretty warm. The voltage seems very stable after 4 hours so far. >>>>That's about 12K joules. >>>> >>>>It's likely it could do that forever, but the data sheets suggest that >>>>high power shots can do cumulative damage. I might set up to try that >>>>somehow. >>> >>>I bet that the duty cycle affects the cumulative damage, with smaller >>>duty cycles (more powerful pulses, but more widely separated) doing >>>more damage than just the cumulative energy. >>> >>>I looked at the Yageo 470KD14 MOV datasheet. It does not seem to >>>mention any wearout effect. Perhaps they figured the mechanism out >>>and remedied it, which would be a good thing. >>> >>>But the "surge life" items under "Reliability" on page 9 only does ten >>>surges and notes no visible damage, so we have no idea what happens >>>beyond that simple surge test's parameters. >>> >>>Joe Gwinn >> >>On page 5, it doesn't say so but I think the curves are parametreized >>on the number of shots, 1 to 1e6. > >Yes, one can certainly read it that way. Probably have to ask Yageo >how to read those plots, and the underlying physical mechanism. > > >>I might have to cut over to using mosfets and resistors to dump my >>overshoot energy. MOVs may be too risky longterm. Pity... they are so >>simple. > > How large are the surges and how long will it be to get to 10^6 >surges in total? > >Joe Gwinn That's tricky. Some user might slam a capacitive load or a motor a lot of times. Here's a Riedon ceramic DPAK 50 ohm resistor. It could absorb at least 50j, 100 with two in parallel. That would work. They will need a mosfet to switch them on when the 48v supply gets over-driven to 58 maybe. https://www.dropbox.com/scl/fi/octctz94vdi4ac4aageit/Dpak-50r-joules.jpg?rlkey=y21a3x8xmkno82ezrb4vefxrr&raw=1 The Caddock TO-220 resistors have a big metal tab like a mosfet and would absorb more joules, but are more expensive. They would be an option.