Path: ...!Xl.tags.giganews.com!local-2.nntp.ord.giganews.com!nntp.supernews.com!news.supernews.com.POSTED!not-for-mail NNTP-Posting-Date: Fri, 26 Apr 2024 18:04:59 +0000 From: John Larkin Newsgroups: sci.electronics.design Subject: Re: PI3USB14-A dual 4>1 USB switch for analog Date: Fri, 26 Apr 2024 11:03:08 -0700 Organization: Highland Tech Reply-To: xx@yy.com Message-ID: References: X-Newsreader: Forte Agent 3.1/32.783 MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit Lines: 86 X-Trace: sv3-xZCzvyz8hEQpYPoRsxX4bIxVMWdBSUxSiuBsj1V99C6R9M9wCKaPyzar2mO1A8V55VcgTCmwsJuhi+L!sptEqbNc4Z5mcWS6dKxvQCYukh722CliRu0lmrHsSbigX7iVMPerovAmqIRunZUeRai2pcnQw4wj!bNHa3Q== X-Complaints-To: www.supernews.com/docs/abuse.html X-DMCA-Complaints-To: www.supernews.com/docs/dmca.html X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly X-Postfilter: 1.3.40 Bytes: 4947 On Sat, 27 Apr 2024 01:12:13 +1000, Chris Jones wrote: >On 27/04/2024 1:01 am, Chris Jones wrote: >> On 26/04/2024 2:28 am, Phil Hobbs wrote: >>> Hi, all, >>> >>> I'm in a bit of a rush, doing revs to a testing board for a prototype >>> lidar ASIC that turns out to be..., um..., well....  Let's just say >>> it's a _tiny_bit_delicate_ electrically.  Yeah, just a little >>> delicate, that's it. (Not our design or our nickel, fortunately.) >>> >>> It runs on a single supply, so the only sequencing issues are with the >>> signal pins. Thus the test board is going to bring up V_DD smoothly >>> and not too slowly: a monotonic edge of about 100 us.  While that's >>> going on, all the inputs and outputs will see 100k ohms to ground. >>> (All the enables are positive-true, which makes this easier.) >>> >>> The digital pins are all inputs, so they're just buffered with >>> SN74LVC1G125s (tri-state Schmitt noninverting), with the >>> aforementioned 100k to ground on their outputs. >>> >>> The analog outputs don't have much drive, but can go pretty fast if >>> you keep the capacitance down.  There are a fair few of them, so I >>> want to mux them down before the op amp buffers >>> (OPA2626es--surprisingly nice chips). >>> >>> Because we're not made of money, and the test boards aren't ITAR, >>> we're getting them made and stuffed at JLCPCB.  Thus we care a lot >>> about what their tame distributor LCSC has in stock. >>> >>> They're fairly light on analog muxes, it turns out.  Going through >>> what they do have, I was reminded just how awful the capacitance of >>> old-timey mux parts is--way over 100 pF at the common pin of an 8-1, >>> blech. >>> >>> They do have these very nice-looking USB switches, type PI3USB14-A, >>> for pretty cheap ($0.45 @ 1ea) and in plentiful stock. >>> >>> They look beautiful--6 ohms R_on, 4 pF C_off typical--but apart from >>> one crosstalk number, the datasheet gives only sketchy hints about >>> their actual analog performance--bandwidth and crosstalk, but nothing >>> about charge injection, switching times, make-before-break, resistance >>> linearity, or distortion. >>> >>> Has any of you used them for actual analog stuff? >>> >>> Thanks >>> >>> Phil Hobbs >>> >> >> As long as you don't need the signals to swing near the positive rail, I >> suggest you google "H-mode mixer" as a lot of amateur radio enthusiasts >> have built mixers using bus switches, basically just nmos fets with a >> gate driver. For example: >> https://martein.home.xs4all.nl/pa3ake/hmode/switches.html >> >> They may not be better than the PI3USB14-A but I have a better guess of >> what is in them. > >Ah, I see they also tried ones with both NMOS and PMOS devices in the >switches. The FSA3157 seems to have somewhat more complete >specifications than the one you mentioned. The 3157 is nice too, SPDT in a small package, with specified charge injection. We pay 7 cents. > >If you don't happen to need it to swing near the positive rail, you >might get less capacitance with a NMOS-only type like the venerable FST3125. Optimistically, complementary P and N fets would have cancelling charge injection. But that doesn't always work (another sad story.) > >I hope the ones that also have PMOS devices in the switches keep the >backgates of the PMOS at the rail (like a 4016), and don't do the >horrible backgate switching trickery to reduce Ron like in the old 4066. >I think some of those might have been so bad that they could momentarily >short your signal to the rail, and even if not, they certainly had >another big charge injection mechanism.