Prev: Friends important message for you
Next: Work at Home - Earn 50,000 Weekly Without Investment Its Not A Fake, 100% Earning Guarantee
From: Joerg on 14 Mar 2010 18:52 Vladimir Vassilevsky wrote: > > > Joerg wrote: > >> Vladimir Vassilevsky wrote: >> >>> Joerg wrote: >>> >>>> Vladimir Vassilevsky wrote: >>>> >>>>> >>>>> >>>>> Joerg wrote: >>>>> >>>>>> Yup, a poly resistor is nice. But you can't put crystals on the >>>>>> chip :-) >>>>> >>>>> >>>>> >>>>> Isn't it what Maxim does in their clock ICs ? >>>>> >>>> >>>> I never use Maxim parts. If you mean the former Dallas series 32kHz >>>> things those are modules that contain a tuning fork crystal. They >>>> may look like a chip but in reality it's modules. And my clients >>>> would probably have my head examined if I placed a timing clock >>>> oscillator that costs several Dollars :-) >>> >>> >>> There are not too many choices when you need an RTC. Besides, some >>> Dallas/Maxim parts offer the ability to adjust the clock frequency >>> digitally, so you can phase lock it to external signal. That's not >>> too bad for few dollars. >>> >> >> If you can get any, that is. Key "Maxim" into the Digikey serch >> engine, then click on oscillators. All but two are zero stock. Now why >> does that not surprise me? > > I am not great fan of Maxim as I had burned with them before like > everybody else. However you must be fair to them: > > "Items in stock: Yes" > "You have selected 60 items, spanning 3 pages" > When you go to oszillators there's 22 hits, 20 of them non-stock. Only the DS32KHZS is stocked in two temperature grades. >> If I needed a fancy RTC with phase-locking or other nifty add-ons I'd >> grab a uC. Much cheaper, and available. MSP430 comes to mind. > > My customers will eat me alive if I design in yet another programmable > device. Especially as a DAC, a temperature sensor and some sort of clock > adjustment circuitry will be required as well. MCU will drain somewhat > x10 standby battery power compared to what is required for a pure RTC. > The MSP430 is in the uA range with just a 32kHz crystal but yeah, needs programming. If you only need a simple PLL-lock there's other options, too. > > Vladimir Vassilevsky > DSP and Mixed Signal Design Consultant ^^^ Hmm, doesn't that imply something programmable? ... :-) > http://www.abvolt.com -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: Joerg on 14 Mar 2010 18:56 Jim Thompson wrote: > On Sun, 14 Mar 2010 14:13:57 -0700, Joerg <invalid(a)invalid.invalid> > wrote: > >> Vladimir Vassilevsky wrote: >>> >>> Joerg wrote: >>> >>>> Vladimir Vassilevsky wrote: >>>> >>>>> >>>>> Joerg wrote: >>>>> >>>>>> Yup, a poly resistor is nice. But you can't put crystals on the chip >>>>>> :-) >>>>> >>>>> Isn't it what Maxim does in their clock ICs ? >>>>> >>>> I never use Maxim parts. If you mean the former Dallas series 32kHz >>>> things those are modules that contain a tuning fork crystal. They may >>>> look like a chip but in reality it's modules. And my clients would >>>> probably have my head examined if I placed a timing clock oscillator >>>> that costs several Dollars :-) >>> There are not too many choices when you need an RTC. Besides, some >>> Dallas/Maxim parts offer the ability to adjust the clock frequency >>> digitally, so you can phase lock it to external signal. That's not too >>> bad for few dollars. >>> >> If you can get any, that is. Key "Maxim" into the Digikey serch engine, >> then click on oscillators. All but two are zero stock. Now why does that >> not surprise me? >> >> If I needed a fancy RTC with phase-locking or other nifty add-ons I'd >> grab a uC. Much cheaper, and available. MSP430 comes to mind. > > I find it so incongruous... Joerg is designing medical electronics > with _cheap_ components :-( > Well, that _is_ the trick. If you don't deliver a device at the best possible cost, your competitor eventually will ;-) Some of my clients retain me for exactly that reason. They know they get a cost efficient design and also one where the purchasers don't have to go on many wild goose chases to find stock. Typically the fact that a similar machine is going to be available for less cost is not known until a major trade show. And then it's definitely too late to react. Way too late. -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: Joerg on 14 Mar 2010 18:58 Jim Thompson wrote: > On Sun, 14 Mar 2010 13:34:49 -0700, Joerg <invalid(a)invalid.invalid> > wrote: > >> Jim Thompson wrote: >>> On Sun, 14 Mar 2010 13:25:18 -0700, Joerg <invalid(a)invalid.invalid> >>> wrote: >>> >>>> Jim Thompson wrote: >>>>> On Sun, 14 Mar 2010 09:51:52 -0700, Joerg <invalid(a)invalid.invalid> >>>>> wrote: >>>>> >>>>>> Jim Thompson wrote: >>>>>>> On Sun, 14 Mar 2010 09:05:27 -0700, Joerg <invalid(a)invalid.invalid> >>>>>>> wrote: >>>>>>> >>>>>>>> Jim Thompson wrote: >>>>>> [...] >>>>>> >>>>>>>>> My rule of thumb, which seems to work with every inverter-style >>>>>>>>> crystal oscillator I've ever tried: Use capacitors as recommended by >>>>>>>>> the crystal specification... if it says 15pF, that means 30pF on EACH >>>>>>>>> end to ground. Then choose the resistor based on 45� phase shift with >>>>>>>>> the first capacitor (at the crystal frequency). >>>>>>>>> >>>>>>>> I am usually happy with around 1M. All it needs to do is to crank it >>>>>>>> over. Kaaaa-chum-pah ... POOF ... vrooooom. Unless I must start it >>>>>>>> really fast for some reason but that can require special kicker circuitry. >>>>>>> Huh? You and I are talking two different resistors. I'm talking the >>>>>>> series drive resistor; you're talking the DC loop resistor. >>>>>>> >>>>>> Ok, different thing. Mostly you can get away without although that's not >>>>>> always the nice way of doing things. >>>>>> >>>>>> >>>>>>> Which brings to mind... does the CD4060 have the DC loop resistor >>>>>>> built-in? It would appear not. >>>>>>> >>>>>> AFAIR not. >>>>>> >>>>>> >>>>>>> And I should pay closer attention to details :-( The OP has 100K >>>>>>> there... needs to be 10-20Meg. >>>>>>> >>>>>> I usually go for 1M. >>>>> CD series stuff doesn't have a lot of gm. But 1Meg is good for >>>>> something like 74HCU04. >>>>> >>>> For the CD4060 they recommend even higher values, page 3: >>>> >>>> http://www.fairchildsemi.com/ds/CD%2FCD4060BC.pdf >>> Didn't I say 10-20Meg ?:-) >>>>> DO NOT USE buffered inverters for crystal oscillators! >>>>> >>>> Depends on the frequency but generally yes, don't use those if you can >>>> avoid it. >>> Makes for lousy stability and possible non-starts. >>> >> I never had those problems. But I also prefer non-buffered inverters for >> that. >> >>>>>> 10M may not work outdoors when humidity is high and >>>>>> some condensation occurs. >>>>> Mine are usually on-chip :-) >>>>> >>>> Yup, a poly resistor is nice. But you can't put crystals on the chip :-) >>> Wanna bet ?:-) >>> >> With MEMS you could, lots of things are possible. But unlikely as good >> as a real crystal. > > You _can_ mount a membrane type crystal onto a chip, though, right > now, we're mounting the crystal separately... hybrid board style. > Sure, you could. But the question is will it be good enough and low enough in cost? I've been somewhat involved in a membrane type custom chip although the membrane was used to measure displacement, not to oscillate. That is not exactly a low cost field. -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: John Fields on 14 Mar 2010 19:10 On Sun, 14 Mar 2010 10:46:33 -0700, Jim Thompson <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: >On Sun, 14 Mar 2010 12:17:10 -0500, John Fields ><jfields(a)austininstruments.com> wrote: > >>On Sun, 14 Mar 2010 09:29:36 -0700, Jim Thompson >><To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: >> >>>On Sun, 14 Mar 2010 09:05:27 -0700, Joerg <invalid(a)invalid.invalid> >>>wrote: >> >>>>I am usually happy with around 1M. All it needs to do is to crank it >>>>over. Kaaaa-chum-pah ... POOF ... vrooooom. Unless I must start it >>>>really fast for some reason but that can require special kicker circuitry. >>> >>>Huh? You and I are talking two different resistors. I'm talking the >>>series drive resistor; you're talking the DC loop resistor. >>> >>>Which brings to mind... does the CD4060 have the DC loop resistor >>>built-in? It would appear not. >> >>--- >>No, but that's probably not a bad thing since you like 10M and I like >>1M. >> >>One rather unfortunate thing, though, is that RESET stops the oscillator >>so you can't use RESET to do a divide-by-n unless you use an external >>clock. >>--- >> >>>And I should pay closer attention to details :-( The OP has 100K >>>there... needs to be 10-20Meg. >> >>--- >>Seems a little high ... I've always used 1 megohm >> >>JF > >My resistors are always on-chip :-) --- Sometimes mine are, sometimes they aren't. ;) JF
From: Joerg on 14 Mar 2010 19:27
John Fields wrote: > On Sun, 14 Mar 2010 10:46:33 -0700, Jim Thompson > <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: > >> On Sun, 14 Mar 2010 12:17:10 -0500, John Fields >> <jfields(a)austininstruments.com> wrote: >> >>> On Sun, 14 Mar 2010 09:29:36 -0700, Jim Thompson >>> <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: >>> >>>> On Sun, 14 Mar 2010 09:05:27 -0700, Joerg <invalid(a)invalid.invalid> >>>> wrote: >>>>> I am usually happy with around 1M. All it needs to do is to crank it >>>>> over. Kaaaa-chum-pah ... POOF ... vrooooom. Unless I must start it >>>>> really fast for some reason but that can require special kicker circuitry. >>>> Huh? You and I are talking two different resistors. I'm talking the >>>> series drive resistor; you're talking the DC loop resistor. >>>> >>>> Which brings to mind... does the CD4060 have the DC loop resistor >>>> built-in? It would appear not. >>> --- >>> No, but that's probably not a bad thing since you like 10M and I like >>> 1M. >>> >>> One rather unfortunate thing, though, is that RESET stops the oscillator >>> so you can't use RESET to do a divide-by-n unless you use an external >>> clock. >>> --- >>> >>>> And I should pay closer attention to details :-( The OP has 100K >>>> there... needs to be 10-20Meg. >>> --- >>> Seems a little high ... I've always used 1 megohm >>> >>> JF >> My resistors are always on-chip :-) > > --- > Sometimes mine are, sometimes they aren't. ;) > Currently most of mine are in neatly labeled bins :-) -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM. |