Prev: Bad Resistors
Next: 4mA current regulator
From: Joerg on 29 Jun 2010 16:20 Spehro Pefhany wrote: > On Tue, 29 Jun 2010 10:13:29 -0700, Joerg <invalid(a)invalid.invalid> > wrote: > >> Winston wrote: >>> On 6/29/2010 12:47 AM, Nial Stewart wrote: >>> >> [...] >> >>> (WRT 'commodity parts':) >>>>> For the right money, that 12 month commodity lead time would >>>>> magically shrink to nearly nothing. >>>>> It's the second oldest game in the world. >>>> I'm not sure, the Altera FAE was genuinely frustrated. He said >>>> the only thing saving their bacon is that Xilinx are in nearly >>>> the same position. >>> I've been out of the game for a long time but I still wonder: >>> Is his response 'code' for "re-design using Lattice, Actel, >>> Cypress, Quicklogic, Silicon Blue, Achronix, etc."? >>> >> The real message is: Try your darndest to design it with discretes, >> simple logic chips, maybe 80C51 and whatever else is needed. No more >> FPGA. Which is what I generally do, then there are half a dozen sources >> and never a shortage :-) > > There might not be a market for something that can't do high > performance signal processing. > A lot of times it turns out that one can get away with much less fancy signal processing. In some cases none at all. BTDT. So far the topper was a self-cal scheme to gang several ADCs. They had a huge DSP on the board, did all kinds of frequency domain finagling, it ran forever, barely ever converged and mostly ended in a calibration error. So, I rolled up the sleeves, pulled out the bush knife and got rid of all that. In came a few analog phase shifters, servo loops and an odd but fixed test signal. Oh, yeah, there were two signal lines which technically constitutes a digital world. One for "start cal", the other to signal "done". "Done" came on a few milliseconds after "start cal", every single time, and calibration was dead on. The fancy DSP was now demoted to an RS232 arbiter because some other part of the board needed that functionality. I think they left it on there for historical purposes and because they didn't want to re-write the RS232 code. -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: Nial Stewart on 30 Jun 2010 05:08 > The real message is: Try your darndest to design it with discretes, > simple logic chips, maybe 80C51 and whatever else is needed. No more > FPGA. Which is what I generally do, then there are half a dozen sources > and never a shortage :-) > > I know this sounds kind of Luddite but it has served my clients quite well. Discretes, discretes, you don't need no steeenking discretes. I only use a single piece of wire when I can get away with it! More often than not I can't. :-) Nial.
From: Joe G (Home) on 30 Jun 2010 10:46 "Nial Stewart" <nial*REMOVE_THIS*@nialstewartdevelopments.co.uk> wrote in message news:88k12sFo0kU1(a)mid.individual.net... >> I should have included a smiley, for I was attempting >> to suggest that any 'component shortage' has got to >> be a product of imagination rather than real. > > > In 15 years using Altera devices I've never had any problems obtaining > them. > > I've been kitting up for a prototype build and my FAE has had to pull > a lot of strings to get me 4 Stratix's that were fairly freely available > when the project started 6 months ago. It'll be October before I see any > more. > > I also have an Analog Devices ADC designed in which was available from > Digikey and a couple of other distributors four months ago but I've > had to get Analog customer services to trawl their system to get me > 4 for debug. Again it's November before these will be available. > > The Altera FAE said that 15 smaller fabs closed last year so all the > big ones are working at more than capacity(?). Altera use TSMC which is > fully booked. They announced the Cyclone IV family in October but > so far only one device is out, the rest were due pretty quickly after > that but are now scheduled for the Autumn (because they can't build > them). > > Shortages are real IMHO. > > > Nial My point is that you have a good relationship with your supplier and supplier FAE. Some engineers on this thread... won't give the suppliers the time of day... until they need them for supply or technical support. Joe
From: E on 30 Jun 2010 13:13 "Spehro Pefhany" <speffSNIP(a)interlogDOTyou.knowwhat> kirjoitti viestiss�:rbck26hjbh229b70qhco514lupov8j287i(a)4ax.com... > On Tue, 29 Jun 2010 10:13:29 -0700, Joerg <invalid(a)invalid.invalid> > wrote: > >>Winston wrote: >>> On 6/29/2010 12:47 AM, Nial Stewart wrote: >>> >> >>[...] >> >>> (WRT 'commodity parts':) >>>>> For the right money, that 12 month commodity lead time would >>>>> magically shrink to nearly nothing. >>>>> It's the second oldest game in the world. >>>> >>>> I'm not sure, the Altera FAE was genuinely frustrated. He said >>>> the only thing saving their bacon is that Xilinx are in nearly >>>> the same position. >>> >>> I've been out of the game for a long time but I still wonder: >>> Is his response 'code' for "re-design using Lattice, Actel, >>> Cypress, Quicklogic, Silicon Blue, Achronix, etc."? >>> >> >>The real message is: Try your darndest to design it with discretes, >>simple logic chips, maybe 80C51 and whatever else is needed. No more >>FPGA. Which is what I generally do, then there are half a dozen sources >>and never a shortage :-) > > There might not be a market for something that can't do high > performance signal processing. > Well x86 is multi-sourced and quite backwards compatible. -ek
From: Joerg on 4 Jul 2010 20:28
Joe G (Home) wrote: > "Nial Stewart" <nial*REMOVE_THIS*@nialstewartdevelopments.co.uk> wrote in > message news:88k12sFo0kU1(a)mid.individual.net... >>> I should have included a smiley, for I was attempting >>> to suggest that any 'component shortage' has got to >>> be a product of imagination rather than real. >> >> In 15 years using Altera devices I've never had any problems obtaining >> them. >> >> I've been kitting up for a prototype build and my FAE has had to pull >> a lot of strings to get me 4 Stratix's that were fairly freely available >> when the project started 6 months ago. It'll be October before I see any >> more. >> >> I also have an Analog Devices ADC designed in which was available from >> Digikey and a couple of other distributors four months ago but I've >> had to get Analog customer services to trawl their system to get me >> 4 for debug. Again it's November before these will be available. >> >> The Altera FAE said that 15 smaller fabs closed last year so all the >> big ones are working at more than capacity(?). Altera use TSMC which is >> fully booked. They announced the Cyclone IV family in October but >> so far only one device is out, the rest were due pretty quickly after >> that but are now scheduled for the Autumn (because they can't build >> them). >> >> Shortages are real IMHO. >> >> >> Nial > > My point is that you have a good relationship with your supplier and > supplier FAE. > > Some engineers on this thread... won't give the suppliers the time of > day... until they need them for > supply or technical support. > And give the FAEs feedback in a friendly way. I discover bugs on a regular basis. Then the style would be "Your so-and-so opamp has the following issue ..." instead of "Your so-and-so opamp really sucks" :-) -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM. |