ON TOPIC... Circuit Design
in [Design]
|
Prev: Custom LCDs
Next: Fun with positive feedback
From: Fred Bartoli on 13 Jan 2010 16:40 Jim Thompson a �crit : > On Wed, 13 Jan 2010 14:00:23 -0700, Jim Thompson > <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com/Snicker> wrote: > >> On Wed, 13 Jan 2010 21:39:12 +0100, Fred Bartoli <" "> wrote: >> > [snip] >>> Have a look at AD8605. Just amazing. >>> > [snip] >> Yep. Judging from the input bias specs, they've got some HUGE input >> devices there. > > [snip] > > Looked it up... L=1u W=1600u !! > Come on, you should know better... -- Thanks, Fred.
From: Jim Thompson on 13 Jan 2010 16:46 On Wed, 13 Jan 2010 22:40:23 +0100, Fred Bartoli <" "> wrote: >Jim Thompson a �crit : >> On Wed, 13 Jan 2010 14:00:23 -0700, Jim Thompson >> <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com/Snicker> wrote: >> >>> On Wed, 13 Jan 2010 21:39:12 +0100, Fred Bartoli <" "> wrote: >>> >> [snip] >>>> Have a look at AD8605. Just amazing. >>>> >> [snip] >>> Yep. Judging from the input bias specs, they've got some HUGE input >>> devices there. >> >> [snip] >> >> Looked it up... L=1u W=1600u !! >> > >Come on, you should know better... That's W=1.6mm, longer than most chips edge dimension. So the input stage dominates the chip area. Not that unusual, check out... http://www.analog-innovations.com/SED/ICE-LM195-LM117.pdf The pass device takes up 2/3 of the chip. ...Jim Thompson -- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
From: Spehro Pefhany on 13 Jan 2010 17:10 On Wed, 13 Jan 2010 22:22:23 +0100, Fred Bartoli <" "> wrote: >Spehro Pefhany a �crit : >> rOn Wed, 13 Jan 2010 21:39:12 +0100, Fred Bartoli <" "> wrote: >> >>> Jim Thompson a �crit : >>>> On Wed, 13 Jan 2010 08:40:14 -0800, "Joel Koltner" >>>> <zapwireDASHgroups(a)yahoo.com> wrote: >>>> >>>>> Did that design make it through fab, Jim, or did the company start imploding >>>>> first? >>>> It was fabbed at Polarfab. The company began disintegration while >>>> debugging the digital was in progress. >>>> >>>>> I thought that combined bipolar/CMOS ("BiFET?") processes were generally >>>>> considered too exotic/unnecessary-due-to-performance-improvements-in-CMOS >>>>> these days? Not so? >>>>> >>>>> ---Joel >>>> Not so for analog. Try making a low noise amplifier with only CMOS. >>> Have a look at AD8605. Just amazing. >> >> 16nV/sqrt(Hz) at 40Hz typical, vs. more like 1nV/sqrt(Hz) for a good >> bipolar op-amp operating from a low source impedance. >> > >Never said it could compete with a low noise bjt opamp. > >But 8nV/rtHz noise floor isn't common for cmos opamps, which often are >at least 5 times that high. And a 140Hz fc with a 8nV/rtHz is pretty >good too, as it is ordinary at least as high, but with a 5-10 times >higher noise floor. > >Sure it can't compete, but so can't LN BJT opamps with my 200pV/rtHz preamp. BIG input JFETs (in paraellel?) >Still, the AD8605 is amazing for a CMOS opamp. Yeah, it's pretty good alright. A far cry from early CMOS amplifiers. >(and I like the tiny 5 balls WLCSP, despite they goofed the pinout).
From: Jim Thompson on 13 Jan 2010 17:06 On Wed, 13 Jan 2010 17:10:14 -0500, Spehro Pefhany <speffSNIP(a)interlogDOTyou.knowwhat> wrote: >On Wed, 13 Jan 2010 22:22:23 +0100, Fred Bartoli <" "> wrote: > >>Spehro Pefhany a �crit : >>> rOn Wed, 13 Jan 2010 21:39:12 +0100, Fred Bartoli <" "> wrote: >>> >>>> Jim Thompson a �crit : >>>>> On Wed, 13 Jan 2010 08:40:14 -0800, "Joel Koltner" >>>>> <zapwireDASHgroups(a)yahoo.com> wrote: >>>>> >>>>>> Did that design make it through fab, Jim, or did the company start imploding >>>>>> first? >>>>> It was fabbed at Polarfab. The company began disintegration while >>>>> debugging the digital was in progress. >>>>> >>>>>> I thought that combined bipolar/CMOS ("BiFET?") processes were generally >>>>>> considered too exotic/unnecessary-due-to-performance-improvements-in-CMOS >>>>>> these days? Not so? >>>>>> >>>>>> ---Joel >>>>> Not so for analog. Try making a low noise amplifier with only CMOS. >>>> Have a look at AD8605. Just amazing. >>> >>> 16nV/sqrt(Hz) at 40Hz typical, vs. more like 1nV/sqrt(Hz) for a good >>> bipolar op-amp operating from a low source impedance. >>> >> >>Never said it could compete with a low noise bjt opamp. >> >>But 8nV/rtHz noise floor isn't common for cmos opamps, which often are >>at least 5 times that high. And a 140Hz fc with a 8nV/rtHz is pretty >>good too, as it is ordinary at least as high, but with a 5-10 times >>higher noise floor. >> >>Sure it can't compete, but so can't LN BJT opamps with my 200pV/rtHz preamp. > >BIG input JFETs (in paraellel?) > >>Still, the AD8605 is amazing for a CMOS opamp. > >Yeah, it's pretty good alright. A far cry from early CMOS amplifiers. > >>(and I like the tiny 5 balls WLCSP, despite they goofed the pinout). I suspect the input stage is actually JFET's, or gate-enhanced lateral bipolar devices. The input current seems high for MOS. ...Jim Thompson -- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
From: Fred Bartoli on 13 Jan 2010 17:47
Spehro Pefhany a �crit : > On Wed, 13 Jan 2010 22:22:23 +0100, Fred Bartoli <" "> wrote: > >> Spehro Pefhany a �crit : >>> rOn Wed, 13 Jan 2010 21:39:12 +0100, Fred Bartoli <" "> wrote: >>> >>>> Jim Thompson a �crit : >>>>> On Wed, 13 Jan 2010 08:40:14 -0800, "Joel Koltner" >>>>> <zapwireDASHgroups(a)yahoo.com> wrote: >>>>> >>>>>> Did that design make it through fab, Jim, or did the company start imploding >>>>>> first? >>>>> It was fabbed at Polarfab. The company began disintegration while >>>>> debugging the digital was in progress. >>>>> >>>>>> I thought that combined bipolar/CMOS ("BiFET?") processes were generally >>>>>> considered too exotic/unnecessary-due-to-performance-improvements-in-CMOS >>>>>> these days? Not so? >>>>>> >>>>>> ---Joel >>>>> Not so for analog. Try making a low noise amplifier with only CMOS. >>>> Have a look at AD8605. Just amazing. >>> 16nV/sqrt(Hz) at 40Hz typical, vs. more like 1nV/sqrt(Hz) for a good >>> bipolar op-amp operating from a low source impedance. >>> >> Never said it could compete with a low noise bjt opamp. >> >> But 8nV/rtHz noise floor isn't common for cmos opamps, which often are >> at least 5 times that high. And a 140Hz fc with a 8nV/rtHz is pretty >> good too, as it is ordinary at least as high, but with a 5-10 times >> higher noise floor. >> >> Sure it can't compete, but so can't LN BJT opamps with my 200pV/rtHz preamp. > > BIG input JFETs (in paraellel?) > Yep, a 6 x IF3602 (duals) LPT input stage running a total 240mA tail current. About 2nV/rt Hz at... 0.1Hz and 1nV/rtHz at 1Hz with a nice low input current noise. Not that it matters regarding the input load, but the measured signal is biased so a series cap was required, which carried its share of funny design consequences... -- Thanks, Fred. |