Ken S. Tucker wrote:> On Nov 2, 12:03 pm, Phil Hobbs > <pcdhSpamMeSensel...@electrooptical.net> wrote: >> Ken S. Tucker wrote: >>> On Nov 2, 10:11 am, 2G<soar2mor...@yahoo.com> wrote: >>>> On Nov 2, 8:12 am, Phil Hobbs<pcdhSpamMeSensel...@electrooptical.net> >>>> wrote: >> >>>>> Ken S. Tucker wrote: >>>>>> On Nov 1, 10:48 pm, Phil Hobbs >>>>>> <pcdhSpamMeSensel...@electrooptical.net> wrote: >>>>>>> So, I had this application requiring really absurd amounts of ripple >>>>>>> rejection--it's a piezo driver in a laser wavelength locker that (for >>>>>>> historical reasons) has to run off a fairly crappy dual isolated 48V >>>>>>> DC-DC converter. The 48V has almost a half volt of ripple at 47 kHz, >>>>>>> and the output noise needed to be in the nanovolts. >> >>>>>> Would 4 x 12V rechargable batteries do (?). >>>>>> I could explain further unless that option has been eliminated. >>>>>> Regards >>>>>> Ken >> >>>>>>> As we've discussed here before, ordinary cap multipliers are limited by >>>>>>> Early effect and collector-emitter capacitance. This one is a bit >>>>>>> complicated--14 parts--but according to LTSPICE it provides 180 dB of >>>>>>> ripple rejection from about 40 kHz to above 10 MHz. (In reality it >>>>>>> won't be this good of course.) The interesting thing is that even with >>>>>>> a it drops only about 1.5V, due to daisy-chaining the RC lowpasses--the >>>>>>> second stage is biased from the base of the first stage instead of the >>>>>>> emitter, which saves a diode drop. Adding the second stage requires >>>>>>> only about 350 mV total. >> >>>>>>> The remaining limitations are due to the combined Early effect of the >>>>>>> transistors--very small, since they form a cascode pair--and all the >>>>>>> interelectrode capacitances in series. >> >>>>>>> Good medicine for the present problem! >> >>>>>>> Cheers >> >>>>>>> Phil Hobbs >> >>>>>>> Q1 Q2 >> >>>>>>> IN DNLS160 DNLS160 3 OUT >> >>>>>>> 0-----*--- --------- ------RR-*--------0 >>>>>>> | \ / \ / | >>>>>>> | \ A \ A | >>>>>>> | --------- --------- | >>>>>>> 390 R | | === 47uF alum >>>>>>> R | | | || 1uF X7R >>>>>>> | 390 | 330 330 | | >>>>>>> | | | GND >>>>>>> *--RR---*---RR---*---RR---*----+ >>>>>>> | | | | | >>>>>>> | | | | R 91k >>>>>>> === === === === R >>>>>>> 1u | 1u | 1u | 1u | | >>>>>>> | | | | | >>>>>>> GND GND GND GND GND >> >>>>> Batteries aren't a good solution for this, because the data runs are >>>>> very long, and they'd have to be charged up every so often, and >>>>> maintained, and they're a lot larger and more expensive as well. The >>>>> cap multiplier will fit in a board space the size of a nickel very >>>>> comfortably. (Other parts of the system do use batteries, mainly >>>>> because of pickup.) >> >>>>> Other EMI paths are probably going to dominate here as well, but once >>>>> the conduction via the supply is gone, we can nail those too. Low noise >>>>> sensor design is always an onion problem. >> >>>>> Cheers >> >>>>> Phil Hobbs >> >>>>> -- >>>>> Dr Philip C D Hobbs >>>>> Principal >>>>> ElectroOptical Innovations >>>>> 55 Orchard Rd >>>>> Briarcliff Manor NY 10510845-480-2058begin_of_the_skype_highlighting 845-480-2058 end_of_the_skype_highlighting >> >>>>> email: hobbs (atsign) electrooptical (period) nethttp://electrooptical.net-Hidequoted text - >> >>>>> - Show quoted text - >> >>>> Here is a circuit that uses a single darlington stage: >> >>>> http://sound.westhost.com/project15.htm >> >>>> Tom >> >>> We'll need to know the rapidity of the load variation in watt units, >>> then design a V-regulator to meet the nanovolt spec. >>> Personally, I think the cap amplifier is a primitive approach, instead >>> I'd go for a fast feed-back V-reg. >>> My background is designing voltages standard for labs, so I'm quite >>> interested in this discussion. >>> Regards >>> Ken >> >> Food and sex are pretty primitive too, but still widely popular. ;) >> >> I'm not very interested in the exact output voltage, because it's >> driving a bunch of op amps, and their PSR is effectively infinite at low >> frequencies. It's the SMPS ripple that's the problem, one that rears >> its ugly head in essentially every ultrasensitive measurement system. > > It's a good thing to solve. > >> If you actually possess a regulator design that will produce 1-nanovolt >> noise output with 1 V of ripple on the input, I'd love to see it. > > To start : (a nano is 10^-9), so we'll need at least 10^10 control, > within the tolerance braces. > Likely need staged 2x OP-amps for that, or 5x Darlington.Not going to happen with only 1.5V drop, though, or in 14 jellybean parts.> > Supposing we build that regulator, how would we measure it, > to ascertain we're in specs? > (shooting for 1 part in 10^10).No problem, using a lock-in amplifier. The ripple is a repetitive signal, so you can narrow the bandwidth as much as you like.> >> BTW by getting rid of the output resistor and putting a 12 ohm/47 uF >> lowpass section in the collector of Q1, you can get the same rejection >> with lower impedance and lower voltage drop. > > Yeah that's a course filter, gotta look at math, the number above are > the spec.It uses the excess headroom of Q1 to do some of the filtering, vs. adding more voltage drop at the end. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 email: hobbs (atsign) electrooptical (period) net http://electrooptical.net
Improved capacitance multiplier
Started by ●November 2, 2010
Reply by ●November 2, 20102010-11-02
Reply by ●November 2, 20102010-11-02
George Herold wrote:> On Nov 2, 1:48 am, Phil Hobbs<pcdhSpamMeSensel...@electrooptical.net> > wrote: >> So, I had this application requiring really absurd amounts of ripple >> rejection--it's a piezo driver in a laser wavelength locker that (for >> historical reasons) has to run off a fairly crappy dual isolated 48V >> DC-DC converter. The 48V has almost a half volt of ripple at 47 kHz, >> and the output noise needed to be in the nanovolts. >> >> As we've discussed here before, ordinary cap multipliers are limited by >> Early effect and collector-emitter capacitance. This one is a bit >> complicated--14 parts--but according to LTSPICE it provides 180 dB of >> ripple rejection from about 40 kHz to above 10 MHz. (In reality it >> won't be this good of course.) The interesting thing is that even with >> a it drops only about 1.5V, due to daisy-chaining the RC lowpasses--the >> second stage is biased from the base of the first stage instead of the >> emitter, which saves a diode drop. Adding the second stage requires >> only about 350 mV total. >> >> The remaining limitations are due to the combined Early effect of the >> transistors--very small, since they form a cascode pair--and all the >> interelectrode capacitances in series. >> >> Good medicine for the present problem! >> >> Cheers >> >> Phil Hobbs >> >> Q1 Q2 >> >> IN DNLS160 DNLS160 3 OUT >> >> 0-----*--- --------- ------RR-*--------0 >> | \ / \ / | >> | \ A \ A | >> | --------- --------- | >> 390 R | | === 47uF alum >> R | | | || 1uF X7R >> | 390 | 330 330 | | >> | | | GND >> *--RR---*---RR---*---RR---*----+ >> | | | | | >> | | | | R 91k >> === === === === R >> 1u | 1u | 1u | 1u | | >> | | | | | >> GND GND GND GND GND >> >> -- >> Dr Philip C D Hobbs >> Principal >> ElectroOptical Innovations >> 55 Orchard Rd >> Briarcliff Manor NY 10510 >> 845-480-2058 >> >> email: hobbs (atsign) electrooptical (period) nethttp://electrooptical.net > > Fun, I would have first tried taking the drive for the base of Q2 from > the collector of Q1, rather than the long string of RC's. Did you > perchance try it that way also? And was it worse? > > George H.It's quite a bit worse, because that makes the whole string only a second-order lowpass, vs. fourth order for the long string. Also you don't gain any headroom, because Q2 only drops 350-450 mV as it is, and you can't go a lot lower than that and still have any beta left. If this were a sane application, of course, I wouldn't need to turn the crank this tight, but it's still nice to see what you can get for less than the voltage drop of a vanilla Darlington cap multiplier. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 email: hobbs (atsign) electrooptical (period) net http://electrooptical.net
Reply by ●November 2, 20102010-11-02
On Tue, 02 Nov 2010 15:57:26 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>George Herold wrote: >> On Nov 2, 1:48 am, Phil Hobbs<pcdhSpamMeSensel...@electrooptical.net> >> wrote: >>> So, I had this application requiring really absurd amounts of ripple >>> rejection--it's a piezo driver in a laser wavelength locker that (for >>> historical reasons) has to run off a fairly crappy dual isolated 48V >>> DC-DC converter. The 48V has almost a half volt of ripple at 47 kHz, >>> and the output noise needed to be in the nanovolts. >>> >>> As we've discussed here before, ordinary cap multipliers are limited by >>> Early effect and collector-emitter capacitance. This one is a bit >>> complicated--14 parts--but according to LTSPICE it provides 180 dB of >>> ripple rejection from about 40 kHz to above 10 MHz. (In reality it >>> won't be this good of course.) The interesting thing is that even with >>> a it drops only about 1.5V, due to daisy-chaining the RC lowpasses--the >>> second stage is biased from the base of the first stage instead of the >>> emitter, which saves a diode drop. Adding the second stage requires >>> only about 350 mV total. >>> >>> The remaining limitations are due to the combined Early effect of the >>> transistors--very small, since they form a cascode pair--and all the >>> interelectrode capacitances in series. >>> >>> Good medicine for the present problem! >>> >>> Cheers >>> >>> Phil Hobbs >>> >>> Q1 Q2 >>> >>> IN DNLS160 DNLS160 3 OUT >>> >>> 0-----*--- --------- ------RR-*--------0 >>> | \ / \ / | >>> | \ A \ A | >>> | --------- --------- | >>> 390 R | | === 47uF alum >>> R | | | || 1uF X7R >>> | 390 | 330 330 | | >>> | | | GND >>> *--RR---*---RR---*---RR---*----+ >>> | | | | | >>> | | | | R 91k >>> === === === === R >>> 1u | 1u | 1u | 1u | | >>> | | | | | >>> GND GND GND GND GND >>> >>> -- >>> Dr Philip C D Hobbs >>> Principal >>> ElectroOptical Innovations >>> 55 Orchard Rd >>> Briarcliff Manor NY 10510 >>> 845-480-2058 >>> >>> email: hobbs (atsign) electrooptical (period) nethttp://electrooptical.net >> >> Fun, I would have first tried taking the drive for the base of Q2 from >> the collector of Q1, rather than the long string of RC's. Did you >> perchance try it that way also? And was it worse? >> >> George H. > >It's quite a bit worse, because that makes the whole string only a >second-order lowpass, vs. fourth order for the long string. Also you >don't gain any headroom, because Q2 only drops 350-450 mV as it is, and >you can't go a lot lower than that and still have any beta left. > >If this were a sane application, of course, I wouldn't need to turn the >crank this tight, but it's still nice to see what you can get for less >than the voltage drop of a vanilla Darlington cap multiplier. > >Cheers > >Phil HobbsUse 100 uF polymer aluminum caps and pick up another 160 dB! John
Reply by ●November 2, 20102010-11-02
"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message news:7mh1d6t4telqg2qsesdf6nfb0o1lnrj782@4ax.com...>>It's quite a bit worse, because that makes the whole string only a >>second-order lowpass, vs. fourth order for the long string. Also you >>don't gain any headroom, because Q2 only drops 350-450 mV as it is, and >>you can't go a lot lower than that and still have any beta left. >> >>If this were a sane application, of course, I wouldn't need to turn the >>crank this tight, but it's still nice to see what you can get for less >>than the voltage drop of a vanilla Darlington cap multiplier. > > Use 100 uF polymer aluminum caps and pick up another 160 dB!Or MIL-39006 tantalums. I wonder if you can get them with heavy gold plating. Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
Reply by ●November 3, 20102010-11-03
John Larkin wrote:> On Tue, 02 Nov 2010 15:57:26 -0400, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > >> George Herold wrote: >>> On Nov 2, 1:48 am, Phil Hobbs<pcdhSpamMeSensel...@electrooptical.net> >>> wrote: >>>> So, I had this application requiring really absurd amounts of ripple >>>> rejection--it's a piezo driver in a laser wavelength locker that (for >>>> historical reasons) has to run off a fairly crappy dual isolated 48V >>>> DC-DC converter. The 48V has almost a half volt of ripple at 47 kHz, >>>> and the output noise needed to be in the nanovolts. >>>> >>>> As we've discussed here before, ordinary cap multipliers are limited by >>>> Early effect and collector-emitter capacitance. This one is a bit >>>> complicated--14 parts--but according to LTSPICE it provides 180 dB of >>>> ripple rejection from about 40 kHz to above 10 MHz. (In reality it >>>> won't be this good of course.) The interesting thing is that even with >>>> a it drops only about 1.5V, due to daisy-chaining the RC lowpasses--the >>>> second stage is biased from the base of the first stage instead of the >>>> emitter, which saves a diode drop. Adding the second stage requires >>>> only about 350 mV total. >>>> >>>> The remaining limitations are due to the combined Early effect of the >>>> transistors--very small, since they form a cascode pair--and all the >>>> interelectrode capacitances in series. >>>> >>>> Good medicine for the present problem! >>>> >>>> Cheers >>>> >>>> Phil Hobbs >>>> >>>> Q1 Q2 >>>> >>>> IN DNLS160 DNLS160 3 OUT >>>> >>>> 0-----*--- --------- ------RR-*--------0 >>>> | \ / \ / | >>>> | \ A \ A | >>>> | --------- --------- | >>>> 390 R | | === 47uF alum >>>> R | | | || 1uF X7R >>>> | 390 | 330 330 | | >>>> | | | GND >>>> *--RR---*---RR---*---RR---*----+ >>>> | | | | | >>>> | | | | R 91k >>>> === === === === R >>>> 1u | 1u | 1u | 1u | | >>>> | | | | | >>>> GND GND GND GND GND >>>> >>>> -- >>>> Dr Philip C D Hobbs >>>> Principal >>>> ElectroOptical Innovations >>>> 55 Orchard Rd >>>> Briarcliff Manor NY 10510 >>>> 845-480-2058 >>>> >>>> email: hobbs (atsign) electrooptical (period) nethttp://electrooptical.net >>> >>> Fun, I would have first tried taking the drive for the base of Q2 from >>> the collector of Q1, rather than the long string of RC's. Did you >>> perchance try it that way also? And was it worse? >>> >>> George H. >> >> It's quite a bit worse, because that makes the whole string only a >> second-order lowpass, vs. fourth order for the long string. Also you >> don't gain any headroom, because Q2 only drops 350-450 mV as it is, and >> you can't go a lot lower than that and still have any beta left. >> >> If this were a sane application, of course, I wouldn't need to turn the >> crank this tight, but it's still nice to see what you can get for less >> than the voltage drop of a vanilla Darlington cap multiplier. >> >> Cheers >> >> Phil Hobbs > > Use 100 uF polymer aluminum caps and pick up another 160 dB! > > John >Wow, attovolt noise, kewl. Fame and fortune await. ;) Out beyond 100 kHz, it's limited by the ESR of the caps and the Early effect. 470 nF would work fine if it were a 100 kHz switcher. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 email: hobbs (atsign) electrooptical (period) net http://electrooptical.net
Reply by ●November 3, 20102010-11-03
On Tue, 02 Nov 2010 15:03:11 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>>>>> On Nov 1, 10:48 pm, Phil Hobbs >>>>> <pcdhSpamMeSensel...@electrooptical.net> wrote: >>>>>> So, I had this application requiring really absurd amounts of ripple >>>>>> rejection--it's a piezo driver in a laser wavelength locker that (for >>>>>> historical reasons) has to run off a fairly crappy dual isolated 48V >>>>>> DC-DC converter. The 48V has almost a half volt of ripple at 47 kHz, >>>>>> and the output noise needed to be in the nanovolts.Does "dual isolated" refer to Class 2 isolation (two isolation barriers) or two 48 V output power supplies in series in order to produce +/-48 V ?>I'm not very interested in the exact output voltage, because it's >driving a bunch of op amps, and their PSR is effectively infinite at low >frequencies. It's the SMPS ripple that's the problem, one that rears >its ugly head in essentially every ultrasensitive measurement system.Opt amps would suggest +/-15 V power supply. Is the capacitance multiplier only intended as a filter on the +48 V to +15 V line regulation and a separate filter (with PNP transistors) on the -48 V to -15 V side or is the Opt amp power supply voltages derived from a single filter ? Is that Gnp connection also used as signal reference ? Getting ripple down to 1 nV can be hard, if there are significant ripple currents flowing in other parts of the same ground plane on the input side of the filter. Circulating the ripple current as close to the power supply as possible (e.g. LC filters), will reduce the problems caused by circulating ripple currents especially in long connection wires, with their inherent impedances.
Reply by ●November 3, 20102010-11-03
Paul Keinanen wrote:> On Tue, 02 Nov 2010 15:03:11 -0400, Phil Hobbs > <pcdhSpamMeSenseless@electrooptical.net> wrote: > > > >>>>>> On Nov 1, 10:48 pm, Phil Hobbs >>>>>> <pcdhSpamMeSensel...@electrooptical.net> wrote: >>>>>>> So, I had this application requiring really absurd amounts of ripple >>>>>>> rejection--it's a piezo driver in a laser wavelength locker that (for >>>>>>> historical reasons) has to run off a fairly crappy dual isolated 48V >>>>>>> DC-DC converter. The 48V has almost a half volt of ripple at 47 kHz, >>>>>>> and the output noise needed to be in the nanovolts. > > Does "dual isolated" refer to Class 2 isolation (two isolation > barriers) or two 48 V output power supplies in series in order to > produce +/-48 V ?The supply is a Pico SIRF48D.> >> I'm not very interested in the exact output voltage, because it's >> driving a bunch of op amps, and their PSR is effectively infinite at low >> frequencies. It's the SMPS ripple that's the problem, one that rears >> its ugly head in essentially every ultrasensitive measurement system. > > Opt amps would suggest +/-15 V power supply.Yes, but this is a piezo driver, as I mentioned. It uses TI HV op amps.> > Is the capacitance multiplier only intended as a filter on the +48 V > to +15 V line regulation and a separate filter (with PNP transistors) > on the -48 V to -15 V side or is the Opt amp power supply voltages > derived from a single filter ? >It's a bridged amp design with a single-ended power supply. The output of the cap multiplier is about +46.5 V with a +48 V input.> Is that Gnp connection also used as signal reference ?This is a differentially-connected piezo driver, but at some point the CMR of the op amps starts to be important.> > Getting ripple down to 1 nV can be hard, if there are significant > ripple currents flowing in other parts of the same ground plane on the > input side of the filter. Circulating the ripple current as close to > the power supply as possible (e.g. LC filters), will reduce the > problems caused by circulating ripple currents especially in long > connection wires, with their inherent impedances. >Yes, there will almost certainly be some inductively-coupled junk. I tested with a scope probe shorted via its ground lead--the pickup is worse with the ground lead wound in two turns than with one turn, and mostly goes away when you reverse one turn by twisting it 180 degrees out of the plane (so that the intercepted flux cancels instead of adding). That last test pretty well rules out capacitive effects as the leading cause. Ultrasensitive instrument design is an onion problem, so when you find one nasty, you nail it to the floor and move on to the next one. If this client gets as far as needing cost reduction, we'll see if we can shave one or even two poles off the cap multiplier. It won't be the first cost reduction step, because the parts are all cheap. (I thought this design was interesting because of its favourable combination of high rejection and relatively low voltage drop, not because it's a complete solution to all SMPS problems.) Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 email: hobbs (atsign) electrooptical (period) net http://electrooptical.net
Reply by ●November 3, 20102010-11-03
On Wed, 03 Nov 2010 03:52:19 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:>It's a bridged amp design with a single-ended power supply. The output >of the cap multiplier is about +46.5 V with a +48 V input.With bridged configurations, shouldn't extreme opposite symmetry between the left and right side layout help in reducing differential ripple voltages ? Anyway, 1 nV ripple requirement seems a bit strange. At room temperatures and 50 kHz bandwidth, the noise power (kTB) is -127 dBm. Thus, the source resistance would have to be a few milliohms. Is this realistic in practice ?
Reply by ●November 3, 20102010-11-03
On Wed, 03 Nov 2010 18:50:48 -0500, Jamie <jamie_ka1lpa_not_valid_after_ka1lpa_@charter.net> wrote:>Tim Williams wrote: > >> "John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in >> message news:7mh1d6t4telqg2qsesdf6nfb0o1lnrj782@4ax.com... >> >>>>It's quite a bit worse, because that makes the whole string only a >>>>second-order lowpass, vs. fourth order for the long string. Also you >>>>don't gain any headroom, because Q2 only drops 350-450 mV as it is, and >>>>you can't go a lot lower than that and still have any beta left. >>>> >>>>If this were a sane application, of course, I wouldn't need to turn the >>>>crank this tight, but it's still nice to see what you can get for less >>>>than the voltage drop of a vanilla Darlington cap multiplier. >>> >>>Use 100 uF polymer aluminum caps and pick up another 160 dB! >> >> >> Or MIL-39006 tantalums. I wonder if you can get them with heavy gold >> plating. >> >> Tim >> >You like flames? > >Tants would be safe in that circuit, at least the ones on the bases. But polymer alums have absurdly low ESRs. John
Reply by ●November 3, 20102010-11-03
Tim Williams wrote:> "John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in > message news:7mh1d6t4telqg2qsesdf6nfb0o1lnrj782@4ax.com... > >>>It's quite a bit worse, because that makes the whole string only a >>>second-order lowpass, vs. fourth order for the long string. Also you >>>don't gain any headroom, because Q2 only drops 350-450 mV as it is, and >>>you can't go a lot lower than that and still have any beta left. >>> >>>If this were a sane application, of course, I wouldn't need to turn the >>>crank this tight, but it's still nice to see what you can get for less >>>than the voltage drop of a vanilla Darlington cap multiplier. >> >>Use 100 uF polymer aluminum caps and pick up another 160 dB! > > > Or MIL-39006 tantalums. I wonder if you can get them with heavy gold > plating. > > Tim >You like flames?
