C-multiplier again
in [Design]
|
Prev: Futuristic weapon question: Anti-matter-gun, would it have a signature ?
Next: EEVblog Live Event
From: George Herold on 27 May 2010 21:37 On May 27, 9:31 pm, George Herold <gher...(a)teachspin.com> wrote: > On May 27, 2:32 pm, Mike <s...(a)me.not> wrote: > > > George Herold <gher...(a)teachspin.com> wrote: > > > OK this is still a sim. However I spiced the C-multiplier (aka Hobbs > > > filter at teachspin) I use to get down to ~1nV. I haven't every > > > bumped into the Early effect (I don't do much on the transistor > > > level.) But this shows attenuation down near 90dB at 100kHz. (Which > > > is where the switcher I use lives.) > > > [...] > > > Seems to be missing some components. I got the 2N4401 and what appears to > > be a AC source in series with 15V. Everything else is just gaps. I browsed > > the file but could not find any references. > > Well that figures, It's my first attempt to post a ltspice circuit. > > Not much to it. a two pole low pass RC filter (R-C to ground. feeds R- > C to ground) and this feeds the base of a npn. the output has a > minimum 1 k ohm to ground with anoter C. R is 1k and C is 100uF Al > electro. There's also 10 ohms between the final R-C cap node and the > base, but this is only to stop oscilations and doesn't change the > spice sim much. I'm going to play hookey (sp) from work tomorrow and > get the homestead ready for our annual Memorial Day party. (Else I'd > try and repost) Summer's here early in Buf. NY. If you're in the > area I grill up some great Polish sausage with Walt's mustard. (Walt > lives down the road.) > > George H. > > > > > > > > George H. > > > Oh, and thanks again Mike for showing me how to post and read the > > > LTspice stuff... I feel so empowered. > > > You are welcome. > > > Mike- Hide quoted text - > > - Show quoted text - Geesh lots of typos there, please forgive me (I'm a terrible speller) George H.
From: John Larkin on 27 May 2010 21:40 On Thu, 27 May 2010 22:23:47 GMT, Mike <spam(a)me.not> wrote: >John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >> >> But come to think of it, an SSR could be a 4-ohm replacement for the >> 50 ohm thing, with correspondingly less Johnson noise. >> >> John > >SSR - Solid State Relay? Too slow to protect the inputs, even if you could >design a circuit to detect a fault instantly. The Clare parts I'm using - to protect electronics - turn off in anout 400 usec. > >Back-to-back schottkys across the inputs, with perhaps 2 ohms in series to >the electrolytics, might be the best solution. Hopefully the schottkys >would turn on fast enough to do the job. Big ones would probably add noise. Little ones wouldn't protect. John
From: Mike on 27 May 2010 22:11 John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: > On Thu, 27 May 2010 22:23:47 GMT, Mike <spam(a)me.not> wrote: > >>John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>> >>> But come to think of it, an SSR could be a 4-ohm replacement for the >>> 50 ohm thing, with correspondingly less Johnson noise. >>> >>> John >> >>SSR - Solid State Relay? Too slow to protect the inputs, even if you >>could design a circuit to detect a fault instantly. > > The Clare parts I'm using - to protect electronics - turn off in anout > 400 usec. What are you protecting, and against what? MOSFET thermal overload? What's the time constant? Seconds? You have plenty of time. How long does it take to destroy a junction in a 1GHz op amp? Nanoseconds? Picoseconds? The AD797 input bias current is 250nA, so it's probably bipolar. Perhaps the problem is not so much destroying the junction as it is breaking it down in the reverse direction. I understand this increases the device noise. If it's reverse breakdown, it may only take a few nanoseconds. >>Back-to-back schottkys across the inputs, with perhaps 2 ohms in >>series to the electrolytics, might be the best solution. Hopefully the >>schottkys would turn on fast enough to do the job. > > Big ones would probably add noise. Little ones wouldn't protect. Zero volts across them. Zero current. Probably high capacitance, maybe 1nF. In parallel with large electrolytics. What's to add noise? Little ones may be fine. The 11DQ05 will take 150 Amp surge, so all it needs is a high enough value resistor in series to keep the voltage drop under the AD797 spec: http://www.datasheetcatalog.org/datasheet/irf/11dq05.pdf Or use a more modern op amp that has lower noise and higher max input differential voltage. Or put a 10 / 0.025 = 400 ohm resistor in series with the input. That should keep the current under 25mA:) > John Mike
From: John Larkin on 27 May 2010 23:22 On Fri, 28 May 2010 02:11:32 GMT, Mike <spam(a)me.not> wrote: >John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: > >> On Thu, 27 May 2010 22:23:47 GMT, Mike <spam(a)me.not> wrote: >> >>>John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>> >>>> But come to think of it, an SSR could be a 4-ohm replacement for the >>>> 50 ohm thing, with correspondingly less Johnson noise. >>>> >>>> John >>> >>>SSR - Solid State Relay? Too slow to protect the inputs, even if you >>>could design a circuit to detect a fault instantly. >> >> The Clare parts I'm using - to protect electronics - turn off in anout >> 400 usec. > >What are you protecting, and against what? MOSFET thermal overload? It's a 12-channel isolated 4-20 mA input/output board. It's got a bunch of resistors, transzorbs, and mosfets in harm's way. We're allowing the customer to apply up to 50 volts at either polarity. There are a couple of 1206 resistors that dissipate something like 20 watts each at max overload, with the current limited by the fets inside the SSR. An ARM processor (one per channel!) is digitizing everything at about 100 KHz, notices the overload, and turns off the SSR. I figure the resistors are OK for a millisecond maybe, so we've got to turn things off before then. > >What's the time constant? Seconds? You have plenty of time. > >How long does it take to destroy a junction in a 1GHz op amp? >Nanoseconds? Picoseconds? Microseconds at least, if the mechanism is thermal. You could vaporize metalization a lot faster, with enough energy. > >The AD797 input bias current is 250nA, so it's probably bipolar. Perhaps >the problem is not so much destroying the junction as it is breaking it >down in the reverse direction. I understand this increases the device >noise. If it's reverse breakdown, it may only take a few nanoseconds. That opamp has back-to-back diodes across its inputs and no current limiting resistors. The problem is too much current from charging those big caps. > >>>Back-to-back schottkys across the inputs, with perhaps 2 ohms in >>>series to the electrolytics, might be the best solution. Hopefully the >>>schottkys would turn on fast enough to do the job. >> >> Big ones would probably add noise. Little ones wouldn't protect. > >Zero volts across them. Zero current. Probably high capacitance, maybe >1nF. In parallel with large electrolytics. What's to add noise? Leakage resistance maybe, and capacitance adding a pole to the feedback loop. I did say "probably." > >Little ones may be fine. The 11DQ05 will take 150 Amp surge, so all it >needs is a high enough value resistor in series to keep the voltage drop >under the AD797 spec: > >http://www.datasheetcatalog.org/datasheet/irf/11dq05.pdf Might work. 100 pF, maybe more, at 0 volts. > >Or use a more modern op amp that has lower noise and higher max input >differential voltage. > >Or put a 10 / 0.025 = 400 ohm resistor in series with the input. That >should keep the current under 25mA:) Mr Johnson would object. John
From: whit3rd on 28 May 2010 01:04
> >I have an interesting idea. How about a blue LED as the reference. > I just used two diodes in series to make a low-noise -1.5 volt > shunt-type supply. I could have used an LED, which would be cool - > they light up! - but I didn't want any stray light inside our box. GAAA! If you want low noise from AN OPTICALLY OPEN DIODE you need to shield from light. Incandescent will cause 120 Hz input, fluorescent 60 Hz and 120 Hz, electronic ballasts and CCFL can go from kilohertz to megahertz. Use a 1.25V bandgap reference like real folk do. |