I guess I was thinking about bootstrapping the end of the resistor that is connected to the inverting input... I'm not sure how to do that and not cause more leakage current.
The app note found by Chris Jones talks about the guard band improving the time constant.
George h.
That file crashes my Foxit reader.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Special case: all PDF files crash my Foxit reader. I applied the universal software fix: reboot Windows.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Phil Hobbs: Yes, but why did you think so? It can sometimes provide a somewhat different view of circuits. Several compensation systems required. George Herold: You need to go back a long time for early compensation techniques (to the days of tubes) e.g.: Pelchowitch I, Zaalberg van Zelst J J 1952 A wide-band electrometer amplifier Rev. Sci. Instrum. 23, 73-5, Praglin J, Nichols W A 1960 High-speed electrometers for rocket and satellite experiments Proc. I.R.E. 48, 771-9 John Larkin: There are epoxy coated 100T resistors (previously Hydrazine but now Japan Fine Chemical Co though their latest version has a ceramic sleeve outer). However the price is about ten times the glass types.
Scott Hamilton.
Because only a physicist would use a resistor that large. My usual example of physicist circuit design is putting a gigohm resistor on a cable. ;)
I agree. Being a bit of a chameleon that way myself, I do crank out some idiosyncratic stuff, which usually works very well. I did do a fairly long apprenticeship in RF PLL design among other things.
Like what? Sounds interesting.
Cheers
Phil Hobbs
This is cool:
The glass resistors seem to usually be $30 to $60 from distributors. Are there cheaper ones?
-- John Larkin Highland Technology, Inc lunatic fringe electronics
I've bought Russian ones off eBay.
Cheers
Phil Hobbs
e of physicist circuit design is putting a gigohm resistor on a cable. ;)
idiosyncratic stuff, which usually works very well. I did do a fairly long apprenticeship in RF PLL design among other things.
You do not have much choice in resistor size for 100T; they are all about t he same size and size does have some circuit advantages. Henry Ford is repu ted to have said 'You can have any colour you like as long as it is black'. If you mean by 'large' high value I will explain if you wish. I do have I think rather longer and wider experience than many physicist/el cctronicists; see for example ISBN 978-0-521-68780-5. Regards, Scott.
Sure, tell us all about it. Sounds interesting.
Looks like a potentially interesting read. I ordered a copy off abebooks.com, so we'll see in a couple of weeks.
When I talk about the way physicists approach circuit design, I don't mean that some of them aren't very good at it. Paul Horowitz is a physicist, and Jan Hall (2005 Nobel) designed all his own stuff, just for starters.
However, there's a certain commonality of style that EEs find a bit weird.
Cheers
Phil Hobbs
Thanks Scott, I'll check out the references. The guard band is a bit like a driven shield.. which is something I've played with before.
George H.
ple of physicist circuit design is putting a gigohm resistor on a cable. ;)
e idiosyncratic stuff, which usually works very well. I did do a fairly lon g apprenticeship in RF PLL design among other things.
the same size and size does have some circuit advantages. Henry Ford is re puted to have said 'You can have any colour you like as long as it is black '. If you mean by 'large' high value I will explain if you wish.
elcctronicists; see for example ISBN 978-0-521-68780-5.
Hah, I've got your book on my shelf... sadly I only browsed it and then fo rgot about it! (I think I read a nice review in Physics Today or someplace when it came ou t.)
I'll read section 5.12 with renewed interest. (5.12 Fast operational picoa mmeter.)
George H.
His book is on my shelf... not sure why it just sat there, but I'll tuck it into my backpack. It starts out with a lot of basic stuff. ~100 pg's of math and then ~50 pgs of physics... So lots of background till the "good stuff".
George H.
The Rad Lab pretty much invented modern electronics, and it was mostly physicists. On the other hand, The Review of Scientific Instruments is often laughable. I think the bottom line is that if you don't have a native talent for circuit design, you shouldn't do it no matter what your titles are.
Not specific to anyone here, but I've seen the worst horrors done by chemists.
The guard on a high-ohms resistor needs to be distributed so it can have continuous AC compensation. I had that problem a while back and discussed it here.
It would be cool if we could see the electric potential along a 100T glass resistor as a voltage step is applied to one end.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
With some major help from the Brits (also mostly physicists). Fixed and air borne radar, magnetrons, radio navigation, etc.
Other applied research journals aren't far behind. I often get inquiries ci ting RSI-published designs for things like op amp TIAs.
sts.
My least knowledgeable noise canceller customers have been chemists, but th ere are also a bunch of really smart ones, e.g. my old colleague RE Moerner .
The clue-challenged ones never can understand why, when running in log-rati o mode, they should crank up the feedback bandwidth and filter afterwards. It wouldn't matter if they'd take advice, but those ones won't.
For non-EO types, it's because in a nonlinear circuit there are a lot of na sty parametric effects if you have large signals up near the loop bandwidth . Normal log amps have that problem too--the bandwidth is very different on the two half-cycles, leading to big DC offsets and suchlike.
I recall. In high-Z designs I'm usually pushing the megahertz per nanoamp e nvelope pretty hard, and unfortunately distributed guarding doesn't help th e noise--in fact it makes it much worse. A pity.
Cheers
Phil Hobbs
I was doing a 1200 volt pulse scope-monitor pickoff, obviously not a noise problem. I wound up using a couple of 22M resistors in series and doing some equalization in the opamp downstream, a serious sell-out on the issue but good enough for the customer, who was getting the monitor output as a freebie anyhow.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Sure, I do a lot of that too. The key advantage from my POV is that it doesn't hurt the SNR.
Cheers
Phil Hobbs
irborne radar, magnetrons, radio navigation, etc.
citing RSI-published designs for things like op amp TIAs.
mists.
there are also a bunch of really smart ones, e.g. my old colleague RE Moern er.
tio mode, they should crank up the feedback bandwidth and filter afterwards . It wouldn't matter if they'd take advice, but those ones won't.
nasty parametric effects if you have large signals up near the loop bandwid th. Normal log amps have that problem too--the bandwidth is very different on the two half-cycles, leading to big DC offsets and suchlike.
envelope pretty hard, and unfortunately distributed guarding doesn't help the noise--in fact it makes it much worse. A pity.
Hmm Well with Scott's permission, I'll post figure 5.12.5a. (AFAICT it doesn't add to the noise... it does load the opamp more.... TIA but with compensating RC low pass driving the FB RC. which rings like the dickens... so he adds some dissipation to the compensating C. )
George H.
Okay, I've used that one too--it's good for a factor of 10 or so, and as yo u say it doesn't hurt the noise. The distributed guard trick can give much more of a speedup, but the current noise of the guard gets into the summing junction via the capacitance.
Cheers
Phil Hobbs
you say it doesn't hurt the noise.
Huh, OK It's new to me. bad ascii art for those following along,
+--Cf---+ | | +--Rf---+--C1--R2--GND | | | |\ R1 i_in---+-|-\ | | >--+---out GND--|+/ |/ R1*C1= Rf*Cf, Scott derives a complicated expression for R2 (which includes opamp GBW and input C.)So what's the downside? Do you lose dynamic range?
George H.
The distributed guard trick can give much more of a speedup, but the curren t noise of the guard gets into the summing junction via the capacitance.
Mainly that it costs you gain bandwidth, and that the feedback resistor RC model is only good to leading order, so the step response starts going squi rrelly if you push it too far.
I used that approach in the nanoamp TIA product that we'll be introducing a s soon as we figure out all the NY state tax and licensing hoops we have to jump through. It gets 1 MHz bandwidth with a nice clean step response, us ing a 10M feedback resistor. The BW of the TIA is wider than that, but I pu t in a 4-pole Gaussian filter to knock out most of the noise peak.
Cheers
Phil Hobbs
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.