Your comment about be being way below 0.65v quiescent tells me you're running class B. Like that it can be stable enough. If you ran it AB as I thought you were it would be very unstable. Why...
Trs idle with 0.65v Vbe, much of opamp's Iq going through the Rs, some to tr bases.
As Trs warm up under exercise, their Vbe drops
You then have more Ib and less I through the psu/base R
So the trs turn on more, get hotter and turn on more. Vicious cycle with no way to reduce quiescent current or overlap of drive to the 2 trs.
LT Spice lets you declare a diode to have a constant forward votage drop, which I guess the zener models have. That can be handy sometimes but it's not real.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
They allow the opamp to run within its voltage rating, but still shoot its supply currents up to the +-16 volt rails.
Those just let us shut down the output stages when the customer wants zero magnetic field, like when he's shimming the superconductive magnet.
D6 and D7 add an output deadband that further reduces current and noise when we want zero field. NMR has parts-per-billion sensitivity, so tiny zero offsets matter.
We don't do NMR instrumentation any more. Agilent acquired our customer, Varian, and eventually killed off the NMR operation. I think other analytical chemistry techniques have mostly replaced NMR; the big magnets were really expensive to buy and operate. Agilent also killed off the Varian FTMS products, which needed even bigger magnets. We were developing a cool FTMS controller too.
formatting link
formatting link
Science is fun, as long as you don't have to do the grad-postgrad-PhD-postdoc-publish thing yourself.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
I met a few people laid off from Varian's NMR group. I know about quadrupole* mass spectroscopy.. kinda cool how it all works. I'm not sure about the Fourier transform part.
I suck at writing, but like to make pretty pictures for publication. The worst part of research is grubbing after the money.
George H.
*I want to spell that quadrapole, quadrupole sounds like you've multiplied your mass spectrometers by four.
I'm bored so here's a very brief intro to FTMS. It is based on ion cyclotron resonance. Start with an ion in a vacuum in a uniform magnetic field. In the direction of the field there is no force on the ion so it is free to keep drifting with whatever velocity it started with. Perpendicular to the magnetic field the ion experiences the usual E cross B force, making it's path curve. If the B field is strong and the velocity low relative to the mass to charge ratio m/q the path becomes a circle and you can show that the time to traverse that circle is 2pi m / q B or the cyclotron frequency in hertz is q B / 2 pi m. Put the ion in a box and bias the ends that are perpendicular to the magnetic field with a few volts of the same sign as the charge on the ion relative to the other walls and you get a potential well inside the box that traps the ion along the magnetic field, and the magnetic field does the trapping along the other two axes so now you have a trapped ion cell. Apply rf perpendicular to the magnetic field and the ion will absorb energy and increase its cyclotron radius. If there is more than one ion of the same m/q they will each move in phase with the applied rf, and if you pump them up to a final cyclotron radius that is much larger than the initial radius before the excite they will each have approximately the same final cyclotron radius and the cloud will orbit coherently. Pick a point on one of the cell walls perpendicular to the magnetic field and as the cloud approaches and then recedes it will induce an image charge that can be detected. The simplest trap is a cube, with six sides. Two end plates do the axial trapping, one pair of opposing side plates is used to apply the excitation rf, and the other pair of opposing side plates is connected to a differential preamp to record the signal. For magnets in the 1-7 tesla range and m/q ratios in the 18 to say 10,000 range the frequencies are in the 1-5 kHz to 5-10 MHz range, and for ion populations of say 1000-1000000 the raw signal is in the microvolt range with cyclotron radii of .5 to 2 cm. In the second picture John posted the tubular object with the ruler next to it is a trapped ion cell. There are two end segments 2" long and a central segment 3" long. The end segments can be simple cylinders but are often split into four quadrants, and the central section is split into four quadrants to provide the excite and detect plates. Cells can be cubic, rectangular, cylindrical, and other shapes, and the length to diameter aspect ratio can vary also, and that discussion fills books and fuels arguments. FTMS supercon magnets usually have room temperature bores of
4-6" so by the time you stuff a vacuum chamber into one a rectangular cell winds up maybe 1.5 to 2.5" across and a cylindrical cell maybe 2-3" diameter.
Anyway, if there is just one m/q in the trap then you will get a simple sine wave after the excite. The amplitude tells how many ions and the frequency tells the m/q. As the ions collide with background gas the amplitude will exponentially decay as individual ionis are knocked out of phase coherence or into plates, with a time constant of about 3 seconds at 10-8 torr. Record the damped sine wave, FFT it, convert Hz to m/q, and there's a mass spectrum. In FTNMR the excitation is a pulse of single frequency rf so the final phase is linear and can be easily corrected but here the frequency range is sufficiently large that the simplest excitation is a swept sine wave, which results in a quadratic phase function after the FFT so the usual answer is to use the magnitude spectrum. Excite chirps are in the range of
0-3 MHz in 1-5 msec applied differentially to the trap so maybe 50-400 volts pp across the trap.
It's been a while since I've given an "intro to ftms" and I tried to keep it short and terse, but I hope it answers the basics. Haven't touched one lately but counting grad school I spent about 30 years working with them, building two completely from scratch, building several different vacuum chambers, ionization sources, and transport ion optics for others, along with a good bit of electronics, computer interfacing, and software to go along with them so if you have any questions just ask :-).
I wanted to detect a single molecule orbiting in the cell, which I think is barely possible. Agilent killed the product line (IonSpec originally, acquired by Varian, acquired by Agilent, killed by Agilent) so I didn't get to try.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Hah! Thanks Carl. When I was a postdoc we detected hole CR in GaAs quantum wells. Some of my favorite data... I've kept a copy taped to my wall. Detection was complicated. We monitored the visible absorption spectra as we blasted the sample with an FIR laser. ~20 cm-1, 500 um.
Bored? I don't suppose you know anything about atomic physics? Farting around at a workshop (for teaching others to use our equipment) I recreated some work I did ~10 years ago. Observing the Zeeman splitting of a Rubidium absorption line. And now, when I tried to explain why it worked, I just make myself more confused. It will take me a while to organize the data and explain my conundrum. (So not today.) I was going to post on Sci.optics, maybe here too?
The UW "geonium" project back in the late '70s, iirc, did much the same thing with a single electron.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
Hah! Thanks Carl. When I was a postdoc we detected hole CR in GaAs quantum wells. Some of my favorite data... I've kept a copy taped to my wall. Detection was complicated. We monitored the visible absorption spectra as we blasted the sample with an FIR laser. ~20 cm-1, 500 um.
Bored? I don't suppose you know anything about atomic physics? Farting around at a workshop (for teaching others to use our equipment) I recreated some work I did ~10 years ago. Observing the Zeeman splitting of a Rubidium absorption line. And now, when I tried to explain why it worked, I just make myself more confused. It will take me a while to organize the data and explain my conundrum. (So not today.) I was going to post on Sci.optics, maybe here too?
George H. =======================================================
Sorry, George, can't help with much physics theory.
The UW "geonium" project back in the late '70s, iirc, did much the same thing with a single electron.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
===================================================================
The usual wide m/q range analytical mass spectrometer FTMS system uses a
matched pair of FET's in a cascode differential pair for the input and have
noise levels in the .5-1.5 nV/rt Hz range with bandwidths up to 5-10 MHz,
with input impedances of 1-1000 MOhm. There are several designs in the open
literature, with the one by Gordon Anderson of PNNL probably the most
popular. With traps in the 2-4" range and fields of 7 T it can detect under
a hundred charges. They "cheated" and used a highly charged 100 megadalton
piece of DNA and detected a single molecule but it had thousands of ions
like Na+, K+, and H+ hanging on. They could even see the cyclotron
frequency jump as individual cations fell off. The geonium work, and
further advances by Gerry Gabrielse's lab at Harvard (and others, of course)
could detect a single electron or positron but they cooled their trap and
preamp to liquid helium temperature and used a LC tuned circuit so they
could only detect the single m/q that they designed the whole system to work
with. The IonSpec system that John was starting to work with cooled the
trap and preamp with the 77 K stage of a closed cycle helium refrigerator
that was already there to power a cryopump. I never heard an actual spec on
that preamp but I heard some uncomplimentary things said about it over the
years by various researchers. I don't think it was ever changed from their
first design in about 1980.
The ESR of the Zener and PN Diode are related to the inverse of their power ratings and should match the power rating of the power rating of the R_be string being biased to match the string of ESR's or Zzt's.
Also critical for thermal stability is thermally coupling to track NTC Shockley effects of bias diodes to load diodes.
There's more to this but, that's the main criteria, not the arbitrary comparison of a Zener vs an ordinary diode.
I've been visiting Sandia for the last couple of days. I was talking with a low-temperature guy who apparently uses SiGe BJTs at 4K. JFETs deteriorate quite badly down there, and other bipolars become open circuits due to car rier freeze-out, so I was quite surprised to hear that heterojunction bipol ars keep working.
The IonSpec preamp that I saw was uncooled. Here it is:
formatting link
It's a pair of fet opamps running as followers. They have no cooling in vacuum except radiation and run at roughly 120C. Gate current was so high that they needed 1Meg resistors to ground on the inputs. Shot and Johnson noise big-time. I figured they gave up at least 30 dB.
Chemists are maybe the worst circuit designers.
I had a pretty promising and suitably weird design but the whole thing died.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
The IonSpec preamp that I saw was uncooled. Here it is:
formatting link
It's a pair of fet opamps running as followers. They have no cooling in vacuum except radiation and run at roughly 120C. Gate current was so high that they needed 1Meg resistors to ground on the inputs. Shot and Johnson noise big-time. I figured they gave up at least 30 dB.
Chemists are maybe the worst circuit designers.
I had a pretty promising and suitably weird design but the whole thing died.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
========================================================
Hmm, I never got to take an IonSpec system apart but I was told by more than
one person that that first stage was attached to the 77 K heat shield. I
was at UC Riverside, just up the road from UC Irvine where Robert McIver,
founder of IonSpec, was on the faculty. We had Nicolet systems and he
visited a few times to see what we were doing. It was very frustrating
because we would basically tell him anything he asked, and show what we
could depending on what we had torn down at the time, but his answer to any
question we asked was "Sorry, that's proprietary". Pretty much every preamp
before Gordon's used 1 Meg input resistors to ground. Going bigger helps
s/n some but not that much, and you have to worry about the input cable
capacitance (50-100 pF by the time you got outside the vacuum system to the
preamp since only IonSpec put that first stage inside) and time constants to
recover from feedthrough of the big excite pulse. Ideally you want to be
able to start digitizing the signal in a msec or so. That's not so critical
at very low pressures but if you need to run up in the 10-6 torr range doing
ion molecule chemistry the signal damps away in 10-50 msec so a slow preamp
recovery really hurts your signal. But what do I know about circuit design,
I'm just a chemist (who was about 6 classes short of a BSEE and 3 math and
physics classes short of a BS Comp Sci so I took that as a second major
:-)).
a low-temperature guy who apparently uses SiGe BJTs at 4K. JFETs deteriora te quite badly down there, and other bipolars become open circuits due to c arrier freeze-out, so I was quite surprised to hear that heterojunction bip olars keep working.
Huh, maybe they would work as low temperature, temperature sensing diodes?
Re: Cable capacitance, sounds like a job for a driven shield. I made a driven shield to look at the Johnson noise (R~1k - 1M) of resistors dwon the bottom of a probe. It didn't make into the final instrument, but worked fine.
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.