Vacuum tubes and Electromagnets: HR100 & HA100
Work in progress! Photographs will be inserted soon!
The practical limit for electromagnets is 2.35T, corresponding to 100 MHz for 1H resonance, and the top-level instrument in the world over (no competitors at that time!) was the HR100, or High Resolution 100 MHz; let’s have a look at it, from left to right.
Photograph from “Varian NMR Story”, an advertisement edited by Varian Inc. shortly before the acquisition from Agilent, hereinafter quoted as “Varian Story”
– Electronics console, in the “single floor” version; clearly visible are the Low Frequency Signal Generator (HP100D foto) and the HP Oscilloscope ; the next rack unit is RF Unit, housing 100MHz Transmitter and Receiver; no chart recorder of any kind in sight, hence we have to assume the spectrum was only observed on the oscilloscope screen and peak positions recorded by writing on a sheet of paper. A (rough!) calibration came from the sideband technique, modulating the field with ab audio frequency of known value, coming from the HP100D, that produced sidebands on the spectrum. The Sideband Modulation technique was widely used on all future Varian spectrometers to obtain Field/Frequency Lock.
– 2.35T Electromagnet with, inserted between the pole caps, the Probe with its coax connection cables to the RF Unit; notice on the probe the housing of the vacuum tube of the first stage of signal amplification, a special quality tube with low noise figure; this amplifier, located very close to the signal receiver coil, was almost essential with tube electronics to improve signal-to-noise ratio. At magnet top-right you ca see another interesting device of the magnet: the kind of handle is the “ultra-coarse” Y homogeneity adjustment, that simply applies pressure on the magnet frame. (On electromagnets and permanent magnets the sample is spun around Y axis, hence the most critical axis to adjust is the Y and not the Z like as in supercons). Other shim control knobs were located on a small panel in the console; controls where simply 5 or 6 potentiometers regulating the currents on pairs of coils mounted on the pole caps.
An interesting and maybe little known historical point. The very first step in otimizing the homogenety of an electromagnet in the factory was to make the pole caps exactly parallel, and this was done inserting thin plates of steel, or “shims”, under the pole pieces inner side; the magnet was therefore “shimmed” to good homegeneity, and the term survived up to today.
At the end of the 60’s, Varian had two competitors in the NMR market, Jeol and Perkin-Elmer; P-E introduced 60 MHz (R-12) and 90 MHz (R-32) permanent magnet spectrometers, and used the term “Golay” for the shim coils, I guess from the name of the designer.
– On top of the magnet sits the Flux Stabilizer; books could be written around this unit, it’s enough to say it was the nightmare of the spectrometer’s operator. Stories on it will follow on the HA100 section.
– Beside the magnet is the Cooling Unit, including a water tank filled with distilled water, a pump circulating water in an inner loop to cool down the magnet and a water-to-water heath exchanger with a simple temperature regulator to keep the temperature constant: quite a lot of tap water was needed.
– The 2,500 V – 5 A Magnet Power Supply; perfect for 100% lethal electrocution of the poor service engineer! Usually it was located in a locked room nearby, but this made the search for the signal even more complex; two men needed, and a openining in the wall, through which orders were shut: up! down! slow! STOP!!!
To make this possible for a single man, in some NMR labs a mirror system was installed, to make it possible for the man at the controls to watch the oscillosope screen. Needless to say, the mirror’s alignment was crytical and unstable, and needed almost daily calibration, but this was not care of the spectrometer’s Service Engineer (who was me…)
Photo courtesy Prof. A.Mazzanti, Univ. Bologna (coming soon, some technical problem to insert it!)
This a shot of the HA100 spectrometer in an early version, 1st floor plus mezzanine, took in the 1960s in the Industrial Chemistry Dept, Bologna University; the standing young man is Prof. Ludovico Lunazzi, who kept running Varian machines up to 600 MHz until (and even after!) his retirement.
The first NMR Spectrometer I met in 1970 was a 60 MHz version of the HA100, named DA60-IL that stood for Dual-purpose, Analytical, 60 MHz, Internal Lock. It was a huge beast with 3 rack units, 2-floor cabinets, and included modules for high-resolution plus wide-line NMR, and all switching units needed. To obtain a stable NMR lock (homonuclear lock on TMS!) the poor operator had to do a dozen of coordinated operations, done in the right sequence, and a kind of dance between pushbottons, switches and knobs. Quite a good training ship, with Prof. Ivano Bertini standing behind my shoulders and screaming orders and imprecations.
?TO BE MOVED SOMEWHERE ELSE?
The original HA100 was improved and modified in several parts, the first being the magnet itself and its power supply; the magnet coils became low impedance type, wound in copper foils and therefore with very low resistance (1 ohm typical each); the magnet power supply was low-voltage, high current (120 V, 80 A max), transistorized (Motorola 2N174, one of the first devices manufactured by Motorola). Several magnets were modified, which implies the replacement of magnet coils, not an easy task, two servicemen where busy for at least tw0 weeks.
The magnet pole pieces were removed, coils replaced, pole pieces re-installed and magnet shimmed back to meet lineshape specifications. The first 3 steps where hard mechanical work, the last operation almost a nightmare. I still remember such a job done in Bari, southern Italy, end of July/first weeks of August; outside temperature was 40+ °C, and room air conditioning didn’t work very well. Coarse shimming was done by adjusting the torque on pole pieces fixing bolts, and the other guy, which was much heavier and stronger than me, did the hard work, while most of the time I was looking for results at the oscilloscope. The poor guy was sweating all the time, and with time a small pool of sweat developed on the floor under his feet. From time to time I took my turn at the long bar and did my best to adjust the torque, with very scarce results. We managed to obtain a
reasonable resolution, just within specs, but this magnet was very good indeed before the operation, and the customer quite demanding, so that he kept complaining for long time about “horrible lineshape”. Then he got married, and had other matters to care about.
Another important improvement was the solid-state Flux Stabilizer; the original one was vacuum-tube unit, and had quite big problems; the working principle was apparently simple:
– a pair of coils placed over the pole caps picked-up all magnetic field fast variations and converted them in a small error current;
– this current was amplified, filtered and applied back to another pair of coils, called buck-out coils, with proper phase and intensity to compensate and null the field variations.
Easy to say, very hard to get. Here is the original Varian schematics of the damned device; to obtain high-gain with reasonable stability with a 100% tube design, they used an optical galvanometer. The current form pick-up coils drives the coil of a galvanometer, which carries a small mirror illuminated by a lamp; a couple of photocells are illuminated by the light reflected by the mirror; no field variations, no movements of the mirror, the cells are equally illuminated, zero correction current; field changes, mirror moves, a cell gets more light, correction current applied to buck-out coils and field is back to original value.
Believe it or not, it worked from time to time; but if you happen to sneeze you can be 100% sure the mirror will jump and lock is lost!
The vacuum-tube Flux Stabilizer block diagram, together with the Probe and associated components, is illustrated in the above drawing from the original HA100 Service Manual.
This damned gear was eventually replaced with a new, much smaller solid-state unit, employing state-of-the-art operational amplifiers from Analog Devices, at that time just born and almost unknown
What was probably the last version appears in all its beauty in the following artwork: a, fully furnished console with the state-of-the-art, transistorized Internal Lock module and, better late than never, a solid-state Flux Stabilizer unit. The console was now 2 floors plus mezzanine, and had eventually the flat-bed calibrated paper recorder. The operator is adjusting the RF phase while observing the oscilloscope screen to maximize the sinewave of the lock signal.
On the console you can see (from l to r):
– Mezzanine: Magnet power supply remote controls
– First floor: Oscilloscope Integrator/Decoupler unit RF Unit
– Ground floor: Decoupler/Integrator Flux Stabilizer (top) Internal Lock (bottom)
The shim coils controls (6 potentiometers), together with others auxiliary control panels, are barely visible just above the recorder housing. At far left you can also notice the HP 100B Audio Oscillator, sometime used for calibration.
Photo obtained time ago from website www.yorku.ca/cliveh/nmr, no longer available.
Shown are Prof. Clive H. Russel and his coworker (Not 100% sure, I apologize for any improperty)
Eventually comes the flat-bed recorder, with the pre-printed, calibrated paper where the spectrum appears in black ink over a celestial millimetric grid.
The recorder had two major drawbacks:
– the ink pen, getting dry and plugging the pen right when a very important spectrum was recording;
– the driving cables for X and Y axis pen movements getting loose or breaking altogheter; this required a recorder restringing. Cables were several meters long and going intricated ways around wheels, pulleys for motors and potentiometers, deep inside the recorder assembly’s most obsure corners.
If anybody interested, I have the original Operational Manual and schematics for HA series NMR Spectrometers and related equipment.
CW NMR: Tricks and Techniques
The main interest in NMR Spectroscopy moved soon to high-resolution spectra of organic compounds, and was immediately realized the need for much better resolution and field stabilization; approaching the second problem, it turned out quickly that the needed stability was far beyond the one obtainable from even the most sophisticated electronics, and the problem was left unsolved until Dr. W.A. Anderson came out with the bright idea: the Nuclear Sideband Oscillator (NSBO), widely used in NMR and ESR CW spectroscopy, with many examples included in every spectrometer.
An oscillator is an electronic feedback circuit where the feedback loop include a frequency filter, and the idea is to put an NMR signal as the feedback filter; this has several big advantages:
– looking at the phase of the NMR signal, you can monitor any small change in the resonance signal frequency and quickly correct it;
– looking at the amplitude of the NMR signal, you can easily monitor and optimize the field homogeneity;
– the narrower is the lineshape, the better is the filter, and the stability of the loop improves.
Easy to say, but how to realize it? The basic trick is the modulation transfer technique: modulation of magnetic field is transferred to a radio-frequency modulation by the NMR effect, of course only when the NMR occurs, i.e. when and only when the magnetic and radio-frequency fields are exactly in the proper gyro-magnetic ratio.
[ Technical short note: modulation of a signal means to change a characteristic of this signal (typically the amplitude) with another lower-frequency signal; there are many variations of the main theme, but talking about NSBOs the fundamental version apply: the RF signal (carrier) is modulated with an audio-frequency. Modulation always produce at least two other signals: one at a frequency (fcarrier – fmodulation), the other at (fcarrier + fmodulation); these new 2 signals are called sidebands; the fcarrier itself may be present or not (balanced modulation]
So that, Dr. Anderson modulated the magnetic field with a couple of coils glued to the sidewall of the Probe (marked as AC Coils in the block diagram; the coils marked DC Coils are the ones used for field stabilization with the Flux Stabilizer); the modulation frequency was 5 kHz typical.
When the sample goes into resonance, the RF signal from the sample is not only the one at the carrier frequency, but also the other two new sideband signals, and it’s easy to detect and process one sideband signal as a 5 kHz audio frequency with all the amplitude and phase information of the NMR signal.
Another key component of the CW-NMR spectrometer enters now in the story, the Phase Sensitive Detector, a.k.a. Lock-In; it’s a detector with two inputs, signal S and reference R, and it gives its maximum DC output when S and R are exactly on-phase and zero output when S and R are exactly 180° out of phase.
Combine now properly one NSBOs and one or more Phase Detectors and you will have a new powerful and interesting unit, the Internal Reference Stabilizer or Internal Lock.
The unit had two channels, completely independent each other:
1) the Lock channel, i.e an NSBO with fixed modulation frequency; the reference signal had the phase calibrated at 180° respect to the modulation, therefore the DC output is 0 when signal is exactly on-resonance, positive when high-field off-resonance and negative when low-field off-resonance. If the Lock loop is closed, this DC is applied to the correction coils in the Probe (and to the Flux Stabilizer unit to correct for low-term drifts) and, if amplitude and polarity are correct, an NMR Field-Frequency Lock is established; believe it or not, it worked very well.
2) The Observe Channel, with modulation frequency variable and controlled by the x-axis of the chart recorder; as the recorder moves, the audio frequency changes, the RF frequency of the sideband changes of the very same amount and an RF frequency scan is done. The reference frequency is adjusted to be on-phase, so that when the sample goes on-resonance a nice positive going peak appears on the recorder.
A hell of a lot of knobs, potentiometers and pushbottons had to be adjusted and calibrated, and usually several test spectra had to be recorded (on blank paper, the calibrated printed paper was quite expensive!) before the “good copy” was produced and recorded. And, as Murphy’s Law states, often the pen run out of ink right in the middle of this final record.
No need to go into further details, a (very!) condensed block diagram is included here,
Sketch drawn in the classroom of the Varian Service Center, Palo Alto, November 1972
and a copy of the original schematic from the Varian manual follows. Please note the date code: 1268 or December, 1968.
The HA100 was a very versatile and easy to modify spectrometer, the only real drawback being its 1H only observe, fixed-frequency configuration. I cooperated with customer’s technicians to perform many modification (very wide sweep width in locked mode, CW time averaging and, with some dirty trick, 19F observe) with reasonable results. Varian also introduced a 13C observe accessory, but producing 13C CW spectra with homonuclear 13C lock (on 13C enriched CS2, if I remember well) on routine basis proved to be unreachable. By the way, the C-1024 Time Averaging computer was used, see A60 section. I did my best to install such a gear on the HA100 at the Italian National Health Institute in Rome, but I was unable to produce a spectra, so that the engineer who designed the stuff came in from Palo Alto and spent a week to record a couple of readable spectra; the customer was “satisfied”, but later on nobody was ever able to get any visible line coming out of the background noise.
The other BIG drawback of HA100, and other spectrometers of the same family, was they were really hard to use; a dedicated and skilled operator was almost necessary, since most researchers on Chemistry (not to talk about top-level Professors!) had little time and enthusiasm to study all the intricacies of operation, and the function of every knobn and switch on the console. The need of a much simpler, easy to use NMR Spectrometer was obvious, so that about 1966 check!! Varian come out with the A60.