Here's a partial list of EOI's lab equipment. The list is heavily weighted towards electronic stuff, but we also have a very complete set of high precision optical and optomechanical parts: mounts, translation and rotation stages, lenses, prisms, polarization optics, gratings, filters, modulators, detectors, lasers, and fibres, for a start. This lets us try out new ideas quickly and build test setups to verify them.
If you're also a fan of classical equipment, you can download manuals at Keysight (formerly Agilent, formerly HP) and Tektronix, liberatedmanuals.com, BAMA, Kurt's Manuals, To-Way.com, ElektroTanya.com, and K04BB. KE5FX's GPIB Tools are pretty serviceable and also free. (NB: Keysight for some reason doesn't allow deep links to their support pages—you have to go through the main page and select "support".)
Send me an email and let's talk about your application.
~Dr. Philip C. D. Hobbs, Principal
125 ps rise time, 3GHz BW, 10 Gs/s simultaneously on all 4 channels, colour LCD
A true beauty. 50 Ω DC-coupled only, of course, but then I have the matching P6249 FET probes for that.
1GHz 4Gs/s, 4 channels, colour LCD over monochrome CRT About the fastest scope that still has a 1MΩ input option. The display is a bit dim, but it works great.
Thanks to Jay at TestPlus for help with the conversion from 744A to 784A. The conversion involves removing one bypass capacitor across the differential output lines of each front end, which recovers the full 1-GHz bandwidth, and then moving a SMT zero-ohm jumper from one position to another, which lets the firmware know what sort of scope it's supposed to be. I was a bit worried about what the step response would look like afterwards, but it turns out to be just fine.
Another 744A conversion, with the EasyScope LCD screen replacement kit, which is beautiful.
Highly recommended. Somebody had tried to do the 784A conversion on this one, but moved the jumper to the wrong position, which turned it from a 2 GSa/s 744A to the 1 GSa/s "export model", so they sold it on eBay for cheap. (Fortunately we knew the trick, of course.)
HP scopes have much better knob response than Teks of the same vintage. This is the first HP scope I've ever owned. It was less than half the price of my workhorse Tektronix TDS 744A, and overall it seems to work fine. One wart is that it doesn't automatically sense when you're using a ×10 probe.
Sampling scopes are stroboscopic; you get one data point per trigger event so that it takes 512 triggers to get a full trace. On the other hand, a digitizing scope this fast (50Gs/s) would cost as much as a house on a lake. This one is nearly pristine; it's my primary tool for really fast measurements and for things like settling time, where stroboscopic sampling prevents overdriving the scope's vertical amplifier and causing artifacts. Takes two 2-channel plug-ins and has two built-in semi-rigid coax delay lines.
More of a workhorse—21k hours, but still nice and bright. The 11801C doesn't have the built-in delay lines, but can take up to four 2-channel plug-ins. (The 50 vs 20GHz doesn't matter much—I don't have the 30- or 50-GHz plug-ins, but a 20GHz scope is enough to be getting on with.) Since I rely on the 11802 so heavily, it makes sense to have a spare.
These are 120-ps rise time sampling heads with built-in high-impedance FET probes on 1-m cables.
Usually I don't need the full speed of the SD-24 and SD-26, so the lower noise level is a win
These and the SD-24s below have beautiful clean 17-ps rise times with almost no overshoot or ripple.
My favourite head of all; produces a 17-ps rise time pulse for TDR measurements, which is also widely useful as an ultrafast pulse generator for driving lasers and testing circuits.
One of the less convenient featurs of sampling scopes is that they don't really have internal triggering. Their front end amplifiers are hidden behind sampling bridges, so they aren't able to watch the signal change except stroboscopically, i.e. after triggering occurs. This head helps with that at high frequencies (in the gigahertz) but not so much lower down.
Besides triggering, another inconvenient thing about sampling scopes is that we frequently want to measure the effect of some amplifier or other device, by measuring its input and output and comparing them. At low speeds this is easy—often we just use a tee connector on the scope, to connect one cable to both the scope and the input of the device under test (DUT). Unfortunately this doesn't work at high frequencies due to transmission line reflections. For narrow-band measurements we can use special splitters or couplers, but they generally don't work down to DC, and most aren't really intended for fast pulses.
The SD-20 is a good solution to this problem: the input signal passes through the SD-20 on its way to the DUT, and the DUT's output goes to another sampling head such as an SD-26.
The SD-4x plugins are InGaAs devices that work from about 800 to 1600 nm. The fastest ones (SD-48) go up to 30GHz, but they're scarce and expensive. Fortunately I found one on eBay for a steal.
This one is the older, 20GHz version of the fast photodiode boxes. It was an even better buy than the SD-48.
These ones go for a lot less, and realistically are fast enough for most of what I'm likely to do.
This is a calibration fixture that mimics an SD-24 head with a pulse source attached, for testing and aligning the 1180x sampling scope mainframes. Just one of those nice-to-haves.
Fast signals attenuate so rapidly in cables that you really have to bring the head to the circuit. Using these cables means you don't have to hang your gizmo off the front of the scope, but instead bring the sampling head onto the bench. These 2m ones are great for when the scope is in the rack.
150MHz, great triggering, no aliasing, very low noise (2 mV/div)
350MHz, 4 channels. Ultra-bright microchannel plate CRT: fastest writing speed of any portable scope ever
Cheap 'N' Cheerful Chinese.
The matching probes for the TDS694C. Normally a bit on the pricy side, but these were quite reasonable—in future I may wind up making my own with some cascoded pHEMTs. (One of these guys is on the sick list—its bias loop is oscillating at 2MHz or so. Probably a dead capacitor.)
900MHz, AC and DC coupling, ×1, ×10, ×100 All probes have probe hooks, ground springs and ground leads. One of these days I'll get round to doing one of these using a pHEMT cascode, and not have to buy these slow ones. They're pretty good for most things, of course.
Hall effect plus inductive pickup for fast DC-coupled current measurements. Unsurpassed for looking at transients in switchmode power supplies, magnets, and so on.
A beautiful instrument, with four source-measurement units (SMUs) and four additional outputs: 1mV to 100V, 50 pA-100 mA. It would be good to have some spare disk drives—4145s boot from the floppy drive, and that's the only thing that ever seems to break. (Of course it has a proprietary interface.)
Most of these are connected to the lab's 10MHz rubidium reference distribution system.
One of HP's best instruments ever: a combination network analyzer, sampling scope, and spectrum analyzer running from DC to millimeter waves. It's based on a dual-channel picosecond sampling system that works from DC to 40 GHz. These are outrageously cheap at the moment, considering what they do, although they do need precision 2.4 mm connectors, which are not cheap.
A real boat anchor, but it has -110dBc phase noise at 100Hz offset, which is 30dB better than you can get in affordable modern analyzers
The 8568's big brother. I've been getting into much faster measurements recently, especially with the Mach 9 ISICL sensor , which generates Doppler signals up to about 8 GHz. The 1.8GHz HP 8568B is too slow, so I got one of these. This one was very very cheap, and has a much brighter display than the 8568B. It arrived with a problem: the display text was all torn and garbled. I happened to have a spare A4A3 memory board, so I swapped it in on spec, and it fixed the problem. (Score one for pack rats.)
The one real drawback of the 8566 is that the RF (0→2.5GHz) and microwave bands can't be displayed at the same time. In most cases this is more an æsthetic than a practical issue. In the microwave band (above 2 GHz) the noise baseline is a little funky-looking, with ~1-dB ripples, so it probably needs calibration fairly urgently before I do any quantitiative measurements with it.
About the best of the classical FFT analyzers, with all the functions needed for optimizing control loops and doing low frequency noise measurements. A factor of 100 or so more bandwidth would be nice.
This is sort of the answer to the "100x more bandwidth" wish above. It's a lot faster but has worse 1/f noise, so it's not a complete replacement. It sure makes narrow-bandwidth RF measurements faster, compared with an HP 8568A.
Option 040: Jitter FFT
The original "modulation domain" device: plots Allan variance, jitter histograms, jitter spectra, frequency vs. time, and so on. A beautiful thing for laser heterodyne stability measurements, moderate bandwidth phase-locked loops, and so on, though one has to remember that it's basically a glorified frequency/time interval counter, so its resolution is less than you'd expect and its jitter is around 200 ps. It's also a bit of a pain to set up, so it doesn't get used as much as it might.
Simple and reliable.
A plain-vanilla 12-digit, 18-GHz counter.
One of the slightly newer ones that can supply frequency locking to a sweeper such as the HP 8620C. The resulting signal still has a whole lot of phase noise, but its drift is much reduced.
Very decent phase noise (-110dBc/Hz @ 1 kHz offset), fast tuning, 100-dB attenuation range.
Reasonably quiet phaselocked generator, though not in the same class as the (much newer) PTS units. Recently repaired by Bill Dino at Electronics Revisited in Massachusetts—he more or less jacked up the name plate and slid a new synthesizer in underneath: A great outfit!
Opt 01: High Stability OCXO
A nice stable high resolution synthesizer with extreme amplitude accuracy. and free of the usual weird spurs that plague DDS-type units. Only goes to 20 MHz. I also have a couple of these for parts mules.
One of the best of the old-line analog pulsers: stable and easy to use.
No chains of monostables for sequencing in this lab, thanks—4 triggered delay outputs with 1-ns edges and 20-ps jitter, about 20dB better than the competition.
Not brilliant, but good for simple jobs where analog controls are convenient.
Microwave sweepers aren't the quietest things in captivity, but they produce a lot of output power (>=50mW), and they're great for low-resolution swept measurements. These ones accept source locking from a counter than can provide it. The higher frequency one is for driving diode lasers for testing the Hypervelocity ISICL .
Amps, attenuators, filters, mixers, splitters, directional couplers, phase detectors, switches,....
In the ultrasensitive measurements business, you have to be able to make really reliable noise and SNR measurements, and 1 dB accuracy won't cut it. A good photon budget will predict the noise in a good measurement very accurately, but you have to be able to verify it.
The 3400A and 3403A are HP's finest AC meters, with accuracies of a percent or two and a crest factor of 10. The Fluke has ranges down to 2 mV full scale (with reduced bandwidth) and a crest factor of 7.
I use these a lot for noise measurements, in conjunction with passive filters: RCs for simple situations and fancier ones from TTE otherwise. That's the most reliable method: A known noise bandwidth and a true-RMS meter.
Spectrum analyzers read noise as much as 2.5dB too low, due to doing average detection and log video averaging, so fast RMS converters are key. (For the reasons behind this, there's a classic HP/Agilent/Keysight app note AN150. ). Peak-reading voltmeters such as the HP400EL have a less pronounced version of this problem—you add 1dB when measuring noise. (It's really 10log(4/π) = 1.049dB.) The problem arises when you're measuring low-SNR signals, because those hacks no longer work—there's no substitute for true-RMS detection then.
Analog, but very sensitive, stable and easy to use. Accepts an externally controlled bias for C-V measurements, and its analog output works well with data acq hardware.
(100 mV excitation, needs contact cleaning)
A 3-1/2 digit version of the 72A above. Very stable: on the 2 pF FS range, once it's zeroed it sits there showing 0.000 pF all afternoon. Wave a hand nearby and see the reading increase to a few femtofarads. These Boontons use low-level (15 mV) 1-MHz AC excitation with phase-sensitive detection, do 2-, 3- or 4-terminal measurements, and allow external DC bias to be applied. That makes them ideal for measurements on active devices; you can even measure the capacitance to ground of a bootstrapped front end. So useful that it makes sense to have at least two.
Especially helpful for measuring unmarked SMD components; also unusually good for doing in-circuit measurements. Expensive at $400, but saves countless hours over its lifetime.
This was actually the first kit I ever built—in 2015! It does a very good job of measuring capacitors and inductors for RF circuits. It seems to be generally better than the Smart Tweezers in that range, though not so general or so convenient. (You can't get one anymore, owing to the untimely death of its inventor, Neil Hecht.)
The best of a long series of total harmonic distortion (THD) bridges. It has a very low distortion sine wave oscillator to drive the device under test (DUT). The DUT's output goes into an auto-tuned notch filter that gets rid of the fundamental, and the meter displays the amplitude of this distortion-plus-noise signal divided by the fundamental amplitude. To DSP folks, this is actually not THD but SINAD (Signal to Noise And Distortion), but that's how THD has always been defined in the audio business.
The classical grid dip meter. A dip meter's job is to find LC resonances. It's a bit specialized, but very useful in hacking together RF circuitry quickly. Usually the resonant frequencies of the individual sections make more difference to a filter's response than the exact L and C values, and that's what dip meters are good for. The way they work is pretty interesting. There's a variable-frequency LC oscillator in the meter, whose tuning knob has a reasonably accurate frequency scale. A resonator such as a parallel LC circuit will absorb energy near its resonant frequency. If the dip meter's coil is placed so that it couples to the unknown LC circuit, the unknown will absorb oscillator power when the oscillator is tuned to its resonance. This will reduce the strength of the oscillation, which is easily detected. Classically, and in the Measurements 59, it's done by measuring the DC grid current of the oscillator tube, but nowadays there are lots of other methods.
A less obvious use is in measuring the coupling coefficients of transformers: a secondary winding that's 10 mH open circuit can drop down into the low microhenries when the primary is shorted. A dip meter is good at measuring low-Q resonances, so with a nanofarad or two in parallel, you can get a decent measurement of the leakage inductance. I recently measured an ISDN transformer's coefficient of coupling at 0.99986 this way.
The Measurements 59 gives a nice sharp dip even for very weak inductive coupling—much better than any later dip meter I've used. That means that it hardly perturbs the circuit under test at all. I have two meter units and three RF heads: two VHF and one UHF. The frequency calibration of my main VHF one is still excellent: within 2% of reading over the whole 2.2-250 MHz range, though I haven't measured the others.
This illustrates one of my old opinions. In situations where both approaches apply, a smart engineer, even working with very limited technology (vacuum tubes, hand calculation, and mechanical everything) can run rings around a less-smart engineer with SPICE and all sorts of great modern parts.
AVO made the best analog multimeters ever, and the Model 8 Mk IV was really the best of the lot. This one came from the estate of a man who also collected Rolls-Royce cars. No trace of stickiness in the meter movement.
More modern looking but not in the same class at all. It's almost like a barometer—you have to thump it to get the meter to settle properly.
Almost as nice as the older AVO, but not quite.
Mostly for nostalgia, but sometimes still useful for looking at slow voltage changes in high-Z circuits; really nice meter movements
This has a 100 fA full scale range. It uses tubes, and takes a good two hours of warm-up to get down that low, but then it's been sitting on some shelf for nearly sixty years. Instruments have come a long way in performance since 1960, but have lost character doing it. I built a little box with CMOS op amps, Teflon standoffs, and Russian 33G vacuum resistors that does more or less the same thing only better, and works right when you turn it on—but where's the fun in that? (It's often forgotten that a garden-variety 4.5 digit DMM running on its most sensitive voltage scale (usually 400 mV) is also a picoammeter: its low-order digit is 10 uV and its shunt resistance is 10 MΩ, so it you ignore the decimal point it reads directly in picoamps.)
Solid state from '68. A beautiful instrument with a nice meter movement. This one was super cheap, but works very well. Its most sensitive ranges are 1 mV, 10 fA, 100 TΩ, 100 fC full scale, but the 10 fA and 100 TΩ ranges aren't stable enough to be useful.
Solid state, runs off six 9-volt batteries. It's about as sensitive as the old tube one, but it wants an unobtainable 1.34V mercury battery for the ohms ranges. It's waiting for a brain transplant—I'm going to use the meter and switch and give it new electronics, probably based on a 100-pF charge dispensing loop. Or maybe I'll hang onto it for 610C parts.
Good to ~1 ppm (relative) over a really wide range of voltage and resistance measurements. (It's 25 ppm ± 2 digits absolute on all DC voltage ranges.) Its big brother, the HP3458A 8.5-digit DMM, remains the standard in cal labs everywhere.
The standard for 30 years—which is about how long I've had this one. It has the separate rubber holster and the flexible tilting bail that you can wrap around stuff to hold the meter in odd places. The more recent versions have better input protection but come with a much cheesier hard plastic bail.
Roughly in the same class as the Fluke, which is saying a lot. These ones came from eBay in 2014/17, and they work great. I did get one duff one, but my eBay average is excellent—way over 90%.
Another nice meter, but this example isn't very accurate, unfortunately—it reads 0.18% high. One minor annoyance is that when you switch it to volts it comes up in AC, so range switching is a two-step process.
A very nice midrange digital multimeter with a really fast continuity beep, which makes it an excellent kick-around bench meter.
Older mains-powered DVMs with excellent DC accuracy (70 ppm ± 2 LSB) and nice big readable LED displays but no current ranges. I mostly use them for instrumenting test setups—I'll put them in a stack on the optical table, with a Velcro strap to keep them from falling off and Post-It notes to remind me what each one is reading.
Another mains-powered meter with full scale ranges of 20 ohms, 20 mV, 20 uA, with 0.03% basic accuracy.
These things are the most beautiful DMM probes I've ever seen, and they're only $16 per set. Thick gold plating, very sharp points, lovely supple silicone-insulated wires, easy to hold, 1 kV CAT III rated. They also sell a bunch of accessories, some of which I also have. Try these out and you'll toss all your others into the back of some drawer and never use them again. They do get a little flakier after a few years as the plating wears off the points, but they're cheap to replace.
A very capable and well-supported ADC/DAC module that also has six PWMs and 20 digital I/Os. They work great, and are an excellent value, though their DACs are PWM-based, so you may have to use an external SPI DAC. There's also a good library for writing my own code. [I am Not A Fan of graphical toolkits like LabView—as a colleague says, "Ah, LabView: spaghetti code that even looks like spaghetti."] On the other hand, LabJacks are brilliant, and I use them all the time; for instance, in Spring 2013, I built a low cost grating SWIR spectrometer proof-of-concept for a medical devices start-up. It was a prototype of a transcutaneous (through-the-skin) blood glucose and blood alcohol sensor, which worked very well and got excellent data. (Unfortunately the client ran out of money before the product could be introduced.)
The LabJack was doing data acq, servomotor control, and temperature control. At this writing (2015), I'm using one to control a laser scanning microscope operating at a numerical aperture of 5.0.
Many of the gizmos I build myself have switching power supplies built in, usually based on the National Semiconductor (now TI) LM259x 150 kHz Simple Switchers. They aren't the very most efficient things you'll find, but they work painlessly and don't produce much in the way of EMI on account of their slow switching edges, especially with toroidal inductors. I tend to power them from random laptop bricks, so I don't need as many lab power supplies as I used to.
Triple DC power supply
Triple DC power supply
Triple DC power supply
150V 1.5A power supply My standard bench supplies. Reasonably beefy, quiet, bulletproof, and well-behaved.
Kepco BOPs are like op amps, if you think a 98-pound weakling is like the Incredible Hulk. This ±50V, ±2A unit is one of the baby 100W ones—they come up to 400W (e.g. ±72V, ±6A.) They work in all four quadrants, which is both useful and rare. Super useful for things like servomotors and thermoelectric coolers.
A slightly stronger adjustable linear supply, nice and quiet.
Tek has a nice method for powering probes, using PCB pads arranged around the BNC connector. This is especially useful because it allows me to use the 4GHz P6249 FET probes with spectrum analyzers.
Probe power supply (±15 V for the P6201s)
These have 4-digit thumbwheel switches and are dead accurate, except for the 6110A which needs calibration. Typical accuracy is 0.025%.
Precision 0-40V, 0.6 A DC power supply
Precision 0-20V, 2 A / 20-40V, 1A DC power supply
Precision 0-20V, 1.0 A DC power supply
Precision 0-100V DC power supply
Precision 0-3kV 6 mA DC power supply.
Generates voltage ramps for three piezos of our RC-110 Fabry-Perot, for scanning, intracavity measurements, and moderate-accuracy wavelength locking.
Earlier, noisier version, kept as a spare.
Highland stuff is all clean, fast, and strong.
2-channel lock-in amplifier
With plug-ins Model 39A-5, 34 and 35. A really capable, digitally programmable filter system for all sorts of baseband jobs where the usual RC in a Pomona box won't cut it. Unfortunately some of the plugins produce a lot of kickout (noise coming backwards out of the inputs) so it doesn't get used as much as it might otherwise.
A dedicated BPF box. Lots of ganged pots and wafer switches.
0-1 MΩ, 0.1 Ω steps
Distributed via a Mini-Circuits 12-way splitter and cables to keep all the lab synthesizers and analyzers locked together
For PALs and retro stuff like that, plus making backups of instrument ROMs.
Very fast and trouble-free; gives your GPIB stuff an IP address. It's especially useful with the HP spectrum analyzers and John Miles' KE5FX GPIB Toolkit.
A pretty nice box all round. I bought it for doing electromagnetic simulations, but it's also my main office machine. After 2-1/2 years running continuously, the power supply started to hiccup, which made the machine cold-boot every 10 minutes or so, but a new supply cured it. One of these times I'll put SSDs in it, but not today.
For running simulations and general workstation use.
Mostly for running EDA software
A pretty nice network-attached storage (NAS) box that runs Linux and allows you to install normal Linux packages. Synology has their own version of RAID that seems to work better than the normal kind. This is configured with four HGST Travelstar 5K1500 1.5 TB 2.5 inch drives, three in a RAID array and one hot spare. One out of eight drives failed in less than a week, but the others have been running flawlessly for a year now. It's set up to send me email if one of the drives fails a S.M.A.R.T. test, so I knew about the drive failure very quickly.We also have a bunch of hand-configured boxes for use as database servers, NASes, and gateways.
A beautiful thing, with a coherence length > 150m. I bought it for testing the new 10MHz laser noise canceller design. I'll push the beam through a Pockels cell and a single mode fibre to make it really single-transverse-mode, then use free-space optics to split it into three adjustable-strength beams with exactly the same modulation and spatial structure.
632.8 nm, 20 mW, 2.3 kV@6.0mA
2 mW green He-Ne (randomly polarized unfortunately) (Plus a bunch of visible diode lasers)
Single frequency YAGs are beautiful devices. I do a fair amount of stuff in the NIR and SWIR, so having a selection of IR lasers is pretty important.
All single frequency, TE cooled, with collimators and circular output beams
15 mW, pulsed with 190 ps rise time
850 or 1550 nm, rise time about 120 ps. Used with a 30-ps Leo Bodnar pulser for testing fast TIAs.
With 4× & 8× objectives. This really speeds up surface-mount prototyping.
I often use Mitutoyo microscope lenses in optical setups, and missed having the microscope to go with them. Mitus are workhorses, and the optical quality is outstanding. This one has 5×, 10×, 20×, and 50× objectives and 10× eyepieces, plus a 4k colour CCD camera.
For inspecting fibre facets before use
Dithered shear plate collimation testers—indispensable. It would be great to have a real wavefront- wavefront-measuring interferometer. I used to have an old Wyko Ladite, which was a thing of great beauty, but had to be put out to pasture because it couldn't communicate with any modern hardware.
Michelson interferometer wavemeter, 0.35-1.0 μm with vacuum chamber
Piezoelectrically controlled, with mirrors for the red; really great for passive-cavity measurements.
A scanning slit instrument for measuring beam diameters. Much faster than either chopper-wheel or razor blade + translation stage methods.
350-1050 nm coverage, 2048 pixels, about 1.5 nm resolution. Primarily for the egg-grading sensor , but will also have applications in future blood constituents work.
A nice small Czerny-Turner monochromator with a 1400-2800 nm range and two sets of slits.
Only slightly faster than the Tek SD-42 plug-ins, but small and battery powered.
700-1300 nm IR viewer handheld IR viewer. Useful alternative to a CCD camera and monitor.
700-1300 nm IR viewing goggles Head-mounted IR viewer: easier to manage while doing alignment, but blind past 1.3 μm
A lead-salt vidicon, covering 0.6-2.2 μm: Big and clunky, so it's conveniently mounted on a Manfrotto Magic Arm for hands-free use. Needs a bit of adjustment because the display contrast is sort of low at the moment
These are good out to 1064 nm, and are a lot more convenient.
For use with the Electrophysics and Toshiba cameras.
Coatings for UV to 1.6μm
Ordinary folding mirrors with dielectric stack, enhanced aluminum, protected gold coatings, in λ/10 and λ/20; also a lot of interferometer mirrors, laser output couplers and HRs.
Dove, Porro, equilateral, wedge, penta.
About a dozen, in 12-mm and 25-mm apertures
Polarizing cube beamsplitters: About 20, 400 nm - 1.6 μm
Beam separator prisms: PBS laminated with a λ/4 retarder, 633 & 750 nm; about a dozen
Waveplates for 532 nm, 633 nm, 750-840 nm, and 1.5 μm; lots
Film polarizers, thin-film waveplates,....
Various lines/mm and blaze wavelengths, UV/Visible/SWIR: about 40
About 50, UV to SWIR
UV, 532 nm, 578 nm, 599 nm, 633 nm, 750-840 nm, 1500 nm
Two sets of 50-mm metal ND filters plus a bunch of black glass ones
A dozen or so SM and MM patch cords, fibre on reels, FC connectors and polishing stuff, collimators and so forth. Both glass and plastic.
SWIR: tunes 1.4-2.45 μm with 3W of 52.5-95MHz drive 80% efficient and 3.5nm resolution at the short end, 45% and 10nm at the long end; optical acceptance ±3°. AOTFs use noncritical phase matching, where the curve of wavelength vs. angle of incidence goes through a maximum, making them widely tunable.
Off-axis TeO2 cell for the visible; 50-110MHz operating frequency, 340 resolvable spots@633nm
Like the EFLD-340, but with a smaller crystal, giving 250 resolvable spots. Both in excellent shape; about 150 kHz modulation bandwidth
200MHz centre frequency, 10MHz analog modulation bandwidth.
80MHz centre frequency, 10MHz analog modulation bandwidth.
5 mm aperture, with 8 plates, which reduces Vπ from 3kV to about 400V, and runs as fast as you can drive it.
Nonmagnetic, which is sometimes a pain and sometimes very useful
Wye Creek Instruments, Burleigh, Newport, and others—piezomikes, flexures, and pushers.
Encoder mikes are useful for moderate-performance applications, where open-loop steppers are too coarse, but you don't need the extreme smoothness of piezos. This control box is easy to talk to over RS-232, and I have a bunch of code from years back to run it.
Indispensable for holding cameras and other things that need to sit at odd angles. It's about the length of your arm, and articulated much the same way, with a ball joint at shoulder and wrist and a hinge at the elbow. Move and point it as you like, then a quarter turn of a handle locks it in place. Mounts conveniently via a 1/4"-20 screw to the breadboard. A big favourite at EOI.
For attaching the arm to places that don't have tapped holes handy. Especially good for holding on to Unistrut and other Meccano-type prefabs.
of Newport, Melles Griot, Klinger, Thor Labs, and Linos mounts Mirror mounts, tilt stages, 1-, 2-, and 3-axis translation stages, damped posts, prism tables, lab jacks, fibre positioners, and so on. All the stuff an optics lab needs to allow bolting optical systems together quickly.
Microbench (Spindler & Hoyer aka Linos aka Qioptiq) is an optical Meccano set based on mounts threaded on four 6-mm centreless-ground stainless rods. Good for more alignment-sensitive jobs, but more prone to etalon fringe problems than random mounts. Thor sells a mostly-compatible series with more selection, but their stuff is much less mechanically stiff.
One of these times I'll declare it to be Fixit Week and get these guys back in harness where they belong.
Just because. The original HP 200A was HP's first product, and this is a lineal descendent. Wien bridge, tubes, and a light bulb for automatic level control. It was the last tube product in the HP catalogue, 1985!
(LO unlock problem)
An early model that reads about 2.5dB low, probably due to a dead emitter bypass cap someplace.
Worked fine until I put it into storage mode for the first time, then keeled over.
The Boonton is a bit of an experiment—it's a lot like an HP3400A, but with a bandwidth of 20MHz. Like the 3400A, its maximum range is 300V, which is a fair amount higher than any other RF thing I have. Boonton's capacitance meters are amazing, but this particular unit reads about 4 dB low.
(Roughly in priority order)
For noise temperature measurements in ultrasensitive front ends, and for extracting the junction parameters of metal-insulator-metal (MIM) tunnel junctions (It looks like I'll be revisiting that work in the near future.)
Single frequency YAGs are beautiful devices. I do a fair amount of stuff in the NIR and SWIR, so having a selection of IR lasers is pretty important. I used to have a JDSU NPRO 126N-1064 700 mW unit, which is a thing of great beauty, but it belonged to a client.
For locking to etalons
(Rebadged Watec 120+ with a cooling system bolted on by CosmoLogic) plus a telecentric C-mount lens for video microscopy Primarily for advanced amateur astronomy, but a sweet solution for high sensitivity imaging in a lab setting too. Not the absolute lowest dark count rate available, but excellent bang for the buck.
The lab is somewhat weaker in fibre devices than in free-space optics, which should be fixed
e.g. Avanex SD-40 Zero-chirp longitudinal Pockels cells—very quick and easy to drive
50:50 and 90:10
850 nm, 1064 nm, and 1550 nm
We have 10 scopes all to ourselves as it is, but you really can never have too many.
For when our ship comes in. The spectrum analyzer's phase noise is no match for better-grade standalone units such as my HP 8566B, but the mixed-domain capability is entirely unique. I'm also a fan of the new Agilent 2000 and 3000 series Infiniium scopes, especially on account of their writing speed (1 M traces/s) and their very responsive user interface, but the three-domain capability gives the Tek the edge for electrooptical work.
Another SD-20 Loop-through sampling head
SD-32 50GHz sampling head to match the SD-48 30-GHz O-E converter
A couple of 012-1220-00 1-m Sampling Head Extenders for benchtop use
Mostly for fast temperature control and motion control experiments.
To connect the 70001A to my 70004A display
A more modern version of the 89441A, used for testing 3G cell phones (which is why they're now pretty affordable used). Needs Option B7C to get the baseband input.
This is an AC-coupled, 300kHz–3GHz high-Z probe specifically designed for spectrum analyzer use.
It has low noise and low distortion, and (most important) no DC output, so it won't blow up your spectrum analyzer's input mixer.
Option 004: 70 dB step attenuator HP 85032B Cal Kit or
HP 85047A S-Parameter Test Set and Cal Kit
Apparently these have all disappeared, so I'll probably have to build something equivalent
Models 30A, 31A
For 8568B. The 8566 and 8568 have a slightly screwy frequency plan, so the tracking generator isn't exactly locked to the LO, but it's a lot better than nothing.
Another 7-foot EIA rack to hold all this extra stuff!