31 July 2011

Scope Sunday 3

Instead of buying a scope this week, I've decided to retell a scope story that I heard from Jim, with a little follow up:

A Tektronix engineer visiting IBM in the early 1960s saw that they were developing an oscilloscope. When he inquired about it, he was told that IBM wanted a scope for their field-service engineers. After some (probably brief!) internal discussions back in Oregon, Tektronix offered to develop a portable scope to IBM's specifications. IBM had two requirements: it had to be fast enough for computer work (where "computer" probably means System/360) and it had to fit under an airline seat. To satisfy these requirements, Tektronix produced the twenty-nine-pound 50-MHz 453. If you buy one today, many of them still say "Property of IBM".


After Jim told me this story, I needed to see for myself. So one time that I was in California at the De Anza Flea Market, I bought a ten-dollar 453 (serial number 23,109), and proceeded to use it as my carry-on for my flight back to Boston. And, yes, it does fit under an airline seat. Here's the proof:


When I sent this picture to Jim, here's the letter he sent in return:


Here's the text:

Ah, yes. The Tek 453.

I had a 453 in building 20 (serial number 7402) that I used for years; it was stone reliable. When I wasn't using it, I studied it. Early 453s had nuvistor triodes at the attenuator output and in the trigger buffers---later versions (serial numbers 20,000 and up) used FETs because it took a while for FETs to be good enough for Tek standards. In about 1966 Carl Battjes figured out a way to use T-coil bandwidth boosting with transistors, which, of course, everybody knew was impossible. He applied T-coils to the 50-MHz 453, tripled the bandwidth, and begat the 454, which had an astonishing 150-MHz bandwidth. But the 453 was the beginning...

If you need any 453 parts, let me know. I have plenty.

Finally, I would like to thank the TSA agents at San Francisco International Airport for their cool professionalism. They didn't freak out all at when I put the 453 into the X-ray machine. You would think that a middle-aged guy carrying a middle-aged scope was a normal occurrence! (They did swab it for explosive residue, but I was expecting a much more invasive search.)

29 July 2011

App Note 13 part 3

The appendices of App Note 13 include a wealth of practical information. Appendix A talks about bypass capacitors and includes five scope traces that warn of potential troubles. Figure A7 is particularly horrifying. (I wish he named some names here; I'd like to know what specific combination of capacitors caused that shameful ringing. I guess I'll have to experiment myself... Personally, I've been using a combination of tantalum and X7R for bypassing. I really should check it out, as Jim suggests.)

Appendix B further discusses probes and oscilloscopes. Again, he doesn't name any specific makes and models of oscilloscopes, but we can guess what he's using (a Tek 547 and a Tek 556). It's funny how he suggests that the oscilloscope should have 150 MHz of bandwidth, after admitting that "90% of the development work was done with a 50MHz oscilloscope." More space is devoted to discussing probes, FET probes, current probes, and (of course) grounding. I think that I will steal the test circuit in Figure B1 to use at the basis of a lecture demo and/or lab project. It is simple, yet instructive. The picture in Figure B5 shows a wide variety of probe types ("Note the ground strap on the third finger."). 

Appendix C discusses some suggestions for ground planes. In short, use them and love them. 

All three of the above appendices will appear again (in one form or another) in App Note 47.

Appendix D shows an interesting and strange circuit for producing very fast pulses. First comment: the LM301A is only specified for a maximum voltage of 36V. The military-grade version, the LM101A, is specified to 44V. I wonder why he didn't suggest the LM101A? Second comment: the circuit uses a TD-263B tunnel diode! That's cool (it's the right tool for the job), but I don't think that Germanium Power Devices even makes tunnel diodes any more. Does anyone? In Figure D2 and the accompanying caption, we learn that the heretical HP scope that we occasionally see is a 275-MHz unit. (I don't know my HP scopes very well. Can anyone identify this model? Is it an HP 1725A?)

Appendix E discusses high-speed level shifters. Figure E2 shows a TTL-inspired level shift with a 15-volt output. I like figure E3 with the speed-up capacitor and the Baker clamp. I really do have a soft spot in my heart for old logic-circuit topologies. I'm curious about what application requires that power FET switching one amp(!) in 9 nanoseconds in Figure E4.

Best quote (page AN13-27): "Probes are the most overlooked cause of oscilloscope mismeasurement." Yep.



Related:

28 July 2011

App Note 13 part 2

The applications section of App Note 13 includes a dozen more of his favorite circuits. There are three voltage-to-frequency converters. The first one (Figure 16) goes to 10 MHz and is based on a feedback charge pump (some similarities to Figure 8 in App Note 7). I really like the trick of replacing the input stage of the LT318A with discrete 2N4393 FETs connected to the op amp's balance terminals. Clever. The second V-to-F circuit (Figure 19) is a quartz-stabilized design that goes to 30 MHz, using a sampled-data DAC in the feedback path. Why doesn't this schematic include the oscillator circuit?

The third V-to-F is shown in Figure 21, with a sine-wave output using an AD639 as a triangle-to-sinusoid converter. I'm sad to see the (now-discontinued) AD639 used this way. The AD639 was a Barrie Gilbert's brilliant Universal Trigonometric Function Converter, and using it as a triangle-to-sinusoid converter is like using a Lamborghini to go get your groceries. It was capable of producing sine over a plus-and-minus 500 degree range, along with cos, tan, sec, csc, cot, arcsin, and other functions, using some really interesting bipolar-transistor tricks. The data sheet and Gilbert's JSSC paper are a fun read. It really was a brilliant chip; too bad nobody really figured out what to do with it.

(B. Gilbert, "A monolithic microsystem for analog synthesis of trigonometric functions and their inverses," IEEE Journal of Solid-State Circuits, vol. 17, no. 6, pp. 1179–1191, Dec. 1982.)

There are also three sample-and-hold circuits, continuing on a theme that he started in National Semiconductor App Note 294, "Special Sample and Hold Techniques." Figure 23 shows a high-speed ramp-and-hold circuit that uses a single-slope-integrator technique to match the input voltage. He uses a 2N5486 JFET to buffer the hold capacitor from the output (in this circuit and the next one). Figure 26 uses a similar topology to implement a track-and-hold circuit that oscillates around the input voltage. Oscillations on purpose? Scary.

Figure 29 revisits a sample-and-hold circuit for repetitive signals from AN-294, although it is improved from a 100-ns sampling window to a 10-ns window. This circuit samples a single voltage at an adjustable delay time after a zero crossing. Just think, with a bank of 32 of these circuits, you could build the front end of an analog sampling oscilloscope.

The next two circuits are analog-to-digital converters. Figure 31 is a two-speed SAR converter (a topology that he will revisit in App Note 17). Figure 33 is a single-slope-integrating converter. Several tricks are used to improve performance, such as the Schottky diode on the latch pin and the temperature-compensating resistor in the current source. Also note Q4, the 2N2222 being used in the reverse-active region to improve the capacitor reset voltage.

Figure 36 shows a high-frequency (2.5 MHz!) precision rectifier that uses a four-diode gate with level-shifted drive like Figure 5 of App Note 10. Figure 38 is a fiber-optic receiver, using a discrete-transistor amplifier for the high-frequency gain stage (Q1 through Q6), and a cute min-max detector (Q6 and Q7) that samples the peaks and valleys of the input signal and calculates the midpoint for the comparator threshold and the level-out monitor.

Figure 40 shows a 12-ns circuit breaker (I like the output stage of Q1, Q2, and the diode that is virtually the same topology as a discrete TTL gate). Figure 42 is a high-speed trigger circuit, using a discrete-transistor FET input buffer and showcasing the high speed of the LT1016 (50 MHz!).

Whew! I'll discuss the appendices tomorrow.



Related:

27 July 2011

App Note 13 part 1

"High speed comparator techniques." 32 pages.

At 32 pages, this app note is the longest one so far, the first with a table of contents, and the first with separate named appendices (there have been "box sections" before, but never appendices, and this note contains five!).  In addition to just being long, this work is also an important piece of foreshadowing for future projects.  I cannot cover this app note in a single day, so I'm going to spread it out over three days.

After a short introductory lament ("Comparators may be the most underrated and underutilized monolithic linear component"), this note covers design, construction, and instrumentation techniques for high-speed comparators (such as the ten-nanosecond LT1016 featured here), and it lays the foundation for his longest (and best?) application note, the fabled forthcoming App Note 47.

The first section of this note is a "Rogue's Gallery of High Speed Comparator Problems" along with thirteen scope photos of various ailments.  The frustration in these comments is really close to the surface (which I totally understand and absolutely agree with).  Imagine how many application complaints are eventually traced to poor bypassing, incorrect scope probes, poor grounding, suboptimal construction, and excessive parasitic capacitances!  The fact that five pages of this app note are dedicated to "You're doing it wrong!" is quite telling.  Nevertheless, this rogue's gallery should be required reading.  Read it twice.

Page AN13-8 has a short (the first!) discussion of oscilloscopes.  He doesn't name any names, but he does admit that "90% of the development work was done with a 50MHz oscilloscope."  I'll bet dollars-to-donuts that he's talking about his Tektronix 547 and 556 (you can see them in Figures 3 though 15).  He sums it up in a good quote: "In general, use equipment you trust and measurement techniques you understand.  Keep asking questions and don't be satisfied until everything you see on the oscilloscope is accounted for and makes sense."

Best quote (from page AN13-1): "In developing such [fast] circuits, even the most veteran designers sometimes feel that nature is conspiring against them.  In some measure this is true.  Like all engineering endeavors, high speed circuits can only work if negotiated compromises with nature are arranged.  Ignorance of, or contempt for, physical law is a direct route to frustration."

I'll discuss the applications section tomorrow.



Related:

26 July 2011

App Note 12

"Circuit techniques for clock sources." 8 pages.

I think this app note is most notable for what it doesn't contain: eight pages of circuits for clocks and oscillators (including a sinusoidal oscillator; see Figure 9), yet no mention of Wien bridges or his treasured HP200. Curious. Most of the circuits here include quartz crystals, but not all of them: Figures 11 and 13 are synchronized to 60-hertz line frequency, and Figure 15 is a "stable RC oscillator." A discussion of a Wien-bridge oscillator would have fit nicely into the theme.

I agree with his characterization of the circuits in Figure 1 as "temperamental". Despite these circuits' common use, there's a reason why Colpitts, Hartley, and Pierce are still household names. (Well, OK, not quite "household names", but you, dear reader, my educated friend, you should know them.) The box section on page AN12-8 is a one page discussion of quartz crystals, which is good reading for the uninitiated, but not really enough information to dispel the mystery.

Two approaches for temperature compensation are shown. Figure 5 shows an ovenized Pierce oscillator, but the best circuit is Figure 6, a Colpitts oscillator with a temperature-compensation loop using a thermistor and a varactor (very similar to Figure 21 back in App Note 3). The results shown for the temperature-compensation loop in Figure 7 are impressively flat.

Best quote (from page AN12-1): "In consideration of these difficulties, gate oscillators are generally not the best possible choice in a production design."  Listen to the man, he knows of what he speaks.

25 July 2011

App Note 11


This app note contains some clever tricks for circuits powered from a single 5V rail, most of them using LT1013/14 low-voltage op amps and LT1017/18 low-voltage comparators. The box section on page AN11-15 tells a little bit of the history of linear supply voltages, starting with the 300-volt supplies used in vacuum-tube analog computers, and going down from there. (How low can you go? Stay tuned: single-cell circuits are coming up in App Note 15.)

Again, the circuits here have interesting sensors. The platinum RTD in Figure 1 and the thermocouple in Figure 5 are old news, but Figure 2 shows a unique methane detector with a voltage-to-frequency output. A little translinear circuit is used to compensate for the sensor's square-root behavior. Figure 7 shows a strain-gauge circuit that uses a voltage inverter (LTC1044) and an op amp to servo the left side of the bridge to zero volts. (There are many more bridge interfaces coming up in App Note 28.)

The schematics in Figure 6 show instrumentation amplifiers for 5V operation, one using the standard-topology three-op-amp instrumentation amp with a LT1014 quad op amp, and the other circuit using the LTC1043 switched-capacitor block and LTC1052 chopper amp from App Note 9 Figure 11.

Figure 8 shows a "tachless" motor speed controller that senses the motor's back EMF to determine speed. This approach makes me nervous. Sensing the voltage this way is really noisy! The design of windings (and commutator brushes) is very different for a motor than for a tachometer. I think the key phrase here is "Q3 turns off and the motor's back EMF appears after the inductive flyback ceases." You can see all of the signal junk in Trace B of Figure 9, and the residual junk after filtering in Trace D. I'm always impressed when an engineer can get this noise-plagued approach to work even this well.

Circuits for systems requiring galvanic isolation are shown in Figures 11, 12, and 14. The best circuit is Figure 14, a fully isolated single-slope analog-to-digital converter. As in Figure 12, the transformer is used both to power the isolated circuits and to transmit data (in this case, the pulses from the A/D conversion) across the isolation barrier.

Best quote (from page AN11-1): "The difficulties encountered in maintaining the lowest possible levels of noise and drift in an analog system are challenging enough without contending with a digitally corrupted power supply."

24 July 2011

Scope Sunday 2

It is not my intention to purchase an oscilloscope every week while I am writing this blog, but when this 547 came up on Craigslist, I couldn't help myself. The 547 is Jim's favorite scope! You can see it on his lab bench on the cover of his first book, and it's the scope used in many of the oscilloscope photos in the app notes.


It's a little dirty (OK, it's a lot dirty), but it looks like it's in pretty good shape. No physical damage and no missing tubes. It came with a Type M and a 1A4, but the seller said that one channel didn't work (probably in the Type M, since that was the one installed).

According to the seller, this one used to belong to Amar Bose, former MIT professor and founder of the Bose speaker company. (Actually, it just has a Bose property tag on it, so I think it's just Bose company surplus, and I doubt that it belonged to the big man himself... so much for a rare provenance.)

I haven't plugged it in yet. I want to clean it up a little bit first, and I'm concerned about the high-voltage transformer. The HV transformer in this scope is known to go bad. The transformer is potted in an epoxy that absorbs water and eventually fails (or causes the circuits around it to fail). Given that this scope was likely stored in a damp basement for who-knows-how long, is there a way to test the HV transformer for deterioration before just plugging it in and powering it on?


That's it. No more scopes (unless it's a 556; I still want one of those, and a 661, and a 576, and maybe another 7104... I also need a working 109 and a 7B87).

22 July 2011

App Note 10


This app note, discussing the measurement of op-amp settling time, marks the beginning of a career-long assignment (or is it obsession?) with measuring the fine settling time of op amps and data converters. This topic will be revisited in several future app notes. More of a measurement manual than an application note, this note explains the proper instrumentation and technique for a precision settling-time measurement. In particular, Box Section A (starting on page AN10-6), on evaluating and avoiding oscilloscope overload response, should be read twice and then replicated in the lab.

For best circuit, there's really just one choice: Figure 2, combined with Figure 5, make possible the ultra-precision measurement of settling time down to the ten-microvolt level.

There are many worthy quotes of good advice here:
  • Unfortunately, oscilloscope overdrive recovery characteristics vary widely among different types and are not usually specified.
  • Previously, being able to see an amplifier settle within 50uV wasn't interesting because its thermal drifts swamped this figure.
  • To maintain low noise, the bridge's output ground return should be routed away from high current returns such as the 74123's ground pin.
  • Some poorly designed amplifiers exhibit a substantial "thermal tail" after responding to an input step. This phenomenon, due to die heating, can cause the output to wander outside desired limited long after settling has apparently occurred.
However, I'm going to pick the best quote for its future value. Best quote (page AN10-3): "Since most amplifiers are not nearly this fast [70 ns], it is reasonable to assume that the circuit will always provide reliable results." Seventy nanoseconds? That may seem really fast in 1985, but stay tuned!

21 July 2011

App Note 9


This app note, discussing the LTC1052 chopper-stabilized op amp, includes a wealth of good advice for designing and building high-precision circuits. (The box section on the final two pages discusses chopper theory in more detail.) In particular, the discussion of the thermocouple effect starting on page AN9-2 is very valuable. Figure 5 is worthy of careful study to understand the deleterious effects of thermal EMFs on precision circuitry. This kind of attention to detail separates man from mouse in the microvolt regime. A thermal gradient across parasitic thermocouples created by copper connections to solder or Kovar can be a significant source of error. (What's Kovar, you ask? It's a magical alloy of nickel that makes vacuum tubes, light bulbs, and other glass packages for electronics possible. While you're looking it up, look up the other magical alloys of nickel, like invar, elinvar, constantan, and mu-metal.)

More vintage hardware: Figures 2 and 7 show strip-chart recordings of the low-frequency noise. A strip-chart recorder is another one of those valuable tools (particularly for low-frequency noise measurements) that I wish I had room for. Figure 9 shows a voltage reference using several components from Julie Research Labs, including a saturated-cell voltage reference and a Kelvin-Varley divider. Julie Research Labs : when you absolutely, positively, have to exceed your research budget.

Figure 15 shows another HP oscilloscope photo (also, Figure 22). Heresy.

Figure 18 shows how to use the LTC1052 to correct for offset and drift in an Analog Devices AD650 monolithic voltage-to-frequency converter. Figure 19 shows a discrete voltage-to-frequency converter (that improves upon the specifications of the AD650 part) that is very similar to the architecture used in Figure 2 of App Note 7. Now we start to see the rest of the V-to-F dynamic range iceberg (see the so-called "best circuit" from App Note 7): Figure 19 has a dynamic range of 120 dB, Figure 21 has a dynamic range of 150 dB ("This is by far the widest dynamic range and highest operating frequency of any V-to-F discussed in the literature at the time of writing."), and even more is promised in App Note 14. See the footnote on page AN9-14.

Figure 25 shows a 16-bit analog-to-digital converter, this time using a sigma-delta architecture.

For the best circuit, I'm going to have to go with the thermometer in Figure 27, which feels like a wicked prank. This application points out that the offset drift of an AD547J (the competition's op amp, now obsolete) is so bad that it can be used as a temperature sensor. (In reality, the max drift was only 5 uV/C, but that is one hundred times worse than the LTC1052 discussed here.)

Best quote (page AN9-2): "Any connection of dissimilar metals produces a potential which varies with the junction's temperature (Seeback effect). As temperature sensors, thermocouples exploit this phenomenon to produce useful information. In low drift amplifier circuits the effect is probably the primary source of error."

20 July 2011

App Note 8


This app note includes useful circuits for battery-powered systems, including voltage doublers, voltage inverters, and rail splitters. The LTC1044 is used in several circuits, as is Bob Widlar's low-voltage op amp, the LM10 (Figures 3, 16, and 18).

Figure 6 is an interesting low-dropout regulator, using a voltage doubler to supply the op amp to drive the base on the NPN pass transistor, achieving a dropout voltage of Vce,sat. Figures 9 and 11 show low-power switching regulators (Figure 11 includes a linear-regulator stage with a JFET pass transistor). Figure 12 shows an "inductorless" switched-capacitor regulator (which is exactly the same as Figure 23 in App Note 3).

Germanium! Excellent! Figure 18 uses germanium diodes and transistors. How often do you get to use 2N1194 transistors and 1N100A diodes? Can you even buy 2N1194 transistors anymore? I bet Jim was hoarding them, waiting for an opportunity to use them. (At first glance, I thought that Figure 18 was awfully strange, given that it doesn't use any Linear Technology parts. After a little more digging, I found out that Linear did second-source the National LM10 for a while (along with some other parts). Interesting. Given the contentious (and litigious) relationship between Linear and National in the early days, I'm surprised there was that much technology transfer.)

(With my over-active imagination, I can imagine how this situation arose: Linear Tech asked Bob Widlar to design a low-voltage op amp for them, and Widlar replied, "Been there, done that, it's called the LM10." When Linear said, "No, that's a National part; we want one of our own to make and sell," Bob gruffly replied, "Then get it from National." Disclaimer: I totally made this scene up.)

The best circuit is the single-inductor, dual voltage flyback regulator in Figure 22. I just think it's a neat topology to get a bipolar 15-volt power supply from a 6-volt battery using a single inductor, although I'm surprised that the clock input on the 74C74 is just labeled "30kHz input". I'm shocked that he didn't design a Wien-bridge oscillator to provide the 30-kHz input. Shocked.

Best quote (page AN8-8): "This circuit [Figure 18] will supply a 5V, 150uA load (about 25 CMOS SSI ICs) for 3000 hours from a single 1.5V D battery." Impressive.

19 July 2011

App Note 7


This app note claims "direct digitization," but the circuits are based on voltage-to-frequency conversion rather than analog-to-digital conversion. Thus, I disagree with the title: V-to-F conversion isn't really a digitization technique; V-to-F conversion translates signals from one continuous domain (voltage) to another continuous domain (frequency). There's no quantization, so it's not really "digital". However, it is discretized in time (at the edges of the output waveform), so that's something.

Terminological quibbling aside, V-to-F conversion is an under-appreciated, but useful, trick in practical system design, particularly in instrumentation circuits, as demonstrated here. V-to-F isn't as sexy as A-to-D, and it doesn't get as much (if any) emphasis in college courses, but it's a life saver in some applications. Transmitting information via frequency (particularly over long cable runs) is often a good idea.

Figure 2 shows a V-to-F converter for a Type-K thermocouple. It's a good example of using a crude V-to-F converter in the forward path (the 74C04 inverter chain) and then linearizing it with a precision charge pump in the feedback path. Feedback is awesome. (Possible schematic error: is there a connection dot missing from the summing junction of A1? I think the 3300pF capacitor, the 33k resistor, and the 150k resistor should all be connected to the minus input of the op amp.)

The other applications are great, too. Figure 4 shows an acoustic thermometer (for those of you reading ahead, acoustic thermometry is also the topic of Jim's last app note twenty-five years later, App Note 131). Figure 6 shows a circuit for strain-gauge that produces a full-scale signal current of only 48 nA. Figure 8 shows a "Pease type charge pump" current-to-frequency converter with 100 dB of dynamic range (20 Hz to 2 MHz). Figure 11 shows a "humidity-to-frequency" converter (compare to Figure 8 in App Note 3). Figure 14 shows a level transducer with AC drive, and Figure 16 shows a circuit for a capacitive accelerometer. Again, the real story is the great assortment of sensors and transducers that Jim comes up with. I'm constantly jealous of his resourcefulness and his junk pile.

I think all these circuits are great, but the best circuit is the photodiode digitizer in Figure 8: Anything with a useful dynamic range of 100 dB deserves respect (and this result is just the tip of the V-to-F dynamic-range iceberg; stay tuned).

Best quote (page AN7-8): "At a minimum, careful layout and a clean PC board are required. The best practice is to use a Teflon stand-off for all summing point connections." Good advice. You know you're down in the picoamps when Telfon standoffs become necessary.

18 July 2011

App Note 6

"Applications of new precision op amps." 8 pages.

This app note is a bit of a mixed bag. I'm not sure if I agree with the last sentence in the first paragraph. While the circuit in Figure 1 is an interesting beast, with its crazy opto-coupled MOSFETs and its 300-volt common-mode range, I'm just not sure if it's an "excellent example" of what to do with a new precision op amp. I think the platinum-RTD signal conditioner and the direct-connection thermocouple-driven battery charger are better examples of what to do with a precision op amp like the LT1001.

The best circuit would be the dead-zone generator in Figure 5, but why is the second LM301A labeled C1? Is he really using it (uncompensated) as a comparator? Bob Widlar would not approve. Why not use an LM311?

Also, the ultra-precision voltage reference is a neat application, but something about "ultra precision" and "chopper step-up transformer" in the same system leaves a bad taste in my mouth. Plus, C1 is another LM301A being used as a comparator. I like how Figure 7 nonchalantly indicates a Kelvin-Varley Divider in the circuit. No problem. Doesn't everyone have one in their desk drawer? (All kidding aside, the careful error budget on page AN6-7 is a skill more engineers should learn.) Note that page AN6-6 includes another call for the General Radio 1432-K precision resistor decade box (see App Note 3). He was a big fan.


Actually, the best circuit is the high-speed op amp in Figure 8, another brilliant discrete design, using 2N3688s, 2N5160s, 2N4440s, and a feed-forward path to improve for high-frequency performance. Finally, note the oscilloscope photo in Figure 9: is that an HP logo in the top left corner? Heresy!

Best quote (page AN6-1): "A2 is trimmed for a 93Hz clock output. This frequency inhibits power line-originated noise from interacting with the switching action because it is not harmonically related to 60Hz." I'm surprised that he didn't discuss the history of 93-cps modulated systems.

17 July 2011

Scope Sunday

I felt Jim's presence at the MIT Swapfest this morning.  I found two Tektronix oscilloscopes that needed a good home, a 465 and a 475A.  Ten dollars each and guaranteed not to work.



Here's the appropriate quote from page 7 of Jim's second book: "It just seems sacrilege to let a good piece of equipment die...  fixing is simply a lot of fun.  I may be the only person at an electronics flea market who will pay more for the busted stuff!"

While I felt no need to pay more for these scopes, I certainly knew they had to come home with me.

15 July 2011

App Note 5


This app note has two best circuits: first, Figure 6 shows a wideband thermal RMS-to-DC converter. This circuit uses the old servo-multiplier (or light-bulb multiplier) technique of using a feedback loop to force two matched parameters (in this case, the heat produced in the two composite thermistors) to have equal value. We'll see this circuit again. The flowmeter (Figure 8) and anemometer (Figure 11) are related tricks.

The second celebrity circuit is on the final page: Jim's obsession with the HP200 makes an early appearance. Figure 12 is an op-amp recreation of the Hewlett-Packard HP200 audio oscillator circuit (and the references include William Hewlett's thesis). While the flowmeter and anemometer are useful applications, I wonder if they weren't just a set up so that he could justify spending time at work playing with Hewlett's light-bulb circuit.

The other applications are also interesting and instructive. Figure 1 shows an interesting temperature controller (but the LT3525A is now discontinued). The explanation includes a simplified discussion of the various delays and time constants in a heat flow problem, but the full story is much worse: it's not just a multiple-time-constant system, it's a diffusion-equation system! Also, his discussion of insulation needs a caveat about the trade-off: you want to keep the losses small to save power, but large enough to allow the system to quickly recover from an over-temperature condition. Regardless, using the 50-ohm resistor and switch to check the loop response is a great idea, but the loop response needs to be checked for positive steps (when the heater heats) and negative steps (when the losses cool). He only shows the former. (Also, I don't think he needs the 100M resistor in the integrator, but that's another discussion.)

The circuit in Figure 4 uses a CA3096 array to thermally stabilize the feedback transistor in a log amp. This neat trick is also used to thermally stabilize a VCO in National Semiconductor's App Note 286. I'll comment here: Yikes! That LM301A has a huge compensation capacitor! A 33nF capacitor gives a unity-gain frequency of 1 kHz and a slew rate of 0.5 V/ms (yes, volts per millisecond). Replace the LM301A and 33nF with an integrator, and I think you'll get better (and more repeatable) performance. Looking back at Figure 1, is Jim afraid of integrators? I'm guessing Bob Widlar didn't consult on this circuit.

Best quote (page AN5-1): "The close relationship between temperature and electronic devices is the source of more design headaches than any other consideration." Word.

14 July 2011

App Note 4


This app note discusses the LT1010, a power buffer designed by Bob Widlar. Bob Widlar contributes the last page and a half to this app note ("The LT1010 at a Glance") discussing the internal details of the design. Incidentally, App Note 16, in which Widlar further discusses this applications of this buffer, is the app note that breaks Jim Williams' initial streak of writing all of Linear Technology's application notes (Jim wrote all of App Notes 1 through 15).

I love the discrete transistor amplifier designs in this note. Figures 4 and 5 are neat designs (even if there is a low-pass filter in the feedback path of the LT1008 in Figure 4). The motor-speed control loop in Figure 13 makes me happy: no mystery transfer functions in this one (the stability of this control loop can be analytically determined!). Also, I can't help but notice that the high-voltage electrostatic piezoelectric fan used in Figure 14 is another example of "Jim finds the weirdest stuff in his junk pile."

The best circuit is the sample-and-hold design with hold-step compensation in Figure 9, and it is worthy of a grad-school lecture in itself. The charge-dump circuit and the TTL interface are valuable examples. My students always have trouble designing a proper interface to TTL levels (dear students, no, a TTL input is not a 0V and 5V voltage source).

(By the way, there's a schematic error in the last figure on the last page (the LT1010 conceptual schematic).  The OUTPUT pin does not connect to the negative rail.  There should be no connection dot on the output: the output pin is only connected to the right side of R1.)

Best quote (page AN4-2): "With C load increased to a brutal 2 uF, the circuit is still stable, even though the large capacitance requires substantial current from the LT1010." Brutal. I love it.

Finally, I note that there are a lot of National LM101A and LM301A op amps used in this app note (Figures 1, 13, and 14). I suppose that if you're working on a project with Bob Widlar, you use his op amps. Full stop.

13 July 2011

App Note 3


First, we've got an oscilloscope photo with eight traces and two different time scales in Figure 16. It must be a dual-beam Tektronix 556 with two 1A4 plugins. Excellent.

This app note discusses the LTC1043, a board-level switched-capacitor building block, with applications involving platinum RTD temperature sensors, capacitive humidity sensors, and LVDT position sensors. You can see Jim's history in physics and biomedical instrumentation rear its head when he starts discussing instrumentation amps (Figures 2 and 3), and the lock-in amplifier (Figure 4). Unless you have background in experimental science, you probably don't even know what a lock-in amplifier is! (Hint: it's a powerful technique for extracting useful signals from noisy environments. It is the hidden ace in many experimental laboratories. See page 10 of the Analog Devices AD630 datasheet, or read the book by Meade (Lock-In Amplifiers: Principles and Applications, 1983).)

Another early appearance of a classic Williams theme: the schematic in Figure 10 (the LVDT signal conditioner) includes a Wien bridge oscillator (misspelled here as "Wein"), with a FET resistance for amplitude control. (We'll see this circuit again and again, I assure you.) More classic hardware: the tuning procedure for Figure 7 suggests a General Radio 1432-K precision resistor decade box (page AN3-7). Good luck finding one (Hello, flea market).

Frequency-to-voltage and voltage-to-frequency converters are shown in Figure 12. More foreshadowing of future topics, here. Another clever trick in these circuits is the use of a resistor with a temperature coefficient that is opposite of the tempco of the polystyrene capacitor. A single-slope analog-to-digital converter (first of many) is shown in Figure 15.

I think the best circuit is either the lock-in amplifier in Figure 4 (previously discussed) or the the temperature-compensated crystal oscillator with varactor in Figure 21, but there's not much discussion of the latter. Too bad.

Best quote (page AN3-4): "In this [lock-in amplifier] application, the signal source is a thermistor bridge which detects extremely small temperature shifts in a biochemical microcalorimetry reaction chamber." Wait, what?

12 July 2011

App Note 2


Ha! Another classic element of the Jim Williams style: Along with more scope photos from his Tek 547, this app note includes a vacuum-tube circuit in Figure 11A. Plus, it's an Eimac 75TH (which is a gigantic bulbous triode, almost eight inches tall; not really a practical tube to use). This is Jim's practical-joker side.


This app note contains more tricks and tips for using three-terminal regulators. Figure 2 (particularly the presence of the big capacitors) supports my assertion that feedback-loop design with three-terminal regulators is hard and ad hoc because good transfer-function models aren't available. In this case, you've got to add a lot of damping to the loop to make it well behaved. Then, once you've got 100 microfarads on the output, you've got to add Q4 to maintain a quick disable. Similarly, Figure 12, showing an LT1001 precision op amp in the feedback path of an LT317AH, gives me some more feedback-induced indigestion. What's the loop gain around that loop? Yikes.

Figure 5 uses a switching regulator to control the voltage across a linear regulator. Neat trick! The best circuit is the switcher-controlled linear regulator in Figure 7, driven from AC using SCRs that is up to 85% efficient. (I must admit that I have never used an SCR in a circuit design. I am embarrassed.) Note the LM301A in an integrator topology with 100-pF compensation cap and a diode clamp on pin 8. Again, the loop compensation here is very, very conservative.

The final circuit is another 110VAC/220VAC dual voltage solution (a much better approach than Figure 5 in App Note 1). Clearly, he was grappling with this problem. Plus, another SCR!

Best quote (page AN2-5): "Because these two feedback systems are interlocked, frequency compensation can be difficult." No argument here.



UPDATE (with respect to Figure 12, and the LT1001 and LT317A):  As Joe alludes to in his comment below, there are basically two ways to build an adjustable regulator, as illustrated here (with much simplification).

The topology on the left would be totally unstable with an op amp in the feedback path, while the topology on the right might be stable with an op amp in the feedback path (depending on the unity-gain frequency of the op amp and the bandwidth of the regulator).  I was thinking that the LT317A was like the topology on the left, but looking at the datasheet, it's like the topology on the right (either way, it'd be nice to have a transfer-function model).  Thanks Joe!

11 July 2011

App Note 1


Right out of the gate, this app note has the classic Jim Williams style, with five oscilloscope photos, with three or four traces each.  His application notes can often be identified just by glancing at them because they include scope photos showing real data taken with vintage Tektronix gear. You can see the round bezel in several of the photos, so I'm guessing that the instrument was his favored Tektronix 547 with a 1A4 four-channel plugin.

This app note includes eight pages of tricks using the (now-discontinued) LT1005, a dual-output regulator with a main-output-enable pin. Several input protection schemes are implemented using this enable pin (the clever positive feedback latch, shown in Figure 2C on page AN1-2, is the basis for several variations). I think the best circuit here is actually the 100-pF speed-up capacitor in Figure 3 to get the base charge into Q2. I have a soft spot in my heart for circuit tricks from the days of RTL (resistor-transistor logic, you know, the Apollo technology).

The final two pages explore using negative feedback around the regulator to implement closed-loop control. Hysteresis-loop-based motor-speed controllers are shown on pages AN1-7 and 8. I've always viewed some of his feedback contraptions (like these two) with suspicion. His design approach to feedback loops was always a little too cavalier for my tastes, especially since good models for voltage regulators (particularly good transfer-function models) are hard to come by. I'm pretty sure that the phase margin of these loops has never been determined (as a control-systems aficionado, I prefer a more analytical approach).

He seems to have a never-ending supply of interesting, random, and often hard-to-find hardware for his application circuits. In the various circuits, this note specifies a thermoswitch, a thermistor, a crystal-oscillator oven, and two different motor/tachometer units. Although he gives part numbers, it is sometimes difficult to independently source these parts to duplicate his results. I've had trouble finding his components in the past. Sometimes substitutions are easily found, sometimes not. Some quick online searches show that only one of the specified parts in this app note is still easily sourced (the thermoswitch). Often, the only result returned for an online search for the part numbers is just the same application note.

I wonder: was the primary source of these parts from customers looking specific application assistance, or from his personal junk pile? (That said, I actually have some of the little Canon motor/tach units in my personal junk pile, and I love them for small experiments and lab projects.)

Best quote (page AN1-7): "For example, the small motor listed... is almost unstoppable by the unaided human hand at 150RPM." I imagine a group of engineers gathered around the lab bench, nursing friction burns on their fingers after failing to stop the motor.

Order of the App Notes

A note on the order of the app notes: I'm going to read these app notes numerically and not chronologically. If you look at the dates of the first app notes, you can see that they weren't published in numerical order. Chronologically, the order of the first twenty app notes is 1, 2, 4, 5, 6, 7, 13, 8, 10, 3, 11, 12, 15, 17, 18, 14, 21, 9, 23, and 22. (My OCD cannot abide this information.) I'm going to ignore this detail and just read numerically. (I assume the app notes were written in numerical order, but only released when the parts discussed were ready.)

10 July 2011

Introduction

Jim Williams passed away four weeks ago on June 12, 2011.

He was a hero, role model, mentor, and friend. I deeply miss our flea-market sprees, our junk-store pilgrimages, and our oscilloscope-repair adventures. Most of all, I miss our effortless random conversations, discussing history, business, pranks, and, of course, circuits.

Thankfully, he will never be gone completely: he left behind over 350 publications relating to analog circuit design, so generations of engineers can continue to learn from his wisdom.

To fill some of the void left by his untimely departure, I’ve decided to (re)read his seven book chapters and his 62 Linear Technology application notes (over 1600 pages!) and to write some commentary as I do. Reading a long book is often compared to having an intimate conversation with the author. In effect, this is my final conversation with Jim.