29 February 2012

App Note 98 part 1

Signal sources, conditioners and power circuitry: Circuits of the fall, 2004. 28 pages.

This app note is another circuits collection of more-or-less random projects, as he has previously done in App Note 45, App Note 61, and App Note 75.

The app note starts with a simple implementation of a current source in Figure 1. The real application circuit in Figure 2, produces "alternating, equal amplitude, opposed polarity linear capacitor ramps". An interesting choice for the output waveform: I wonder what this signal was used for?

Figure 4 shows a simple sine-wave oscillator (well, really a heavily filtered triangle-wave oscillator). I think the interesting piece of this circuit is the use of the one-sided output from the single-supply A3 to implement half-wave rectification in the amplitude-control loop. Cheesy, but clever.

Figure 6 shows a wide-band high-voltage level shifter, capable of producing 50-volt pulses without overshoot. The footnote exposes a brave design choice:
Transistor data sheet aficionados may notice that the –50V potential exceeds Q1, Q2, Q3 VCEO specifications. The transistors operate under VCER conditions, where breakdown is considerably higher.

The next few circuits include several different pulse generators. He seems to be exploring a wide array of alternatives to his usual avalanche-based pulsers (such as Appendix B in App Note 79). Figure 8 uses a single-chip oscillator that achieves a 400-ps rise time (as shown in Figure 10) and a 320-ps fall time (Figure 11). Nice and simple. Figure 12, inspired by the calibrator circuit from the Tektronix 485 oscilloscope, produces a 850-ps rise time, with a flat-top pulse. Clean and pretty. Figure 15 uses a unique (and discontinued) HP tunnel diode to produce a 20-ps (twenty! picoseconds!) rise time. Yikes. Figure 19 produces pulses with controlled widths, down to 1 ns.

The next three circuits are instrumentation applications. Figure 25 shows a single-supply amplifier, which uses the chopper-clock output to drive a charge pump to allow true-zero-volt output swing. This circuit is them used in the milliohmmeter in Figure 27 and the instrumentation amplifier in Figure 30.

Figure 32 show a wide-band low-feedthrough switch using transconductance amplifiers, developed for settling time measurement (we'll see this circuit again in App Note 120).

I'll cover the rest of the circuits next time.

26 February 2012

Vintage scopes are better part 4

Vintage scopes are better (also see part one, part two, and part three).

Reason number 4: Repairabililty and inspiration.

It is nearly impossible to get a useful service manual for a modern oscilloscope, and even if you could, the parts to repair it are generally unobtainable custom ICs. However, many vintage scopes are repairable, and complete service manuals are readily available for many models. While some vintage parts are getting hard to find (cough, tunnel diodes, cough), many repairs are possible with easily obtained parts. In his second book, Jim explains,
Older equipment offers another subtle economic advantage. It is far easier to repair than modern instruments. Discrete circuitry and standard-product ICs ease servicing and parts replacement problems. Contemporary processor-driven instruments are difficult to fix because their software control is "invisible", often convoluted, and almost impervious to standard troubleshooting techniques. Accurate diagnosis based on symptoms is extremely difficult... Additionally, the widespread usage of custom ICs presents a formidable barrier to home repair. [1]
In addition, vintage scopes are extremely well-designed and well-constructed. Much can be learned from examining and exploring the internals of classic oscilloscope.
The inside of a broken, but well-designed piece of test equipment is an extraordinarily effective classroom... The clever, elegant, and often interdisciplinary approaches found in many instruments are eye-opening, and frequently directly applicable to your own design work. More importantly, they force self-examination, hopefully preventing rote approaches to problem solving... The specific circuit tricks you see are certainly adaptable and useful, but not nearly as valuable as studying the thought process that produced them. [2]
His love of vintage scopes was a continual source of inspiration. He studied, referred to, and borrowed extensively from classic instruments and their service manuals. For example,
  • The high-speed ECL logic in the "King Kong V/F" in App Note 14 was inspired by the trigger circuitry of the Tektronix 2235.
  • The adaptive threshold trigger circuit in Figure 130 of App Note 47 was also inspired by the Tektronix 2235.
  • In his discussion of CCFL power supplies in [3], Jim explored the high-voltage resonant power supplies in the Tektronix 547 and the Tektronix 453.
  • In his discussion of low-noise power conversion (App Note 70, Appendix A), he discusses circuits from the Tektronix 454 and 7904 (Figures A1 and A2).
  • Figure 12 in App Note 98 was derived from the calibrator circuit in a Tektronix 485.


[1] Jim Williams, "There's no place like home," in The Art and Science of Analog Circuit Design, Jim Williams, Ed. Boston: Butterworth-Heinemann, 1995, ch. 17, pp. 269–277.

[2] Jim Williams, "The importance of fixing," in The Art and Science of Analog Circuit Design, Jim Williams, Ed. Boston: Butterworth-Heinemann, 1995, ch. 1, pp. 3–7.

[3] Jim Williams, "Tripping the light fantastic," in The Art and Science of Analog Circuit Design, Jim Williams, Ed. Boston: Butterworth-Heinemann, 1995, ch. 11, pp. 139–193.

22 February 2012

App Note 95

Simple circuitry for cellular telephone/camera flash illumination: A practical guide for successfully implementing flashlamps. 12 pages.

This short little app note with a long title discusses the LT3468, which is a monolithic power-converter chip for driving camera flashtubes. As Figure 1 testifies, flashlamps are much brighter with better color temperature than LEDs, and so are the preferred illuminator for digital cameras. After a brief discussion of the physics, Jim discusses the support circuitry and construction requirements. "Capacitor, wiring and lamp impedances typically total a few ohms, resulting in transient current flow in the 100A range."

This app note contains a single application schematic, shown in Figure 6. Most of the text is dedicated to measurements of the circuit performance (Figures 7 to 10). Figures 9 and 10 are particularly interesting, with vertical scales of "relative light per division". Surprisingly, Jim does not reveal how he made the "relative light" measurements.

A carefully worded caution appears near the end of the text:
In cases where the lamp is triggered with a user-selected transformer and drive scheme, it is essential to obtain lamp vendor approval before going to production.
That quote sounds like the voice of unfortunate experience.

The app note ends with a cartoon. "Nice seeing you again. Yeah, nice seeing you again, too."

19 February 2012

Vintage scopes are better part 3

Vintage scopes are better. (See part one and part two.)

Reason 3: Overdrive resilience.

Several of Jim's App Notes were concerned with the measurement of settling time (for examples, see App Notes 120, 86, 79, 74, 47, and 10). In App Note 79, he measures the 0.1%-settling time of a high-speed op amp (5-millivolt settling on a 5-volt step). In App Note 120, he attempts to measure the LSB-settling time of a 20-bit DAC (10-microvolt settling on a 10-volt step). In these kinds of measurements, the primary concern is avoiding overdrive in the oscilloscope. An input-gain setting that allows the user to see 5-mV-settling details also allows the 5-V waveform to travel off the screen (and overload the vertical amplifier).

In Appendix B of App Note 120, he explains the problem,
Oscilloscope recovery from overdrive is a murky area and almost never specified. How long must one wait after an overdrive before the display can be taken seriously? (App Note 120, page 14)
The horrors of overdrive are readily apparent on page 16.

The waveform looks fine in Figures B2 to B4, but when part of the transient is off the screen, as in Figures B5 to B7, the oscilloscope is clearly having trouble. "It is obvious that for this particular waveform, accurate results cannot be obtained at this gain."

However, in this application, vintage scopes are better. Specifically, the classical sampling oscilloscope is immune to overdrive. To explain, Jim compares the architecture of and analog scope, a digital scope, and a classical sampling scope, all shown in Figure B1.

Describing the advantages of the classical sampling scope, he explains,
The classical sampling oscilloscope is unique. Its nature of operation makes it inherently immune to overload. Figure B1C shows why. The sampling occurs before any gain is taken in the system. Unlike Figure B1B’s digitally sampled ‘scope, the input is fully passive to the sampling point. Additionally, the output is fed back to the sampling bridge, maintaining its operating point over a very wide range of inputs. The dynamic swing available to maintain the bridge output is large and easily accommodates a wide range of oscilloscope inputs. Because of all this, the amplifiers in this instrument do not see overload, even at 1000x overdrives, and there is no recovery problem. Additional immunity derives from the instrument’s relatively slow sample rate—even if the amplifiers were overloaded, they would have plenty of time to recover between samples. (App Note 120, page 14)

A final word from Jim, "Unfortunately, classical sampling oscilloscopes are no longer manufactured, so if you have one, take care of it!"

17 February 2012

App Note 94

Slew rate verification for wideband amplifiers: The taming of the slew. 12 pages.

This app note discusses slew-rate measurement, which is another great application for Jim's high-speed pulse generators. The text is only 12 pages long, but it is not a "brief" project. Previously, he has used a variety of pulse generators for measuring oscilloscope bandwidth and amplifier settling times. For examples of the circuits used in some of these other applications, see
Here he is directly measuring slew rate, which requires even faster pulse rise times.

The app note begins with a quick history lesson in footnote 2,
The term "slew rate" has a clouded origin. Although used for many years in amplifier literature, there is no mention of it on the Philbrick K2-W (the first standard product op amp, introduced in January 1953) data sheet, dated 1964. Rather, the somewhat more dignified "maximum rate of output swing" is specified.
He quickly gets to the heart of the matter, the discussion of the necessary subnanosecond-rise-time pulse generator. The schematic in Figure 8 shows his 360-ps design, which is a great piece of work. This design is an improved version of Figure B1 in App Note 79.

Although the complete schematic is shown, a picture of the actual construction is missing. Much of the "secret sauce" of high-speed pulse generation is in the construction technique (as alluded to in footnote 11), so it is unfortunate that a photoessay isn't included. (I've seen one of these boxes, and it makes Figure B3 in App Note 72 look like child's play.)

Figures 11 to 17 show the calibration of pulse damping, the 360-ps rise time, and the measurement of the LT1818 slew rate. Several of these figures (such as 11, 12, 13, 16, and 18) were made with his Tektronix 7104 with 7A29 and 7B15 plug-ins that he discusses on page 4. These figures are the first appearance of his 7104 (note the legend in Figure 2). I discussed this marvelous scope in a previous post.

Appendix A discusses measurement integrity, using a time-mark generator (Figure A1) and a 20-ps rise-time pulse generator (see the table in Figure A3). Yes, Figure A4 is produced by a pulse with a twenty-picosecond rise time! Appendix B discusses level-shifting (no active circuits here; your only hope is to use excellent transformers).

Appendix C discusses the horrors of connectors, cables, adapters, attenuators, and probes at these high speeds. The best quote is on page 11:
Skepticism, tempered by enlightenment, is a useful tool when constructing a signal path and no amount of hope is as effective as preparation and directed experimentation.
The cartoon shows off his 7104 scope, with a 7M13 alphanumeric readout plug-in (which produced his "signature" WILLIAMS03 in the corner). "I'm going as fast as I can."

12 February 2012

Vintage scopes are better part 2

Vintage scopes are better. (See the introduction in part one.)

Reason number 2: Sensitivity and bandwidth. With the appropriate plug-ins, analog oscilloscopes provide superior sensitivity compared to digital scopes. In discussing low-level noise measurements in App Note 70, Jim describes the oscilloscope requirements and laments,
Current generation oscilloscopes rarely have greater than 2mV/DIV sensitivity, although older instruments offer more capability. Figure B11 lists representative preamplifiers and oscilloscope plug-ins suitable for noise measurement. These units feature wideband, low noise performance. It is particularly significant that the majority of these instruments are no longer produced. This is in keeping with current instrumentation trends, which emphasize digital signal acquisition as opposed to analog measurement capability. (App Note 70, page 29)
While 2 millivolts-per-division is commonplace in digital oscilloscopes, plug-ins are available for 500-series and 7000-series scopes with sensitivity down to 10 microvolts-per-division. Yes, microvolts. In Appendix D of App Note 124, Jim lists the high-sensitivity, low-noise amplifiers of choice.

Of course, sensitivity and bandwidth are related (the wider the bandwidth, the higher the expected noise floor). However, in conjunction with superior noise floor, some vintage analog scopes also provide very large bandwidths. Some of Jim's favorites were
  • Tektronix 556 with a 1S1 sampling plug-in, 1-GHz bandwidth (App Note 72, page 9, Figures 16 and 17)
  • Tektronix 547 with a 1S2 sampling plug-in, 3.9-GHz bandwidth (App Note 79, page 19, Figure B4)
  • Tektronix 661 with a 4S2 sampling plug-in, 3.9-GHz bandwidth (App Note 72, pages 34 and 35, Figure 77 to 82)
  • Tektronix 7104 with 7A29 and 7B15 plug-ins, 1-GHz real-time bandwidth (App Note 94, page 4, particularly Figures 2, 11, 12, 13, 16, and 18)
There are modern digital scopes available today with wider bandwidths, but "vintage" and "analog" does not mean "slow." Of course, Jim would want me to point out that bandwidth isn't everything. In App Note 47, he explained,
Intimate familiarity with your oscilloscope is invaluable in getting the best possible results with it. In fact, it is possible to use seemingly inadequate equipment to get good results if the equipment’s limitations are well known and respected. All of the circuits in the Applications section involve rise times and delays well above the 100MHz-200MHz region, but 90% of the development work was done with a 50MHz oscilloscope. Familiarity with equipment and thoughtful measurement technique permit useful measurements seemingly beyond instrument specifications. A 50MHz oscilloscope cannot track a 5ns rise time pulse, but it can measure a 2ns delay between two such events. Using such techniques, it is often possible to deduce the desired information. (App Note 47, page 20)
To be honest, the first sentence of that quote applies no matter what oscilloscope you have.

Footnote: One last comment while we're discussing plug-in oscilloscopes. The Tektronix 556 dual-beam instrument provides flexibility that is not found in modern instruments.
The Tektronix 556 offers an extraordinary array of features valuables in converter work. This dual beam instrument is essentially two full independent oscilloscopes sharing a single CRT. Independent vertical, horizontal and triggering permit detailed display of almost any converters operation. Equipped with two type 1A4 plug-ins, the 556 will display eight real time inputs. The independent triggering and time bases allow stable display of asynchronous events. Cross beam triggering is also available, and the CRT has exceptional trace clarity. (App Note 29, pages 43-44)
In App Note 65, he exploited these dual-beam advantages in a number of measurement. Figure 36 shows six waveforms, with independent triggering of the top two versus the bottom four traces. Figure 42 shows the ringing bursts at the resonant frequency of the Royer converter, with the explanatory footnote
The discontinuous energy delivery to the loop causes substantial jitter in the burst repetition rate, although the high voltage section maintains resonance. Unfortunately, circuit operation is in the "chop" mode region of most oscilloscopes, precluding a detailed display. "Alternate" mode operation causes waveform phasing errors, producing an inaccurate display. As such, waveform observation requires special techniques. Figure 42 was taken with a dual-beam instrument (Tektronix 556) with both beams slaved to one time base. Single sweep triggering eliminated jitter artifacts. Most oscilloscopes, whether analog or digital, will have trouble reproducing this display. (App Note 65, page 38)
Finally, the flexibility of the Tektronix 556 allows for some great measurement displays. In Figure 34 of App Note 35, he showed a 115-volt sine wave, its distortion products, and its frequency spectrum all in one shot.

In the accompanying footnote, Jim teased,
Test equipment aficionados may wish to consider how this picture was taken. Hint: Double exposure techniques were not used. This photograph is a real time, simultaneous display of frequency and time domain information. (App Note 35, page 16)
This picture was (most probably) produced with his trusty Tektronix 556 with a vertical-amplifier plug-in in one bay (perhaps a 1A2 or 1A4), and a spectrum-analyzer plug-in in the other bay (perhaps the 1L5 50Hz-to-1MHz spectrum analyzer).

10 February 2012

App Note 93

Instrumentation applications for a monolithic oscillator: A clock for all reasons. 20 pages.

This app note discusses applications of the LTC1799 oscillator chip. The approach here is classic Jim: Linear Technology produces a simple oscillator chip, and Jim's first thought is "temperature measurement!" The frequency of the LTC1799 is set with a single resistor, so Jim immediately replaces it with a platinum RTD (Figure 5) or a thermistor (Figures 7 and 9), in a variation of his normal voltage-to-frequency-converter theme. Figure 9 also includes galvanic isolation. Other unique sensors also appear. Figures 11, 13, and 15 use a fragile capacitive humidity sensor (which we've seen before), using the LTC1799 as either a reference oscillator or a clock source.

Figures 17 and 19 continue the instrumentation theme with a chopper-stabilized amplifiers. Figure 17 uses a bipolar-op-amp input, while Figure 19 uses JFETs for lower bias current, but slightly higher noise. He also includes two sine-wave generators, Figure 21 uses a resonator loop and Figure 23 is a memory-based DDS (see also Appendix E in App Note 35).

A few almost-digital circuits wrap up the note. A clock-tunable notch filter is shown in Figure 28 and a clock-tunable interval generator in shown in Figure 30. The last circuit (Figure 32) is a single-slope analog-to-digital converter that shares some topology with the interval generator.

The two appendices discuss the LTC1799 in a little more detail. Appendix A discusses the internal topology and the master oscillator, and Appendix B discusses the care and feeding of the RSET pin (and potential bootstrapping).

The best quote is the pedantry in the first footnote,
Strictly speaking, an oscillator (from the Latin verb, "oscillo," to swing) produces sinusoids; a clock has rectangular or square wave output. The terms have come to be used interchangably and this publication bends to that convention.
The cartoon extols the simplicity of the LTC1799, "Everything should be as simple as possible, but not too simple."

08 February 2012

App Note 92

Bias voltage and current sense circuits for avalanche photodiodes: Feeding and reading the APD. 32 pages.

More fiber-optic-laser support circuits (see App Notes 89 and 90) are discussed here, in particular, the design of voltage-biasing supplies and current-monitoring circuits for avalanche photodiodes (APD). These circuits are used on the receiving end of an optical communications system.

Monitoring supply current is hard, especially in a high-voltage, single-supply application. Jim works through several approaches, starting with a direct instrumentation amplifier in Figures 2 and 3, both of which are impractical. His first real solution is the AC-coupled lock-in amplifier in Figure 4, using an LTC1043 and a charge pump to create a negative power-supply voltage (remember Figure 4 in App Note 3?). A DC-coupled amplifier is shown in Figure 5, with an amplifier floating up at the high-voltage rail.

Several high-voltage power supplies are presented, with output-noise measurements. A programmable flyback (with voltage-tripler) APD voltage supply is shown in Figure 6. A careful measurement of the 200-microvolt output noise in shown in Figure 7. Figure 8 combines the features of Figures 5 and 6. Figure 9 implements the features of Figure 8 using a transformer instead of a voltage-tripler circuit for the high-voltage output.

Figure 11 is a simple inductor-based design, but it requires a cascode on the LT1946, which gives Jim an excuse to reference the 1939 paper by Hickman and Hunt again (see reference 11). Figure 13 and 15 produce the low-noise supply voltage using the LT1533 controller (which was discussed, at length, in App Note 70). Again, the pedantry in footnote 11 makes me very happy,
Noise contains no regularly occurring or coherent components. As such, switching regulator output "noise" is a misnomer. Unfortunately, undesired switching related components in the regulated output are almost always referred to as "noise." Accordingly, although technically incorrect, this publication treats all undesired output signals as "noise."
Figures 16 and 18 use a crazy flying-capacitor scheme, reminiscent of Figure 23 in App Note 3, this time with optically driven switches. Figures 19 and 20 discuss digital schemes for output-current monitoring.

The various circuits are summarized in the table in Figure 21, along with Jim's usual love for summary tables,
Figure 21's chart is an attempt to summarize the circuits presented, although such brevity breeds oversimplification. As such, although the chart reviews salient features, there is no substitute for a thorough investigation of any particular application’s requirements.

Appendix A discusses two schemes for deriving the feedback voltage on the high-voltage supply, while minimizing the induced error in the current measurement. Appendix B discusses a few reasons why vintage oscilloscopes are so prized in his laboratory.

Appendix C reprints Appendix C from App Note 70. (He is getting quote a bit of mileage out of this Appendix, but it makes good sense: once you've written the definitive treatise on noise measurement, you might as well use it!)

Appendix D discusses the LT1150 chopper-stabilized amplifier, which has a clock-output pin that can be used to charge-pump a negative supply voltage in single-supply applications. "The circuit provides a simple way to obtain output swing to zero volts, permitting a true "live at zero" output."

Appendix E lists a few protection circuits for the expensive APD module, including a voltage clamp, a current limiter, and a bias-voltage crowbar.

The app note ends with a cartoon, betraying some obvious frustration with coworkers, perhaps? "Someday, when I don't have to make anything work, I'm gonna wear a tie too."

05 February 2012

Vintage scopes are better part 1

I was discussing these application notes with a colleague, and he commented, "Jim sure did love his vintage oscilloscopes. I wonder, is there anything that you can do with a vintage scope that you can't do with a modern digital one?"

"Yes!" I cried, and I listed off several things, but I don't think I convinced him. Over the next four Sundays, I'm going to enumerate and explain the list of reasons why vintage scopes are better than modern digital abominations, including
  1. Trace clarity (resolution and spot size)
  2. Sensitivity and bandwidth (and noise floor)
  3. Overdrive resilience (of sampling plug-ins)
  4. Repairability and inspiration
Mostly, I'll just be quoting the relevant passages in Jim's app notes. Truly, they don't make them like they used to...

Reason number 1: Trace clarity. The low-level-measurement resolution of a oscilloscope is limited, in part, by the minimum size of the trace on the screen. A well-designed (and well-calibrated) vintage scope can have a vanishingly small spot size on the CRT. With a digital scope, the resolution of the input analog-to-digital converter often becomes apparent as you increase the vertical gain on a digital scope (or you are limited by the size of the LCD pixels).

Discussing oscilloscope selection for low-level noise measurements in Appendix B of App Note 70, Jim commented,
The monitoring oscilloscope should have adequate bandwidth and exceptional trace clarity. In the latter regard high quality analog oscilloscopes are unmatched. The exceptionally small spot size of these instruments is well-suited to low level noise measurement. (Footnote: In our work we have found Tektronix types 454, 454A, 547 and 556 excellent choices. Their pristine trace presentation is ideal for discerning small signals of interest against a noise floor limited background.) (App Note 70, page 29)
He continues this train of thought to say,
The digitizing uncertainties and raster scan limitations of DSOs impose display resolution penalties. Many DSO displays will not even register the small levels of switching-based noise. (App Note 70, page 29)
Again discussing noise measurements in Appendix C of App Note 90, Jim says,
Diehard curmudgeons still using high quality analog oscillscopes routinely discern noise presence due to trace thickening. Those stuck with modern instruments routinely view thick, noisy traces. (App Note 90, page 22)
The detail that is visible with well-focused oscilloscope trace is evident on Jim's Tektronix 556 in Figures B9 and B10 in App Note 70.

I'd like to see these plots replicated on a modern all-digital scope.

03 February 2012

App Note 90

Current sources for fiber optic lasers: A compendium of pleasant current events. 24 pages.

This app note continues the fiber-optic-laser-diode theme from App Note 89. Now, attention is paid to the power supply of the laser diode itself, a precision current source. The app note begins with a discussion of performance requirements and protection issues for the laser. "The delicate, expensive load, combined with the uncertainties noted, should promote an aura of thoughtful caution."

Figure 2 shows a simple current-source topology, which is refined in various ways throughout the app note. Figure 3 uses a switching regulator for improved efficiency. I like his intuitive explanation that "It is useful to liken the switching regulator’s input (VCC), feedback (FB) and output (VSW) to the transistor’s collector, base and emitter."

Figure 4 uses an LT1970 "power op amp" to allow a grounded cathode on the laser diode. I have to laugh whenever I see a "power op amp" that comes in a TSSOP body. A real power op amp, like Bob Widlar's LM12, comes in a TO-3 package! (It also drinks scotch.)

The best circuit is Figure 6, which includes a full array of protection circuits: low-voltage drop out, a self-enable regulator, an input current clamp, and an open-circuit lock out. The oscilloscope traces in Figures 7 to 11 demonstrate the performance and operation of these features.

The rest of the circuits in the app note implement low-"noise" switching regulators to provide higher-power output (I heartily endorse footnote 2 on page 8). Figure 15 uses the LT1683, while Figures 17 and 19 use the LT1533 (discussed in App Note 70).

Appendix A discusses using regular diodes and a simple circuit (Figure A2) for simulating a laser diode. Appendix B discusses switching regulator "noise" measurement, using excerpts from App Note 70. Appendix C discusses current transformers for low-level measurements. The Tektronix CT-1 is very nice. The best quote is the footnote in Appendix C,
Diehard curmudgeons still using high quality analog oscillscopes routinely discern noise presence due to trace thickening. Those stuck with modern instruments routinely view thick, noisy traces.

The app note ends with a cartoon, ghost-written by Jim's wife.

01 February 2012

App Note 89

A thermoelectric cooler temperature controller for fiber optic lasers: Climatic pampering for temperamental lasers. 12 pages.

This app note discusses thermoelectric coolers (TECs) for laser-diode temperature control. The Figure 3 shows the main circuit, built around the LT1923 TEC-controller chip. Most of the app note (pages 3 to 7) discusses compensating the temperature-control feedback loop. As Jim says, "The unfortunate relationship between servo systems and oscillators is very apparent in thermal control systems." Figures 6 through 9 show a trial-and-error approach to setting the loop gain.

Best quote (with footnote):
Figure 6 shows large-signal oscillation due to thermal lag dominating the loop. A great deal of valuable information is contained in this presentation. (When a circuit "doesn’t work" because "it oscillates," whether at millihertz or gigahertz, four burning questions should immediately dominate the pending investigation. What frequency does it oscillate at, what is the amplitude, duty cycle and waveshape? The solution invariably resides in the answers to these queries. Just stare thoughtfully at the waveform and the truth will bloom.)
Wise words.

Once the loop is "optimized", as in Figure 9, the temperature "stability" is measured (Figures 10 through 12). The short-term stability is good (Figure 10), but disturbance rejection (the response to changes in ambient temperature) in Figure 11 could be better. In this application, proportional control is good, but proportional-plus-integral control would be better. Jim likes to mock "control aficionados" (see footnote 7), but really, we can help!

The last section of the app note discusses switching-regulator "noise" (ripples and spikes) in the voltage to the laser diode. Like the LT1533 discussed in App Note 70, this controller also uses slew-rate control on the switching edges to minimize the high-frequency harmonic content. Figures 13 to 15 and Figure 17 to 19 show the advantage.

The app note ends with a cartoon. "That kind of talk makes me feel so coherent."