Build From Scratch (open-source crt-driver)

Every other build volume in this series hands you a finished design and a parts list and asks you to assemble it faithfully. This one is different. The from-scratch path is for the builder who wants the analog electronics under their own control — a stiff kilovolt supply, true differential deflection amplifiers, a 6 MHz video amp on the cathode — paired with a CRT of their own choosing and a vector source of their own design. There are two honest ways to walk this path, and they sit at very different points on the effort curve. The first is to adopt Eric Schlaepfer’s (TubeTime’s) open-source crt-driver board set: three small PCBs that together turn a 12 V input and a handful of 3.3 V analog signals into a glowing vector display on almost any 2-to-5-inch electrostatic tube. The second is to design the whole analog chain yourself from the engineering principles in Vols 3 through 5, treating the TubeTime design as a reference architecture rather than a kit. This volume covers both, plus the alternative open references (Sgitheach, NYC Resistor) and a frank assessment of where this path beats buying a kit (Vol 9) or building one of the two owned units (Vols 7 and 8).

10.1 Why the crt-driver is the recommended foundation

The crt-driver project is the cleanest open-hardware starting point for a from-scratch scope clock that exists. It was released by Eric Schlaepfer in May 2014 under the Creative Commons Attribution-ShareAlike 3.0 license, and it is the same board set he used to drive the CRTs in his Asteroids and Flappy Bird vector arcade cabinets at Maker Faire. That pedigree matters: this is not a paper design, it is a proven, demonstrated, field-driven amplifier chain that has been built and rebuilt by the community for over a decade.

Three properties make it the right foundation. First, it is general: the deflection board takes 0-3.3 V analog X and Y plus a 0-3.3 V video/blanking signal and outputs the high-voltage plate drive — exactly the interface a microcontroller-plus-DAC vector source naturally produces (see Vol 5). Second, it is modular: power, deflection/video, and bias are three separate PCBs wired together, so you can build, bring up, and debug each one independently, and you can substitute your own design for any single board later. Third, it is broad: the compatibility list covers most popular 2-, 3-, and 5-inch tubes — the 3BP1, the precision 3RP1A, the 2AP1, and dozens more — so you are not locked to one CRT (tube selection is Vol 6’s subject).

What you do not get is the schematic and layout source. Schlaepfer designed the boards in Altium Designer and chose not to publish the native project files, on the (reasonable) grounds that Altium is not a tool hobbyists own. What the GitHub repository ships instead is the fabrication output for each board — Gerbers, NC drill files, an OutJob print configuration — together with a CSV bill of materials carrying Mouser part numbers, and a PDF schematic print of each board. That is enough to order the PCBs, buy every part, and understand the circuit; it is not enough to edit the layout. If you want to modify a board, you re-draw it. This is the one real friction point of the project and it shapes the decision in § 10.6.

FIGURE SLOT 10.1 — Photo of the three assembled crt-driver boards (power, deflection/video, bias) wired to a small round CRT showing a vector image. Suggested source: the two photographs already staged in 02-inputs/A - Open Source/ (Dx30yHGU8AAkLDC.jpg, Dx3zyfyU8AAAZpv.jpg), or a Photo Helper pagefetch from tubetime.us/?p=183.

10.1.1 The repository at a glance

The repository (02-inputs/A - Open Source/crt-driver-master/, mirror of github.com/schlae/crt-driver) is organized as one Altium design-workspace (ScopeClockDrivers.DsnWrk) referencing several board projects. Each board lives in its own folder holding a .PrjPCB project, a .SchDoc schematic, a .PcbDoc layout, a CSV BOM, a PDF schematic print, and a zipped Gerber set (SP1.zip, SD1.zip, SV1.zip, and so on). Two top-level PDFs tie the project together: ScopeWiring.pdf, the interconnect drawing that shows how the three boards and the CRT join up, and BaseDiagrams.pdf, a sheet of tube-socket pinout diagrams keyed to the CompatibilityList.txt. The folders you will actually use for a scope clock are ScopePower (the 1 kV supply), ScopeDefl (the deflection and video amplifier), and ScopeVideo (the 60 V bias supply). The remaining folders — ScopeMag and ScopeVideoOnly — are siblings from Schlaepfer’s wider CRT work (a magnetic-deflection variant and a standalone video-only board) and are not part of the electrostatic clock chain; ignore them unless you have a specific reason.

10.2 The three boards in depth

The whole design reduces to a clean signal story: 12 V DC and a 6.3 V AC filament feed come in; the power board makes the kilovolt anode rail; the bias board makes a stiff 60 V rail for the video amplifier; and the deflection board consumes all three rails plus the 3.3 V vector signals and produces everything the CRT’s deflection plates and electron gun need. ScopeWiring.pdf draws it exactly that way — four input sources (12 V, 6.3 V AC, and the two HV/bias boards) all converging on the deflection board, which fans out to the CRT’s deflection pins and gun pins.

10.2.1 Board 1 — the 1 kV power supply (ScopePower)

This board takes a 12 V input and steps it up to roughly 800-1200 V DC, adjustable. It is the single most dangerous thing in the entire build, and the README states the danger plainly: the supply can source several milliamps, and “voltages this high with even just a few milliamps can stop your heart.” Treat that sentence as the governing safety constraint of the whole volume. The kilovolt rail is not a feeble high-impedance bias that a finger-brush shrugs off; it is a genuine, heart-stopping, low-impedance source. Every rule in Vol 12 — one hand in your pocket, discharge before you probe, never work alone, plastic-handled trimmer tools only — is in force the moment this board is energized.

Topologically the supply is a Royer oscillator driving an off-the-shelf CCFL transformer (T1, a CTX210605), with the secondary rectified and filtered into the kilovolt rail. The design is explicitly inspired by Jim Williams’ CCFL circuits in Linear Technology’s Application Note 65; where Williams used a switching regulator to drive the Royer tank, Schlaepfer uses a simpler linear FET approach (Q1, an FDB3502) that works well but dissipates more heat — which is why the board has a plated through-hole above Q1 for a screw-down heatsink. Critically, the README warns that the Q1 tab node is not ground, so if you want the heatsink electrically isolated, use a mica washer and a nylon screw. A ZTL431A adjustable shunt regulator (U1) and a TS271 op-amp (U2) close the regulation loop around a 10 MΩ-class sense divider, and the output is tuned by the R1 10 kΩ trimmer. The rectifier (D1-D4, RGP02-20E) and the 2-3 kV-rated film capacitors (C3, C7, C8) live on the secondary side.

Two design notes from the BOM and blog are worth carrying forward. The supply produces stray magnetic fields that will pull the electron beam around in ways you do not want, so the board must be kept physically far from the CRT and the tube shielded with mu-metal or soft steel (see Vol 6 and Vol 11). And the design supports post-deflection acceleration (PDA): by tapping into diode D1 and adding a 2-stage voltage multiplier (shown on the schematic), you can generate roughly 3 kV for the larger 3-inch and 5-inch tubes that require it. The compatibility list flags which tubes are PDA types.

FIGURE SLOT 10.2 — Annotated schematic-print excerpt of the ScopePower 1 kV board highlighting the Royer tank (T1/Q1), the rectifier/filter, the sense divider, and the PDA multiplier tap on D1. Suggested source: ScopePower/ScopePower.pdf in the repo.

10.2.2 Board 2 — the deflection and video amplifier (ScopeDefl)

Schlaepfer calls this “the heart of the circuit,” and it is. It contains two fully differential high-voltage amplifiers (one for X, one for Y) that act as both amplifier and level shifter, driving the deflection plates, plus a high-speed video amplifier of about 6 MHz bandwidth driving the CRT cathode. All inputs are 3.3 V analog: you feed it 0-3.3 V on X, 0-3.3 V on Y, and either a 0-3.3 V analog video signal or a 3.3 V logic-level blanking signal on video-in. This is the board your vector source talks to (§ 10.4).

The deflection amplifier architecture is worth understanding because it is exactly the differential-pair-plus-cascode topology Vol 4 develops in the abstract. The core of each channel is a matched dual NPN transistor (Q1/Q2, a DMMT3904 dual) forming a differential pair, with a bias network setting gain (width/height) and offset (position) — these are the XPOS/XWID/YPOS/YWID trimmers (R1-R4). A cascode stage built from two high-voltage NPN transistors (the STN0214 parts, Q3-Q6) increases bandwidth slightly and level-shifts the signal up to about 1 kV. The plate loads are the four 220 kΩ high-voltage resistors (R30-R33, PR01-series); the README is emphatic that these must not be substituted with ordinary carbon-composition parts — they have to be rated for the high voltage they actually see. The amplifier bandwidth is only about 10-15 kHz, which is plenty for clock-face vector work and even enough to draw an NTSC raster sweep if you wanted, but it is a reminder that this is a deflection amplifier, not an RF amplifier — slew limits matter when you plan the firmware’s drawing order (Vol 5).

The video amplifier takes the opposite, cleverer approach to beam modulation. The naive method — ground the cathode and drive the grid negative — leads to an awkward circuit. Instead the design holds the grid at an adjustable brightness voltage (the BRIGHT trimmer R7) and drives the cathode from the video amplifier output. The result is the same beam modulation with a far simpler amplifier, at the cost of a slight interaction with the focus electrode. It is a class-A design with a cascode speed-up stage (Q7 PZTA42, Q8 PZT2222A), and it draws fairly high current to hit 6 MHz — which is precisely why it needs the stiff 60 V bias rail from board 3. The board also carries the focus (R5, 500 kΩ) and astigmatism (R6, 500 kΩ) trimmers: focus sets the focus-grid bias over a range covering most 2-, 3-, and 5-inch tubes, and astigmatism trims the final-accelerator voltage relative to the average plate voltage — adjust it if the spot looks oval instead of round. There is a CONT (contrast) trimmer (R8) on the video path as well.

A subtle but important detail lives on the filament connector (J4, 6.3 V AC): pin 1 is tied to ground. For tubes whose heater and cathode are internally connected (the “HC” tubes in the compatibility list), you must route the common leg to the correct heater pin or you will short the filament transformer — the compatibility list spells out that this leg goes to pin 5 of J2, not pin 4. The README also notes that filament transformers commonly run high (7 V is not unusual under load), so measure the actual voltage and add a series resistor to bring it down and prolong tube life.

10.2.3 Board 3 — the 60 V video bias supply (ScopeVideo)

The deflection board’s video amplifier needs a stiff 60 V supply because its drive current is high (that is the price of 6 MHz bandwidth). This third board provides it from the same 12 V input as everything else. It is a compact boost converter built around a MAX1771 controller (U1) switching a SI4434DY MOSFET (Q1) through a 68 µH inductor (L1), with a MURA240 Schottky rectifier (D1), a ZTL431A reference (U2), and a 22 µF 160 V output cap (C5). The output voltage is set by the feedback divider, principally R1 (390 kΩ).

The README’s “easter egg” is genuinely useful: this board can be adapted as a Nixie-tube power supply by changing the value of R1 and increasing the voltage rating on C5. If you are building a hybrid clock — a scope-CRT main display with Nixie sub-dials, say — that one substitution turns a spare bias board into a 170-180 V Nixie rail. (The BOM lists C5 as a 22 µF 160 V part already, so for a Nixie rail near 170-180 V you would step the rating up to 200 V or beyond and recompute R1.)

FIGURE SLOT 10.4 — Schematic-print excerpt of the ScopeVideo 60 V bias board with the R1/C5 “Nixie supply” modification annotated. Suggested source: ScopeVideo/ScopeVideo.pdf.

10.2.4 Ordering the boards from OSH Park

You do not have to wrangle the Gerbers yourself — Schlaepfer published each board as an OSH Park shared project, so ordering is a matter of opening the link, clicking order, and receiving three of each board in purple solder mask a couple of weeks later (OSH Park always ships in threes). The three shared-project URLs are:

• 1 kV power supply: oshpark.com/shared_projects/PMsJvDGz
• Deflection & video amplifier: oshpark.com/shared_projects/stZ49KK7
• 60 V video bias supply: oshpark.com/shared_projects/F2lrTWvo

If you prefer another fab (JLCPCB, PCBWay), upload the zipped Gerber set from the corresponding board folder (SP1.zip, SD1.zip, SV1.zip). Every component is stocked at Mouser and the CSV BOMs carry the Mouser part numbers directly, so a cart is a copy-paste exercise — note the blog’s “CRT Board BOM Updates” caveat (June 2014) and reconcile any since-obsoleted line items against current Mouser stock before you order.

10.3 Pairing the board set with a CRT

The board set is only half a clock; it needs a tube. The repository’s CompatibilityList.txt is the authoritative pairing guide, and Vol 6 is where tube selection, sourcing, and economics are treated in full — read it before you buy. Here the job is narrower: understand what the compatibility list is telling you and how the wiring follows from it.

10.3.1 Reading the compatibility list

The list enumerates tubes by type (913, 1EP1, 2AP1, 3BP1, 3RP1A, 5ADP1, and so on), giving for each one the base (socket type, keyed to BaseDiagrams.pdf), the filament voltage (6.3 V for most, 2.5 V for some 3- and 5-inch types and the 905/907), a compatible flag, and a maximum A2 (anode) voltage. Several annotations recur and each carries a build consequence:

• HC (Heater/Cathode) — the cathode is internally tied to one leg of the heater. As noted in § 10.2.2, the common leg must go to pin 5 of J2, never pin 4, or you short the filament transformer. Many 5-inch tubes (905, 907, 5BP1) and several 3-inch types are HC.
• PDA (Post Deflection Acceleration) — the tube needs a multiplied anode (3 kV+), which means installing the optional voltage-multiplier on the power board (§ 10.2.1). The 3FP*, 3JP*, 5CP*, and 5ADP* families are PDA.
• Untested / Poor performance / Very poorly / Unlikely / Fuzzy — Schlaepfer’s honest field notes. “Untested” means the board should work but he has not personally verified it; “Unlikely” (the 5RP*/5XP* PDA types) means the PDA voltage is probably beyond what the supply can safely reach. The 3XP* gives “poor performance,” and the 5UP* is “fuzzy.”

Two voltage cautions govern the small end: for tubes 2 inches and under, keep A2 below 1 kV to avoid damaging the tube, and the 1DP* types need a power-supply modification (their 600 V limit is below the board’s natural floor). The circuit has not been tested with tubes larger than 5 inches. For a first build, the sweet spot is an untested-free 3-inch non-PDA, non-HC tube — the classic 3BP1 (14A base, 6.3 V filament) or the 3RP1/3RP1A (12E base) — which is exactly what the blog recommends and exactly the kind of tube Vol 6 helps you source.

10.3.2 Wiring per ScopeWiring.pdf and BaseDiagrams.pdf

ScopeWiring.pdf is the master interconnect. The 12 V supply feeds J1 on each of the three boards. The 6.3 V AC filament transformer feeds J4 on the deflection board. The 1 kV board’s HV output (J4) goes to the deflection board’s HV input (J5); the 60 V board’s output (J2) goes to the deflection board’s 60 V input (J6). The deflection board then presents two connectors to the tube: J1 to the CRT deflection pins and J2 to the CRT electron-gun pins (filament, cathode, grid, focus, A1, A2). The 3.3 V vector signals (X, Y, video from your DACs) arrive at J3 on the deflection board.

BaseDiagrams.pdf gives the socket pinout for each base type — 11A, 11L, 12E, 14A, 14B, 14G, 14K, and the rest — labeling each pin H (heater), K (cathode), G (grid), A1, A2 (anodes/focus), D1-D4 (the four deflection plates), and IC/NC (internal-connection / no-connection). To wire a given tube you cross-reference three documents: the compatibility list tells you the base and any HC/PDA caveat; BaseDiagrams.pdf tells you which physical pin is which electrode for that base; and ScopeWiring.pdf tells you which board connector each electrode lands on. Build a small pin map on paper before you cut a single wire, and double-check the HC heater leg and the deflection-plate orientation (swapping D1/D2 with D3/D4 mirrors the image — a nuisance, not a hazard).

10.4 Supplying the vector source yourself

The crt-driver board set deliberately stops at the analog interface: it expects 3.3 V analog X and Y plus a 3.3 V video/blanking signal on J3 and asks no questions about where they come from. That is your job, and it is the same job whether you adopt the TubeTime boards or roll your own analog chain. Three approaches are practical, in rising order of capability, and all of them are developed in depth on the firmware side in Vol 5.

10.4.1 A microcontroller plus a dual DAC

The canonical scope-clock vector source is a microcontroller emitting X and Y through a pair of DACs, with a GPIO pin (or a third DAC channel) carrying the Z/blanking signal. Because the crt-driver wants 3.3 V analog, a 3.3 V microcontroller is the natural fit — an STM32, an RP2040, an ESP32, or any modern part with two fast DAC channels or an external dual-DAC over SPI (an MCP4822-class part, or a faster parallel DAC for higher refresh). The firmware keeps time, walks a display list of stroke endpoints, and updates X, Y, and Z thousands of times per second; the deflection amplifier’s ~10-15 kHz bandwidth sets how fast you can move the beam between points without visible slewing, which in turn caps how complex a face you can draw flicker-free. This is Vol 5’s entire subject — display-list structure, blanking discipline, font and hand rendering, and the slew-vs-flicker trade — and the crt-driver is the ideal analog back end for it because its input is a clean 3.3 V analog bus rather than a proprietary connector.

10.4.2 An FPGA

If you want more points per frame, a deterministic refresh, or you are driving a demanding face (dense text, anti-flicker dithering, multiple sub-dials), an FPGA driving external DACs removes the microcontroller’s interrupt-jitter ceiling. The FPGA generates the X/Y/Z stream from a display list in block RAM with cycle-accurate timing; the DAC outputs feed the same 3.3 V J3 inputs. This is overkill for a wall clock but is the right tool if the scope clock is also meant to run vector games or animations (which is, after all, what Schlaepfer built the boards for). Vol 5 discusses where the FPGA’s determinism earns its complexity.

10.4.3 The Nuts & Volts Arduino XY Graphics Interface

A ready-made, well-documented vector source is Edward Andrews’ Arduino Graphics Interface (AGI) from Nuts & Volts (the two-part article and the XYscope library, held in 02-inputs/Arduino XY Graphics Interface/). The AGI uses the dual DACs inside the Atmel SAM 32-bit CPU of an Arduino Due, driven by the chip’s DMA so the X/Y waveform streams out with little CPU overhead, and the XYscope library provides the plotting and font primitives — the article even shows it configured as a “CRT CLOCK.” The AGI was written to drive a bare analog oscilloscope’s X/Y inputs directly, but its DAC outputs are exactly the X/Y/Z vector bus the crt-driver deflection board wants. The one caution is level: the Due’s DACs swing roughly 0.55-2.75 V, not a full 0-3.3 V, so you may want a small scaling/offset buffer between the AGI and J3 to use the deflection amplifier’s full range — a trivial op-amp stage. Used this way, the AGI is the fastest route from “I have the boards” to “it draws the time,” and it cross-references directly to Vol 5’s firmware discussion.

10.5 Designing your own board entirely from scratch

If you want to go all the way — your own schematic, your own layout, your own firmware — the crt-driver becomes a reference architecture rather than a kit, and Vols 3 through 5 become the engineering manual you live in. The work is real, but the architecture is well-trodden. Use the following checklist as the skeleton of your own design; each item points back to the volume that develops it.

10.5.1 Architecture checklist

• Supply rails (Vol 3). You need four: a kilovolt-class anode rail (800 V-1.2 kV for 2-3-inch tubes, 3 kV via a PDA multiplier for larger ones), a focus rail (a high-impedance bias divided from the anode), a stiff mid-voltage rail for the video amplifier (the crt-driver’s 60 V), and a filament supply (6.3 V or 2.5 V AC/DC per the tube). The kilovolt rail is the lethal one — design it with a bleeder, a sense divider that bleeds the output in a second or two after shutdown, and generous creepage. The Royer-plus-CCFL-transformer topology the crt-driver uses (after Jim Williams’ AN65) is a proven starting point.
• Differential deflection amplifiers (Vol 4). Two channels, each a matched differential pair feeding a high-voltage cascode that level-shifts to the plate operating point, with HV-rated plate-load resistors (not carbon comp) and trimmers for gain (width) and offset (position) per axis. Budget bandwidth honestly: ~10-15 kHz is enough for a clock, but it bounds your drawing rate. Plan the symmetric (push-pull) drive so the beam stays centered as it deflects.
• Video / Z amplifier (Vol 4). A class-A cathode-drive stage with a cascode speed-up to reach a few MHz, biased from the stiff mid-voltage rail, with the grid held at an adjustable brightness voltage. This is what blanks retrace cleanly; get it wrong and you see the connecting lines.
• Vector source (Vol 5). MCU+dual-DAC, FPGA, or the AGI, per § 10.4. Decide your X/Y/Z signal levels (the crt-driver’s 3.3 V analog is a sane convention) and the connector early — it is the contract between the digital and analog halves.
• Grounding, HV layout, and creepage (Vols 3, 11). Single-point ground between the analog and digital sections; keep the kilovolt nodes physically far from the CRT (the supply’s stray field pulls the beam) and far from low-voltage nodes (the blog’s 1 cm-minimum clearance rule); slot the board or use standoffs under HV tracks; specify HV-rated parts everywhere the kilovolt rail touches; plan the mu-metal/soft-steel CRT shield (Vols 6 and 11).

10.5.2 Trade-offs versus adopting the TubeTime set

Designing from scratch buys you exactly two things: the ability to modify the layout (which the crt-driver, lacking published source files, denies you), and a complete, first-principles understanding of every node. It costs you months. The crt-driver is a known-good design that other people have debugged for a decade; your board is not, until you debug it. Unless you have a concrete requirement the crt-driver cannot meet — an unusual tube outside its compatibility list, a form factor it does not fit, a feature like programmable intensity or a second gun — the rational move is to adopt the boards and design only the vector source, which is where the creative clock-design work actually lives anyway. Reserve full-custom analog for the case where you have outgrown the reference design and know exactly why.

10.6 Alternative open references

The crt-driver is the recommended foundation, but two other open references are worth knowing, both linked from the project’s Notes.txt.

10.6.1 The Sgitheach DIY route

The British Sgitheach scope-clock project (sgitheach.org.uk/scope2.html, with open-source code under #software) is a complete, independently developed scope clock — a different point in the design space from TubeTime’s. Where the crt-driver gives you a polished, general, three-board amplifier set and leaves the firmware to you, Sgitheach publishes a full integrated design including the vector-drawing firmware, with extensive write-ups of the design reasoning. It is the best second opinion on the analog electronics and a valuable firmware reference for Vol 5, and it is flagged as a primary open source back in Vol 1’s history. If you find the crt-driver’s missing layout files frustrating, Sgitheach’s openness is a useful counterweight.

10.6.2 The NYC Resistor vector-display approach

NYC Resistor’s “Vector Display Introduction” (nycresistor.com/2012/09/03/vector-display/) is the gentlest on-ramp of all, and it deliberately sidesteps the high-voltage problem: it drives the X/Y inputs of an existing analog oscilloscope in X-Y mode, using a Teensy 2.0 microcontroller and a pair of home-built R-2R ladder DACs (sixteen resistors per channel — seven of value R, nine of 2R; the write-up uses 500 Ω and 1 kΩ). Because the scope’s own deflection amplifiers and HV supply do the dangerous work, this is the safest possible way to learn vector drawing before committing to a tube and a kilovolt supply. It is not a standalone clock — it borrows your oscilloscope — but as a proof-of-concept for the vector source you will later feed into the crt-driver’s J3, it is genuinely useful, and its R-2R DAC is the same idea Vol 5 formalizes.

10.7 Honest effort and skill assessment

This is the hardest path in the series, and it is worth being blunt about who should take it. The adopt-the-boards route (build the three crt-driver PCBs, wire a tube, supply a vector source) sits between the owned-unit builds (Vols 7 and 8) and full custom: you are hand-soldering three boards of mostly fine-pitch SMD parts, hand-wiring a CRT socket from three cross-referenced documents, bringing up a lethal kilovolt supply, and writing or adapting vector firmware. It demands a real bench — an oscilloscope, a multimeter rated for kilovolts, an isolation transformer, a variac is nice — and the HV discipline of Vol 12 is non-negotiable. The full-custom route adds months of analog design and debug on top.

Where this path wins over Vols 7-9 is control and reach. The kit and finished-unit options (Vol 9) and the two owned builds (Vols 7 and 8) are all built around particular tubes and particular, few-hundred-volt supplies. The crt-driver gives you a stiff 1 kV (or 3 kV PDA) supply, true differential deflection amps, and a 6 MHz video amplifier — meaning you can drive an unusual or large tube, get a brighter and crisper trace, and even run vector animations or games, none of which the turnkey options reach. If your goal is the satisfaction of understanding and controlling the analog chain, or you are building around a tube nobody sells a kit for, this is the path. If your goal is a working clock on the shelf with the least risk and effort, it is not — go back to Vol 1’s decision tree and pick path 1, 2, 3, or 4. Choose this path because you want to build the electronics, not merely to own the clock.

10.8 References (Vol 10)

• E. Schlaepfer (TubeTime), “Electrostatic CRT Driver Design,” tubetime.us/?p=183 (31 May 2014; with the June 2014 “CRT Board BOM Updates” addendum). Held in 02-inputs/A - Open Source/TubeTime » Blog Archive » Electrostatic CRT Driver Design.pdf.
• crt-driver open-hardware repository (CC BY-SA 3.0), github.com/schlae/crt-driver. Local mirror: 02-inputs/A - Open Source/crt-driver-master/ — README.md, CompatibilityList.txt, ScopeWiring.pdf, BaseDiagrams.pdf, and the ScopePower / ScopeDefl / ScopeVideo design folders (Gerbers, CSV BOMs, schematic PDFs).
• OSH Park shared projects (order the bare PCBs):
  • 1 kV power supply — oshpark.com/shared_projects/PMsJvDGz
  • Deflection & video amplifier — oshpark.com/shared_projects/stZ49KK7
  • 60 V video bias supply — oshpark.com/shared_projects/F2lrTWvo
• J. Williams, “A Fourth Generation of LCD Backlight Technology,” Linear Technology Application Note 65 (the CCFL/Royer reference behind the 1 kV supply).
• E. Andrews, “The Arduino Graphics Interface, Parts 1 & 2,” Nuts & Volts, and the XYscope library. Held in 02-inputs/Arduino XY Graphics Interface/ (201802-Andrews.zip plus the article PDFs).
• Sgitheach scope-clock project and open-source code, sgitheach.org.uk/scope2.html (software at #software).
• NYC Resistor, “Vector Display Introduction,” nycresistor.com/2012/09/03/vector-display/. Held in 02-inputs/PDF Resources, Ideas, and Schematics/Vector Display Introduction » NYC Resistor.pdf.
• LabGuy’s World — wiring diagrams for the RCA 913 and 3JP7 and related electrostatic-CRT material (referenced from the TubeTime blog’s 23-Dec-2014 update).