Deflection & Video (Z) Amplifiers

Between the microcontroller’s digital-to-analog converters and the glowing spot on the phosphor sits the part of a scope clock that does the heavy lifting in the analog domain: the deflection and video amplifiers. The vector firmware (Vol 5) hands these amplifiers a trio of small, friendly, single-supply signals — an X value, a Y value, and a Z (brightness/blanking) value, each a few volts at most — and the amplifiers must turn those into the much larger, push-pull, bias-centred voltages the deflection plates demand, and into a fast cathode or grid swing that can switch the beam on and off cleanly between strokes. This is where a low-voltage logic signal becomes high-voltage analog motion, and it is where most of the subjective quality of the display — sharpness of corners, evenness of brightness, freedom from flicker, absence of retrace lines — is won or lost. This volume covers the deflection-drive problem in general, the video/Z-axis amplifier, the two owned and open-source implementations (the OSC4.4 transistor chain and the TubeTime fully differential design), and the slew-rate and settling limits that ultimately cap how fast and cleanly a face can be drawn.

4.1 The deflection-drive problem

An electrostatic CRT steers its beam by the electric field between two parallel deflection plates (Vol 2). The deflection sensitivity of a typical small tube is on the order of tens of volts per centimetre of deflection at the screen — and the spec is given differentially, as the voltage between the two plates of a pair. To swing the spot from one edge of a 3-inch screen to the other you therefore need a plate-to-plate difference of perhaps a hundred to a few hundred volts, and you need it on both of the X plates and both of the Y plates. The source signal, by contrast, is a DAC output in the 0–3.3 V or 0–5 V range. The amplifier’s job is to bridge that gap, and four constraints shape every design:

• Voltage gain and high-voltage swing. A small input range must become a swing of tens to a couple of hundred volts. The output devices must survive that swing, so the deflection amplifier is necessarily a high-voltage circuit even though its input is logic-level.
• A bias centre, not ground. The deflection plates do not sit near ground; they sit at a high common-mode voltage — often a hundred or more volts positive — because the beam is accelerated through that region and the plates’ average potential sets the geometry. The amplifier must therefore level-shift: take a signal centred on ~1.65 V and re-centre its output on the tube’s deflection-region bias.
• Push-pull (antiphase) on each pair. Both plates of a pair are driven, one rising while the other falls about the common-mode point. This is covered in § 4.1.3 and matters more than it first appears.
• Centring and size. Tube-to-tube variation in deflection sensitivity and in mechanical centring means the amplifier needs adjustable gain (size/width-height) and adjustable offset (centring) on each axis.

4.1.1 Differential vs single-ended drive

The simplest conceivable scheme drives one plate of each pair and grounds (or fixes) the other. It works, and some minimal designs do exactly this, but it has two penalties. First, the full deflection swing must appear on a single plate, so that one amplifier needs the entire voltage range; driving both plates antiphase halves the swing each amplifier has to produce for the same deflection. Second — and more important for picture quality — a single-ended drive moves the average potential of the plate pair as the beam deflects, which shifts the local accelerating field and produces geometric distortion and a small defocus that varies across the screen. Driving the pair differentially, so that one plate goes up by exactly as much as the other goes down, keeps the average plate potential constant and the beam optics stable wherever the spot lands.

4.1.2 Level-shifting: getting from 1.65 V to the bias point

The DAC sits on the low-voltage analog rail and swings about its mid-scale (Vol 5). The plates sit near a high positive bias derived from the main HV string (Vol 3). The two ways the owned and open-source designs bridge that are worth contrasting:

Approach	How it level-shifts	Used by	Trade-off
Common-emitter gain + HV load	A low-voltage input stage drives a high-voltage transistor whose collector load returns to the HV bias; the collector swings about that bias	OSC4.4 transistor chain	Few parts, but gain and centre interact; per-tube resistor selection
Differential pair + HV cascode	A matched dual transistor sets gain/offset; a cascode of HV transistors level-shifts the swing up to ~1 kV across a pair of HV load resistors	TubeTime crt-driver	More parts, better symmetry and bandwidth, true differential output

In both, the centre (bias) voltage and the gain are set by resistor values that change with the tube, because deflection sensitivity is a property of the specific CRT. That per-tube dependence is why the volumes keep returning to component-value tables (and why Vol 6 tabulates them by tube).

4.1.3 Why both plates are driven antiphase

It is worth stating plainly because it explains several later design choices. If only one plate moves, the beam does deflect — the field between the plates still changes — but the beam also passes through a region whose average potential is moving, so it is accelerated or retarded slightly differently at different deflections. The visible result is a spot that defocuses and that bends the geometry of straight lines (pincushion/keystone-like effects), worst at the screen edges. Antiphase (push-pull) drive holds the mean plate voltage fixed: the deflecting field changes while the average potential the beam sees does not. The cost is that every axis needs two output drivers of opposite polarity — which is exactly the “complementary” or “fully differential” structure both designs in this volume adopt.

   SINGLE-ENDED (avoid)              DIFFERENTIAL / PUSH-PULL (preferred)

   +Vbias ──[load]──● X1             +Vbias ──[load]──● X1 ↑ rising
                    |                                 |
        beam ······>|  (mean moves)      beam ······>|   (mean fixed)
                    |                                 |
   X2 ── fixed ─────●                +Vbias ──[load]──● X2 ↓ falling

4.2 The OSC4.4 deflection chain (owned)

The OSC4.4 is the owner’s through-hole build, and its assembly instructions document the deflection chain part by part — which makes it the most concrete worked example in the series. The analog signal path begins at a TLC7528 dual 8-bit DAC (U3, a 20-pin part) that produces the X and Y voltages, with a pair of 4132 dual op-amps (U4 and U5) and a brace of transistors forming the gain and level-shift stages that follow. The blanking/shift behaviour is handled by a small 12F629 (U6) that the instructions call “the shifter,” and the whole face is generated by an 18F26K20 PIC (U1). The deflection output devices are a chain of high-voltage transistors.

4.2.1 The wave-shaper (Q7/Q8)

Before the deflection chain proper, the OSC4.4 has a small wave-shaper built from two 2N3904 NPN transistors, Q7 and Q8, in Part 2 of the build alongside the relay control. Q7 sits with its emitter toward the Q7 silk and Q8 with its collector toward the Q8 silk — the assembly notes the orientation explicitly because the two are wired as a complementary pair rather than two identical stages. The wave-shaper conditions the timebase/blanking signal feeding the rest of the analog section; it is associated on the board with R38 (2.2 K), R33/R27 (10 K), R29 (1 K) and the brightness-related R26 (470 Ω, or 10 K “for less brightness,” per the instructions). This is the OSC4.4’s video/Z front end — the stage that ultimately determines whether the beam is on or off — and it is the part the builder tunes for brightness via P3 (the 50 K/47 K top-adjust pot) during bring-up.

4.2.2 The deflection transistor chain Q1–Q4 and the reversed Q11

The deflection drive itself is a chain of MPSA44 / KSP44 high-voltage NPN transistors — Q1, Q2, Q3, Q4, and Q10 — these are 300–500 V-class small-signal devices, the natural choice for swinging a deflection plate about a high bias. The one transistor that breaks the pattern is Q11, an MPSA92 / KSP92 high-voltage PNP. The assembly instructions flag it twice: it is “the only transistor on the board that the flat side faces opposite of the rest (the collector is closer to the IC’s).” That reversal is not a quirk of layout — it is the complementary device that lets the chain push and pull, i.e. drive a plate in both directions about the bias point. An all-NPN chain can only pull a plate one way actively and relies on a resistor to return it; the PNP gives the complementary half of a push-pull pair.

  OSC4.4 deflection chain (per axis, simplified from the assembly order)

   TLC7528 DAC ──► 4132 op-amp ──► Q1/Q2 (MPSA44 NPN) ──┬──► X/Y plate
                                                         │
                          Q11 (MPSA92 PNP, reversed) ────┘  push-pull half
                                   gain set by R99/R4 ; centre by P6/P7
                                   size by P4/P5

4.2.3 Gain-setting resistors that change per CRT

The deflection gain in the OSC4.4 is set by two resistor pairs whose values are explicitly selected for the tube — this is the per-tube dependence of § 4.1.2 made concrete. The assembly instructions and the two documented CRT pinouts give:

Resistor pair	DG7-32	6Lo1i	Role
R99 / R4	330 K (orange-orange-yellow)	220 K (red-red-yellow)	Sets deflection gain / input scaling
R6 / R8	220 K	180 K (grey-band)	Sets the complementary load / swing

(The instructions also list a 2BP1 option for R99/R4 at 220 K, consistent with the larger, less-sensitive tube needing more drive — a useful sanity check when adapting to a third tube; see Vol 6 for the per-tube table.) The principle is simple: a less sensitive tube, or one run at higher accelerating voltage, needs a bigger plate swing for the same screen deflection, so the gain resistors come down to raise the gain. Do not guess these by ear — they interact with centring, and the correct values for a documented tube are the starting point.

4.2.4 Centring (P6/P7) and size (P4/P5)

Two kinds of adjustment finish the deflection chain. P6 and P7 are 50 K “side-adjust” pots — P6 for X centre, P7 for Y centre — that inject an offset into each axis to place the undeflected spot at the geometric centre of the screen, compensating for tube and amplifier asymmetry. The assembly’s bring-up procedure deliberately starts them at the extremes (P7 fully CCW, P6 fully CW) so the spot is findable, then moves both to mid-travel once the focus and a centred blob are confirmed. P4 and P5 are 50 K top-adjust pots that set the size (amplitude/gain trim) of the X and Y deflection respectively — “Adjust P4 & P5 to the desired size, then readjust P1 & P2 for the best possible focus,” as the final bring-up step puts it. Note the ordering: size and centring interact with focus (P1) and astigmatism-like sharpness (P2), so the calibration is iterative. Vol 12 treats the full centring/size/focus loop as a procedure; here the point is only that the deflection amplifier exposes both a gain trim and an offset trim per axis, exactly as § 4.1 predicted.

  OSC4.4 deflection adjustments (per the assembly bring-up)

   P1  focus            (1 M)   ── round, sharp spot
   P2  anode/sharpness  (1 M)   ── crispness
   P3  brightness       (50K)   ── beam current / Z level
   P4  X size           (50K)   ── horizontal amplitude
   P5  Y size           (50K)   ── vertical amplitude
   P6  X centre         (50K)   ── horizontal offset  (start fully CW)
   P7  Y centre         (50K)   ── vertical offset    (start fully CCW)

4.2.5 Relay control and the bring-up sequence

The OSC4.4 sequences its HV through a relay (with an optional second relay for battery backup) controlled from the wave-shaper/low-voltage section. During bring-up the builder tests the relay by momentarily jumpering the V1 node to the +5 V side of C44 — “the relay will click and the neon bulb should light.” Functionally this lets the controller gate the high voltage and the beam, and it is part of the default-off discipline: nothing dangerous is live on the plates until the supply has come up and the controller has asserted the rails. The detailed power-sequencing and the 300 V supply that feeds all of this are Vol 3’s territory; the deflection chain simply consumes the ~300 V string and the +5 V analog rail.

4.3 The TubeTime crt-driver deflection + video board

The open-source TubeTime crt-driver board set takes the same problem and solves it with a more elaborate, more symmetric topology aimed at higher performance and a wider range of tubes (the compatibility list spans most 2-, 3-, and some 5-inch electrostatic CRTs). Where the OSC4.4 economises, the crt-driver doubles down on differential structure. Its deflection- and-video board carries two fully differential high-voltage amplifiers — one for X, one for Y — plus the high-speed video amplifier, all driven from 3.3 V analog inputs.

4.3.1 The differential deflection amplifier core

Each axis is built around a matched dual NPN transistor (the schematic uses a DMMT3904 dual) forming the differential pair, with a bias network that sets gain and offset — these are the width/height and position controls. The pair’s outputs feed a cascode of two high-voltage NPN transistors that, as the design notes put it, “increases the bandwidth slightly and level shifts the signal up to about 1 kV.” The load for the circuit is a pair of 220 K high-voltage resistors — and the documentation is emphatic that these must be parts rated for high-voltage use, not ordinary carbon-composition resistors, because a 1 kV swing across a physically small resistor will arc or drift in a part not specified for it. The result is a true differential output: both plates of the pair are driven antiphase about the bias, satisfying § 4.1.3 directly.

  TubeTime crt-driver deflection amplifier (one axis)

   3.3V X in ──► [ matched dual NPN diff pair ]
                   │            │
            gain/offset bias (width + position)
                   │            │
            HV NPN cascode   HV NPN cascode      ← level-shift to ~1 kV
                   │            │
        +HV ─[220K HV]─●     ●─[220K HV]─ +HV
                       │     │
                    plate X1  plate X2   (antiphase about bias)

4.3.2 Bandwidth: why the deflection amps are slow on purpose

The crt-driver’s deflection amplifiers have a bandwidth of only about 10–15 kHz. That is not a shortcoming — it is a deliberate match to the job. Deflection moves the position of the spot, and position changes at the rate the firmware steps from one vector endpoint to the next (a few kHz of distinct points for a clock face). A wider deflection bandwidth would only let more high-frequency noise and ringing onto the plates and would make the spot jittery. The design even notes you could feed ramps into X and Y at this bandwidth and generate an NTSC sweep if you wanted. The contrast with the video amplifier — which must be hundreds of times faster — is the whole reason X/Y and Z are separate amplifiers.

4.3.3 Astigmatism, focus, and the bias context

Two trims on the crt-driver are worth naming because they interact with the deflection. The astigmatism pot changes the final-accelerator voltage relative to the average deflection- plate voltage; you adjust it when the spot looks oval rather than round, and it is the analog of the OSC4.4’s sharpness trim. The focus pot sets a bias voltage for the focus electrode, adjustable over a range covering most 2-, 3-, and some 5-inch tubes. Both are bias-supply adjustments rather than signal-path adjustments, but they live on the deflection board because the correct astigmatism setting depends on where the average plate voltage sits — which the deflection amplifier sets. The HV string that feeds all of this comes from the crt-driver’s 1 kV (Royer-oscillator) supply, covered in Vol 3.

4.4 The video / Z-axis amplifier

If the deflection amplifiers say where the spot goes, the video amplifier says whether the spot is on — and how bright. It is the Z axis, and it is a fundamentally different animal because it must be fast.

4.4.1 Cathode drive vs grid drive

There are two ways to modulate beam current. The textbook approach grounds the cathode and drives the control grid negative to cut the beam off; but, as the TubeTime design notes, “the circuit design ends up being tricky” because the grid must swing to a substantial negative voltage and the drive sits at an awkward potential. The easier approach — used by the crt-driver and, in spirit, by the OSC4.4 — is to hold the grid at a fixed (adjustable brightness) voltage and drive the cathode. Pulling the cathode negative relative to the grid turns the beam on; letting it rise turns the beam off. The end result is the same beam modulation, with one caveat the design calls out: there is “a slight impact on the focus electrode,” because the cathode potential is part of the gun’s optics. In practice that is a minor, correctable interaction.

4.4.2 Bandwidth: ~6 MHz and why it matters

The crt-driver’s video amplifier is a “pretty simple class-A design with a cascode stage to increase the speed,” and its bandwidth is about 6 MHz — enough, the design notes, to “easily handle NTSC video.” Six megahertz is roughly 400 times the deflection bandwidth, and the reason is the edge. Blanking is a step: the beam must go from fully off to fully on (or back) in the gap between two strokes, ideally in well under a microsecond, or the transition itself will paint a visible smear. A 6 MHz amplifier can switch the beam in roughly 60 ns (rise time ≈ 0.35 / bandwidth), fast enough that an unblank edge is invisible at normal draw rates. Anything much slower and you would see the beam fade on at the start of each stroke and trail off at the end — strokes with soft, blooming ends instead of crisp ones, and faint retrace lines where blanking lagged the deflection.

  Deflection vs video bandwidth — why two amplifiers

   Quantity        Deflection (X/Y)        Video (Z)
   ----------------------------------------------------------
   Controls        spot POSITION           beam ON/OFF + brightness
   Bandwidth       ~10–15 kHz (crt-driver) ~6 MHz (crt-driver)
   Drive           push-pull HV plates     cathode (or grid)
   Edge speed      vector step rate        < ~100 ns blank/unblank
   Failure mode    jitter, ringing         smeared/soft strokes, retrace

4.4.3 The stiff 60 V bias supply and beam-OFF-by-default

To get 6 MHz the video amplifier runs a fairly high drive current, and that current must come from a stiff +60 V bias supply — stiff meaning low impedance, so the supply voltage does not sag when the amplifier slams the cathode during a fast edge. The crt-driver provides this from a dedicated small board (12 V in → 60 V out); the design explicitly notes the 60 V “needs to be rather stiff since the drive current is fairly high (required to get 6 MHz bandwidth).” A floppy bias supply here shows up as brightness that droops during long bright strokes or dims when many segments are lit at once.

Two safety-relevant behaviours fall out of the video design. First, the crt-driver’s default state is beam OFF — you must either tie the video-in pin to 3.3 V or advance the brightness pot to see anything, which means a freshly powered, un-commanded board does not park a bright stationary dot that could burn the phosphor. Second, brightness is a bias (the grid/cathode operating point), separate from the modulation (the video signal), so the firmware can blank and unblank at full video speed while the human-set brightness stays put. That separation is why brightness control lives on the video amplifier’s bias network and not in the firmware’s fast path.

4.4.4 Why video bandwidth limits draw quality

The video amplifier’s bandwidth is the hidden governor on how fast the firmware can draw. Even if the deflection amplifiers could whip the spot to a new endpoint instantly, the brightness of each stroke is the product of beam current and dwell time, and the video amp’s finite edge speed means very short strokes never reach full brightness before the next blank arrives. Draw too fast and short segments (the serifs of a digit, the tick marks of a dial) come out dim or missing; draw slow enough to brighten them and the whole face flickers. The clean compromise lives inside the 6 MHz envelope — fast enough that edges are invisible, slow enough at the vector level that every segment gets its dwell. This is the trade-off § 4.5 formalises.

4.5 Slew rate, settling, and the brightness/flicker trade-off

The two amplifier families impose two different speed limits, and a good face lives inside both.

4.5.1 Slew rate and plate capacitance

A deflection plate, its wiring, and the amplifier’s output node form a capacitance of tens of picofarads. To move the spot, the amplifier must charge that capacitance through its output impedance; the rate at which it can do so is the slew rate, S = I / C, where I is the current the output stage can deliver and C is the load capacitance. With tens of picofarads and a few milliamps of available output current, a deflection amplifier slews at perhaps tens of volts per microsecond — which is why a full-screen jump takes a few microseconds, not nanoseconds. After the slew comes settling: the output rings and then asymptotes to its final value, and the spot is not truly at the endpoint until it has settled. Draw the next stroke before settling completes and you get a curved or overshooting line where a straight one was intended. The practical levers are: keep plate-wire capacitance low (short, thin leads, no shielded cable on the plate runs), and do not demand more slew than the output current allows.

4.5.2 The brightness vs draw-rate vs flicker triangle

Three desires pull against each other:

Want	Push it by	Costs you
Brightness	longer dwell per stroke, more beam current	slower total frame → flicker; phosphor wear if static
Fast draw	move the spot quickly between endpoints	less dwell → dimmer strokes; needs more slew rate
No flicker	refresh the whole face often (≳ 40–50 Hz)	less time per stroke → dimmer; demands fast amps

The beam must dwell at and along each stroke long enough to be visible, yet the firmware must refresh the entire face often enough — comfortably above the flicker-fusion threshold, so roughly 40–50 Hz or faster — that the eye sees a steady image. The total time budget per frame is therefore divided among all the lit segments, and a complex face (many digits, a seconds sweep, text) has less time per segment than a sparse one. The designer’s knobs are: reduce the number of points, prioritise dwell where brightness matters, raise beam current (brightness) to shorten the needed dwell, and — at the hardware level — provide enough slew rate and a stiff enough video bias that the amplifiers are never the bottleneck. When a face flickers, the fix is usually fewer/optimised vectors or more brightness, not faster amplifiers; when strokes are uneven or smeared, the fix is usually in the amplifiers (slew, settling, video bias stiffness). Vol 5 covers the firmware side of this budget; Vol 12 covers tuning it on the bench.

  Per-frame time budget (illustrative structure, not measured values)

   frame period (≈ 1 / refresh, e.g. 1/50 s = 20 ms)
     ├─ stroke 1: slew + settle + DWELL  ← brightness ∝ dwell × beam current
     ├─ stroke 2: slew + settle + DWELL
     ├─ ...
     └─ blanked retrace between strokes (video amp OFF, must switch fast)
   Too many strokes → dwell shrinks → dim OR refresh drops → flicker

4.6 Comparing the two implementations

The OSC4.4 and the TubeTime crt-driver bracket the design space, and seeing them side by side clarifies the engineering choices.

Aspect	OSC4.4 (owned)	TubeTime crt-driver
Deflection topology	NPN chain (Q1–Q4, Q10) + reversed PNP Q11 push-pull	Two fully differential amps (matched dual NPN + HV cascode)
HV string	~300 V	~800–1200 V (Royer supply, Vol 3)
Deflection gain set by	R99/R4, R6/R8 — per-tube values	width/position bias network — per-tube trim
Centring / size	P6/P7 (centre), P4/P5 (size)	position / width-height pots
Video / Z drive	wave-shaper Q7/Q8 (2N3904) → cathode	class-A cascode video amp → cathode
Video bandwidth	modest (clock strokes)	~6 MHz (NTSC-capable)
Video bias	from main LV/HV string	dedicated stiff +60 V board
Beam default	gated via relay; off until controller asserts	OFF by default (tie video-in high to see anything)
Best for	a documented clock on a documented tube	wide tube range, arcade/vector graphics, headroom

Both are legitimate. The OSC4.4 is a clock optimised for two known tubes with the minimum parts to do the job; the crt-driver is a general-purpose vector engine that happens to make an excellent clock. If you are building the owned OSC4.4, build to its documented resistor values for your tube (Vol 8). If you are rolling your own or chasing graphics performance, the crt-driver’s differential architecture is the better starting point (Vol 10).

4.7 Bench notes and failure modes

A short field guide, all of it diagnosable with the scope the audience already owns. Probe the plates only through a properly rated high-voltage probe — these nodes sit at hundreds of volts (OSC4.4) to ~1 kV (crt-driver), and Vol 12’s HV discipline is in force.

• Spot won’t centre to the middle. Centring pots (P6/P7) at the wrong end, or asymmetric deflection-output offsets. Start P6/P7 per the assembly (P6 CW, P7 CCW), find the blob, then bring both to mid-travel.
• One axis deflects more than the other / image squashed. Size-pot mismatch (P4 vs P5) or a wrong gain resistor on one axis (check R99/R4 vs R6/R8 against the tube).
• Geometry bows or defocuses toward the edges. Symptom of effectively single-ended drive — a dead or mis-oriented complementary device (on the OSC4.4, suspect Q11’s orientation, which the assembly flags as the one reversed transistor).
• Strokes have soft, blooming ends; faint retrace lines visible. Video/Z amplifier too slow or its bias sagging — check the 60 V bias stiffness on the crt-driver, or the wave-shaper and R26 brightness resistor on the OSC4.4.
• Brightness droops on long or busy frames. Video bias supply not stiff enough under load, or the frame is over-budget (too many vectors) so dwell is starved.
• Whole face flickers. Refresh below flicker fusion — reduce/optimise vectors or raise brightness so each segment needs less dwell (firmware side, Vol 5).
• Nothing on screen at all, board otherwise alive. On the crt-driver this is the default (beam OFF) — tie video-in to 3.3 V or advance brightness. On the OSC4.4, confirm the relay has clicked in and the HV is present.

4.8 References (Vol 4)

• OSC4.4 — Assembly Instructions for OSC4.4, Parts 2–3 (wave-shaper Q7/Q8, deflection chain Q1–Q4/Q10, reversed PNP Q11, gain resistors R99/R4 and R6/R8, centring pots P6/P7, size pots P4/P5) and Part 4 CRT-socket wiring with the DG7-32 and 6Lo1i pinouts and per-tube resistor values. Held in 02-inputs/OSC4_4 (I have this)/AssemblyInstructionsForOSC4.4.txt.
• OSC4.4 — Parts List (TLC7528 dual DAC, 4132 dual op-amps, MPSA44/KSP44 NPN, MPSA92/KSP92 PNP, 2N3904). Held in 02-inputs/OSC4_4 (I have this)/PartsListForOSC4.4.txt.
• E. Schlaepfer (TubeTime), “Electrostatic CRT Driver Design,” tubetime.us/?p=183 — differential deflection amplifier (matched dual NPN + HV cascode, 220 K HV loads, ~10–15 kHz), the ~6 MHz class-A video amplifier driving the cathode, grid-vs-cathode drive, astigmatism/focus, and the stiff +60 V video bias. Held in 02-inputs/A - Open Source/.
• TubeTime crt-driver open-hardware repository — README and board set (ScopeDefl, ScopeVideo, ScopeVideoOnly, ScopePower) and the CRT CompatibilityList.txt. Held in 02-inputs/A - Open Source/crt-driver-master/.
• Dutchtronix AVR Scope Clock — schematics and operating manual, for the alternate ATmega-based deflection/Z architecture referenced in comparison. Held in 02-inputs/A - Dutchtronix AVR Scope Clock (I have this)/.
• Cross-references: Vol 3 (the 300 V / 1 kV HV strings and the 60 V bias rail), Vol 5 (the DAC source and the vector firmware’s time budget), Vol 6 (per-tube deflection-sensitivity and resistor values), Vol 12 (centring/size/focus calibration and HV bench discipline).