The plate regulator was stuffed first and is being tested here.
A closeup of the plate regulator test showing the temporary heatsink for the pass-MOSFET at the back. A pot to set control voltage is at left. It seems remarkable that it's so compact, yet it's a 52W, class-A amplifier which swings 0 to 400V and has good 1kHz transient performance.
The plate regulator was rebuilt on a protoboard to investigate and fix problems found.
Next, the heater switching regulator was stuffed and tested. This is long before the large inductor and the two large caps at the back were moved to the bottom of the board. The caps made it difficult to attach the semiconductors to the heatsink in the final system. The large inductor was blocking the fan.
After stuffing the grid regulator, generator and the plate AC-current detector, the Gm measurement could finally be tested. Notice that 6-7 lab supplies were needed!
The completed Main Board being tested.
"The Grand Hookup," connecting the main transformer, power supply board and main board for the first time. The two pots stand in for the front-panel plate and grid voltage controls. The plate regulator pass MOSFET is mounted on the large heatsink and the grid MOSFET has an oddly-shaped piece of metal for a heatsink.
At this time, the solder-side mods were relatively modest, but despite good performance on the bench, storm clouds were brewing...
These sample mods include the 1/4-watt resistor on the left. Below it, is the surface mount kludge which forms the ground sense bus buffer (3 resistors and 2 caps). At right, a minimum-voltage pot for the heater supply is added, allowing it to cover 1.25V (for portable radios, etc.)
The Meter Mod board, mounted to the back of a panel meter. It improves accuracy by providing a high-stability voltage reference, precision gain adjustment, regulated supplies and input protection. Mods were also made on the panel meter board itself.
The Meter Mod board in its raised position for servicing. You can see the improved caps (brown and white) on the panel meter board. To the left of the main IC is the addition to make the clock oscillator crystal-controlled.
The top and bottom of the Meter2 Function Switch board. The diodes do the decimal point and units-indicator logic. In the upper right corner topside is the precision, relay-controlled, 10:1 divider, which provides high/low ranges for Gm and Gp measurements.
Two views of the Meter2 Function Switch. The board is mounted and wired and local switch wires are added. Wires to the board are arranged so it can be tilted up for servicing.
A lot of effort went into producing and affixing the panel art. At left, a print from my letter-size printer is tested for fit and appearance. At right, a print from a commercial, large-format printer is prepared by giving it a clear coating of "Preserve Your Memories II". A paper towel is taped to the can to prevent drips from the spray nozzle.
Affixing the panel artwork with spray adhesive: At left, you can barely see the edge of the back side of the artwork, which is taped to the bottom of the back panel and covered to avoid overspray. We’ll spray the back panel, remove the cover sheet and press the artwork onto the panel. At right, the back and top panels are affixed and we're about to spray the front panel.
The flaps have been pasted down at left and the finished piece is shown at right.
First, (at right) holes near the sides are made with a hand punch. It's easier, more accurate and leaves cleaner holes. Holes which can't be punched are drilled as seen at left. For accuracy, I used up to three drill bits for each hole. For example, a 3/8" hole would be drilled successively with 1/16, 3/16 and 3/8" bits. The yellow and blue protective tape is discussed in the next slide.
The chassis with all round holes drilled or punched. Only the starter holes for the large chassis punches are done at this stage. Yellow and blue masking tape was applied before machining to protect the artwork, which also serves as a drilling template. I used weak masking tape and weakened it further by sticking it to my jeans.
The drilling makes a mess of the artwork around the holes as at left. In the center, the ragged edges are first pushed into the center and then cut away. The result at right is reasonable but must be covered with a knob, bezel or washer for decent appearance.
Rectangular holes for the panel meters were cut slightly undersize with the jigsaw (after drilling starter holes). Then a metal file was used to produce a nice edge. The overlap of the panel meter bezel is only about 0.05" (as I recall), so it had to be quite clean.
The TO-220 devices at the rear of the PCB needed to be mounted directly to the main heatsink, so clearance holes had to be punched. To get those aligned accurately, the main board is temporarily mounted in place and the TO-220 devices are outlined on the rear panel.
At left, clearance holes for the TO-220 devices are punched and are seen in the top middle photo. At middle-bottom, the heatsink is attached and the TO-220 mounting holes are marked. At right, those are drilled and tapped for 4-40 screws.
As a result, the TO-220 devices conveniently attach to the heatsink and are well-aligned with the clearance holes.
In spite of attempts to weaken the masking tape, getting it off without lifting clear-coat, lettering or even a layer of paper is tricky. Moving slowly is essential. At the edges of holes, the direction of pull is critical. As shown here, the best direction to pull is toward the center of the hole.
As this picture of the finished panel shows, it all came out well in the end. One, tiny spot did need to be patched but that was done, almost invisibly.
The Electroswitch rotaries are good but hard to turn. We can see the detent mechanism in the cast aluminum housing at upper left. The ball bearing in the upper right pic slides in and out of the channel, causing the detents but the bottom part of the channel is a bit too narrow, causing annoying stiction. The fix is to widen the slot with a rotary tool as shown in the lower pics.
Another reason for the stiff switches is that the spring which tensions the ball bearing is too stiff. I wound new ones, as seen here. The thinner wire is from stock springs. A large-size X-Acto blade holder provided the proper form (0.437" diam).
Production mods of twelve pin switches and two meter function switches, were done in batches of four. Items shown include long screws to hold the switches together, cleaned-up switch shafts and housings and newly-made springs. Near top center are four, brownish springs used for wire. Lube Gel at bottom left replaced the messy black grease cleaned out of the switches.
The new spring is put in place and the shaft is inserted. At right, you can see the winged washer which slides in the slots and contains the ball bearings.
To get the skirts of the knobs within 0.1" of the front panel, switch shafts were cut to length by rotating the shaft in a drill and using a stationary hack saw, providing a clean cut. The flat section of the shaft had to be extended as shown in the middle and upper right pics. I also had to bevel the shaft hole on the bottom side of the knobs, as seen at lower right.
The major system pieces, ready to be assembled into the chassis. From left to right, you can see the Power Supply board, Main board, Main heatsink, main transformer, Meter2 and its function switch and Meter1 with its function switch. Back, by the heatsink, are spare panel meters.
The tube socket field has been wired and front panel components are mounted, ready for chassis wiring.
All wiring except the main transformer and AC power has been completed. It was last because handling the chassis is much easier without that heavy weight. This was before the switcher caps and toroid were moved to the bottom. The trunk of the main wiring harness still has the temporary orange wires, binding it together.
The main board wiring was arranged so the board can be propped vertically for service. In fact, the board can be laid back against the switches.
Major meltdown in the calibrator circuit, caused by an obscure MOSFET foible: High slew rate on the drain can turn on its parasitic bipolar transistor, causing catastrophic failure. Connecting power to the drain with a switch required a series resistor mod. The calibrator is sensitive, so all parts in the area had to be removed and the board scrubbed clean.
When servicing the VTA, it's sometimes helpful to turn off the switching heater supply. Lifting a control pin of IC7 (the switching regulator chip) and adding a jumper provided the disable feature.
When the Main board is in the service position, a temporary heatsink must be provided for the outboard semiconductors. The L-shaped aluminum heatsink seen here is sufficient for short testing periods, with a 6-inch fan blowing on it. Two MOSFETs were moved from the board to the main heatsink. When the board is in service position, the jumper cables seen here are required.
Development continued for a long time and more kludges were needed.
Two switcher caps and the main switcher inductor were moved to the bottom of the board for easier access to the power devices along the back edge and to unblock the fan.
A major plate regulator mod removed a lot of high-side regulator components, including the AC plate voltage generator for Gp tests. Now, the plate voltage amp itself handles the AC, requiring very good settling specs. Here, the new plate regulator is protoboarded for testing. Precision settling tests required special circuitry, occupying the left two inches of the protoboard.
Part of the drastic mod of the plate regulator included a new, discrete-transistor current limiter and dual frequency band feedback components. Those were built on this perfboard and mounted to the bottom of the Main board.
Here, is the final cleanup of the wiring harness. You can also see the two MOSFETs which were moved to the main heatsink. Their blue wires go to connectors at their original board positions.
Tubes with high Gm caused 200MHz oscillations. Tests pointed to the socket field wiring. The fix was to put a ferrite bead at every socket pin, requiring all socket field wiring to be removed. Rewiring used thinner, 600V wire, allowing more direct routing, better beads and much less capacitance. It nailed the oscillation problem, as seen in the next slide.
The problem with high-Gm oscillation also brought up the fact that the original 20000umho limit wasn't quite sufficient. An extended Gm range was added, which kicks-in automatically for Gm greater than 20000umho. It's good to at least 40000umho. Here, a 7722 pentode is showing its expected Gm of about 26000. The extended range is indicated by the added X10 LED.
This is the final configuration of the VTA, with one exception you will see in the next slide. Here, the auto-range circuit board is seen near center, attached to the bottom edge of the Meter2 Mod board. Two chips and a blue cap are visible on it.
The last mod was to swap the plate regulator MOSFET for a beefier one, rated for continuous current and higher voltage. Compare the new TO-247 device with the TO-220 above it. Leads won't fit the TO-220 pads so they go under the board. This mod was due to another MOSFET oddity: Switching MOSFETs can fail with continuous currents at high voltage, even within other ratings.
The finished Vacuum Tube Analyzer