This project began as a simple but good quality vacuum tube tester. But it soon morphed into a quest for a highly accurate instrument, such as the legendary instrument maker, Hewlett Packard (now Keysight), might have built...

Now, HP (Keysight) never made a tube tester and I’m not an alumnus of that hallowed company, but I love the way they designed their gear: It was built to last and always seems to outperform expectations. Thus, as the design process went on and questions of accuracy and stability goals came up, it was natural to try to do the best I reasonably could. But that’s a slippery slope and it took seven years to finish this project. It was a cliffhanger to the end but the result is pretty much everything I’ve always wanted in a tube tester! You can see the results produced by the unit in the on-going series of 7591 tube evaluations here.

Could You Build It?

All of the documentation and PCB design files for the Vacuum Tube Analyzer (VTA) are available for download at the end of this article. [It’s called “analyzer” to set it apart from tube testers which don’t use constant DC supplies. Unlike those, you can use it to plot DC and AC characteristics.] I had always intended it to be something that others could build. For that reason, all of the parts (except for a kludge) are through-hole types. I even have extra PCBs that I can make available. But truthfully, the answer to the question is that it would be difficult and expensive to re­pro­duce. Nevertheless, I’m putting all this out there in the hopes that parts of the system could be useful to folks or that some might find it to be a useful or interesting example. Hence, this will be an overview, rather than the detailed type of construction article you typically find elsewhere on this website. See “Advice for Builders” near the end of page 2.

A Simple Tube Analyzer

Simple to use, that is. The front panel layout at right shows the general organization. After setting the pin connections in gold at the bottom, you use the controls in red to set heater, grid, plate and screen volt­ages. Meter-1 in blue shows the voltage values for that. Turning on the Power and TEST switches in green starts the test. Results like currents, Gm and Gp are read from Meter-2 in magenta. Also, Ip is duplicated on Meter-1 for con­ve­nience. Test parameters can come straight from the original tube data books or from the given table.

Basic Specs

Measurements performed: DC Voltage and current for heater, grid, screen and plate. Transconductance (Gm), plate con­duct­ance (Gp = 1/Rp) Overall accuracy goal:  0.1% in primary ranges. 20,000-count meters. Gm and Gp AC test levels: Grid ±50mVpk, Plate ±1Vpk. Physical: Weight 23.1-lbs, Overall dimensions 15”W 6”H 18”D. Voltage Metering Burden: 10Mohms Current Metering Burdens: (None affect readings or regulation.) Plate, Screen 10.2ohms. Grid Hi 10.2ohms, Lo 1020ohms. Heater 0.03 ohms Supply voltage and current limits: Supply Voltage Current Spec Current Limit Heater 1.25 to 13V 3A < 8.4V,  2A > 8.4V 3.3A < 8.4V,  2.2A > 8.4V* Grid +3 to -100V -10mA, +3mA -20mA, +6mA Plate 0 to 400V 100mA 130mA Screen 0 to 400V 20mA 30mA *Heater current “limit” shown is a warning LED. Actual limiting is at about 5-6A. Measurement Ranges and Resolutions:  (Two, 4.5-digit meters provided.) Measurement Range Resolution Heater Voltage 20V 1mV Heater Current 20A 1mA Grid Voltage 200V 0.01V Grid Current 20mA 0.01uA Plate Voltage 2000V 0.1V Plate Current 200mA 0.01mA Screen Voltage 2000V 0.1V Screen Current 20mA 0.001mA Transconductance 40000umho 0.1umho Plate Conductance (1/Rp) 2000umho 0.01umho Plate Cond. in terms of Rp 500ohms min 1% at 1Mohm Notable Features Dual 4.5-digit meters make it easy to plot things like Gm vs Ip, Ip vs Vp, Ip vs Vg, Is vs Vp, Gp vs Vp and other common pairs. Tube voltages are precision regulated to high-stability voltage ref­er­ences (as are the Gm/Gp AC levels), so there’s no need to readjust voltages. Voltage-setting pots are 3 turns, including the grid-fine control, making settings precise yet efficient. Virtual ground at the plate eliminates Gm error from low Rp and improves Gp resolution by 3.2X over typical techniques. (justification) Current limiters with indicators on all four supplies for protection. Built-in 0.01% calibration standards at Gm=10000 and Gp=1000umho. Square-wave, AC test signals and synchronous detector improve ac­cu­ra­cy, reject hum and noise, and avoid AC detection flaws. Low AC signal levels (for Gm and Gp) reduce error from nonlinear tubes. Grid test level is just 100mVpp and plate is only 2Vpp. Separate drive and sense lines eliminate error from wiring voltage drops. Separate heater and cathode grounds prevent heater current from af­fecting other readings. Separate ground sense lines eliminate errors due to ground current. Closed-case, internal fan and baffle cools the system while keeping circuitry dust free. Uses the entire chassis as a radiator. Precision, switching, heater power supply minimizes heat while de­liv­er­ing up to 3.3A (5A briefly). Any pin can be switched to the external connection binding post for use with lab supplies and meters. Ferrite beads on all socket pins insure stability for high-Gm tubes. Additional auto-range feature for the Gm high range extends max trans­con­duct­ance reading to 40000umho. The VTA front panel was designed using “natural” sym­bols rather than the ones from the old days. Example: using Vp instead of Va for plate volt­age. Hence, we’re using the nat­u­ral symbols in this article, as defined below. Item Natural Classic Heater h f Cathode c k Grid g g1 or c1 Screen s g2 or c2 Plate p a or b Voltage V E Test results are mostly read on Meter2, so its function switch shows what’s available. The meter formats below indicate ranges and resolutions: Item Format Units Ih 19999. mA Ig lo 199.99 uA Ig hi 19.999 mA Ip 199.99 mA Is 19.999 mA Gm lo 1999.9 umho Gm hi 19999. umho Gm over* 19999. X10 umho Gp lo 199.99 umho Gp hi 1999.0 umho *Gm hi auto-ranges by 10X above 19999, valid up to 4000 x 10 (40000). The 100X step from Ig hi to Ig lo takes advantage of the 4.5-digit meter to provide 1% res­o­lution for a 1uA value.

Physical Layout

Construction is based on a 15 x 15 x 4.5-inch, sloped, aluminum chassis made by LMB-Heeger, which you can see a little better in the photo below right.

All electrical components and modules are mounted to the top cover and the bottom is mainly just a cover. This way, there are no connectors or dangling leads between the top and bottom. Below left is the inside of the unit. (Click for a larger pic). Labeled parts are listed below.

Main board with DC regulators, measurement system and dither generators. Internal fan mounted on baffle for controlling air circulation. Air flow is upward here, across the PCB. Since the baffle is sealed to the top and bottom of the chassis, air must pass through a triangular area near (K). There is no vent, so the whole aluminum chassis becomes a heatsink. Power Supply Board includes unregulated plate-screen, grid and heater supplies. Also includes un­reg­u­lated auxiliary supplies for plate, screen and heater regulators, plus unregulated supplies for each meter. Main transformer. Meter-1 includes stock PM-328 4.5-digit meter plus the visible add-on board with precision reference, split-supply regulators, high-stability gain adjustment and input protection. Meter-1 function switch with mount-on board for decimal point and units indicator logic. Auto-ranging addition for Meter-2 extends Gm limit to 40000umho. Meter-2 with similar qualities to Meter-1 (E). Meter-2 function switch with features of (F) plus a relay-controlled, precision, 10X divider. Control switches: (top to bottom) Power, TEST, Pentode/Triode, Gm/Gp Calibrate. Pin switches connect each pin to one of eight functions: NC, Ground, Heater, Heater-ground, Grid, Screen, Plate, External-connection. Voltage controls: (top to bottom) Screen, Plate, Grid-fine, Grid, Heater. Main heatsink. Power devices are mounted directly to the heatsink, through holes in the chassis.

Block Diagram

Overview

Please click the block diagram above for a high-resolution version. The basic job of the VTA is to provide plate, grid, screen and heater supplies and measurement facilities for all DC voltages and currents, plus Gm and Gp. (Rp = 1/Gp) The tube under test is inserted in one of the sockets represented in the small block at the top. Pin connections are set in the matrix of switches in the large block to the left. Lines from the switch matrix to the rest of the VTA represent the selected tube pins as plate, screen, grid, heater and ground (cathode/suppressor grid). Two, separate lines are switched to each tube pin: drive and sense. Supply current uses the drive line and the sense line is used for metering and remote sensing for the plate and heater supplies. This scheme virtually eliminates errors due to voltage drop in the VTA wiring. It’s as if you put a meter right at the tube socket.

Switching and Power Supplies
The lines from the switch matrix (except heater) pass through the TEST switch to disconnect the tube when the test is not in progress. Heater remains connected in Standby mode and Meter-1 looks at the supply voltages instead of the voltages at the tube pins. This supports warmup and power supply setup. We’ll discuss the Gm/Gp cal switch below. Along the bottom of the block diagram we see the plate, grid, screen and heater supplies, as well as the auxiliary ±15V supplies which are used internally in each of the precision tube supplies. (The heater supply develops its own, internal operating voltage.) Each tube supply is provided with current limiting and a front panel overload indicator to let you know when it cannot meet the current demand.

DC and Gm Measurements
Meter-1 can show any of the tube supply voltages, so it’s used for setting up voltages. It also can display plate current. Meter-2 is used to display results and can show any tube supply current, as well as Gm and Gp. When Gm is selected, a precision, 100mVpp, 1kHz signal is imposed on the grid voltage. The Gm/Gp Detector block above the plate supply measures the amount of AC current generated by the tube and thus reads Gm. There are two selectable ranges (2000umho, 20000umho) and also an auto-range feature which extends readings to 40000umho.

Measuring Gp instead of Rp
When Gp is selected, a precision, 2Vpp, 1kHz, dither signal is imposed on the plate voltage. Now the Detector in the plate circuit shows the AC current caused by the dither and this is read as Gp (plate conductance). Plate resistance, Rp, is 1/Gp. Why read Gp instead of Rp? It’s because Gp is what you get if you want to test with a fixed AC voltage. To read Rp directly, we would have to provide an AC current source into the plate and be able to read a wide range of AC voltage that results. This is more difficult and prone to error. By reading Gp, we can keep the plate swing at a very low ±1Vpp, yet accurately measure plate resistances from 500ohms to 1Mohm or higher.

0.01% Gm and Gp Calibrator
To help in achieving the overall accuracy goal of 0.1%, I decided to include a calibrator for Gm and Gp. While DC measurements can be checked easily enough with lab equipment, insuring AC accuracy with the measurement approach used, would be difficult without the calibrator. When the Cal switch is turned on, VTA circuits are switched away from the tube, to the calibration circuit. For Gm, it uses a precision opamp and a high-voltage MOSFET to create a precise value of Gm, which is set by a 0.01%, low-tempco resistor. The Gm value is 10000umho, half of full-scale. Checking it just now, after months of use, it shows 9997umho, showing just 0.03% drift. The Gp cal circuit is similar, except the input signal comes from the dither voltage on the plate. It provides a Gp reading of 1000.0umho, half of full-scale.

Photo Tour of the Development

Click the image for a captioned photo gallery spanning six years of development: the good, the bad and the ugly! Expand the window to full screen to enjoy the high-res pics. You can navigate with the keyboard: Home, , , End.

Coming up next: Three interesting schematic topics, complete docs and more!