Meet Andrew’s blast from the past
Back to the future
Sometimes the best way to take a step forward is to take a step back in time. So Andrew Smith designed a fully-functional toy oscilloscope, made out of parts he found in his junk box, such as the EF91, EF80 and EF184 valves. Using a DC-DC converter to power the old (but still working) 7cm CRT he discovered in his loft, Andrew housed it in the same wooden box as the rest of the circuitry. The whole system runs from a single regulated 12.6V DC supply, which can be derived from a “wall-wart” PSU. Doc Brown would be proud.
This project is not intended to be a truly functional oscilloscope, although it does actually work. Rather it is intended to be a decorative conversation piece.
I discovered that I had a really cute little 7cm CRT in the loft, together with several other valves of different types, and it seemed a good idea to make something decorative with them. For me, a lot of the charm of this piece resides in the unconventional layout and design, using a polished wooden box instead of the more usual metal chassis.
I have made several other projects in this genre which can be found here. I use pieces of mahogany strip left over from a hardwood floor; scrap material of this kind is readily available from local flooring companies, and with simple hand tools very satisfying results can be gained. There is an intended reference to the form of construction used for electrical equipment in the early decades of the last century, but in my hands the genre has gained a life of its own, and has become completely anachronistic.
Nowadays it is difficult to find mains transformers for glow-bug projects, and the 800v EHT supply required for the CRT could be particularly difficult. To overcome this problem, and to achieve a neat and compact design I have used a dc inverter technique, which turned out to be relatively simple. The converter is housed in the same wooden box as the rest of the circuitry, and the whole system runs from a single regulated 12.6v dc supply, which can be derived from a ‘wall-wart’ PSU.
The circuit for the CRO is shown below.
It uses three different small-signal rf pentodes – an EF91 in the timebase circuit, an EF184 in the Y amplifier, and an EF80 in the X amplifier.
The choice of tubes is somewhat arbitrary. The EF91 would be very suitable in all three functions, since among small-signal pentodes, it is unusual in having suitable characteristics for providing a relatively large anode voltage swing with little distortion. But I do not find the EF91 visually attractive, so I opted for the other tubes in the main functions, relegating the ’91 to the back of the chassis, where its smaller size fits in nicely.
The EF184 has high transconductance, so is used in the Y amplifier where large voltage gain is required, while the EF80 is used with cathode-degeneration feedback in the X amplifier which requires much less gain, but a larger voltage output with good linearity.
These two valves run from an HT supply of 600v in order to achieve linear amplification over a wide output voltage range (400v pk-pk for the X amplifier, 220vpk-pk for the Y amplifier), but they operate within the permitted maximum anode voltage for both tubes of 550V. The large output voltage swings are a consequence of the low deflection plate sensitivity of the DG7-6 cathode ray tube. Variable cathode resistors are used in both amplifiers so that the operating points of the tubes can be set for best large-signal linearity.
The timebase circuit is a phantastron. This is a classic pentode circuit which has great flexibility; it can operate either as an astable or a monostable and is able to create a linear voltage ramp by Miller feedback involving the capacitor joined between anode and grid 1.
All three grids of the pentode, including the suppressor grid, are used as signal-carrying control electrodes, which is most unusual. In this realisation the circuit operates as a free-running astable, with a synchronising signal applied to the anode from the Y amplifier. In practice the circuit locks on to a Y input signal which is an integral multiple or sub-multiple of the free-running frequency of the phantastron, so it acts like a phase-locked oscillator – one of the classic applications of the circuit. The locking range is quite wide, so adjusting the ‘Speed’ control to achieve a firm lock is easy and reliable.
The timebase output, taken from the anode of the EF91, is about 20vpk-pk, so it is attenuated with a preset ‘X Amplitude’ potentiometer before being applied to the grid of the X amplifier tube. The other preset control, from grid 3 to ground is adjusted for best linearity, which can be excellent. Traces of the timebase waveform direct from the phantastron (upper), and after amplification to +/-200v (lower) are shown, the original version being inverted for comparison.
C9 and R17 have been added to suppress the CRT beam during the timebase flyback period.
The circuit of the inverter PSU is shown below.
This converter was originally designed for 175v output for a different project, but was adapted to its present use by adding more windings to the transformer.
When I started on this project I had no previous experience of designing converters, but decided to concentrate on the push-pull type of self-exciting power oscillator shown in the circuit diagram, as this seemed to be likely to be efficient, and does not rely on any trick techniques.
A difficulty in applying rational design techniques was that I had several likely-looking cores in the junk-box, but absolutely no technical specification for any of them. So I decided to experiment with a ferrite E-core salvaged from a dead PC monitor. This was obviously capable of handling much more power than the 5W or so required for the project, but using a large core has the advantage that a large volt/turn ratio can be used, and therefore few turns are needed for the high-voltage secondary winding. A picture of the completed converter is included, together with a regulation curve for the original 175v bridge-rectified output (now used with a doubler to give 350v).
Three secondary windings are used, each with a charge-pump doubler constructed from high-speed rectifier diodes. The rectifier outputs are connected in series to produce the required outputs at 250v, 600v, and 800v. The required secondary turns are as follows –
250v – 121 turns: 250v output used for the timebase
350v – 170 turns: 600v output used for X and Y amplifiers
200v – 97 turns; 800v output used for CRT
The negative grid voltage for the CRT, required to control the brilliance of the display, is derived from the 12.6v heater supply.
Diodes 2 and 3 have been added to allow the inverter to operate efficiently at low output currents, and to reduce the maximum output voltage on no-load.
The system oscillates at about 25kHz, which is well outside the range of audibility, and the residual ripple at the output is 50kHz. T1 is included in the dc input connections to assist in reducing emc problems, but is not strictly necessary. It can be wound on a toroidal core recovered from a SMPSU, for example. Is wound with the two windings arranged on opposite halves of the circumference of the toroidal core, the beginnings of the windings entering the core from the same side, and progressing in opposite directions around the core. The 2SD1723 switching transistors were from the junk box: any small, low-power transistor with a transition frequency in the tens of MHz range or above would suffice.
With an efficiency in the range of 85 per cent, no heatsinks are required. The 1N4937 is a fast-recovery diode, but 1N4007 diodes perform almost as well as these, and could be substituted with little loss of performance.
Textbooks suggest that for best regulation, the output winding should be in two halves, with the primary and feedback windings sandwiched between them.
The wooden box has dimensions 155mm x 88mm x 43mm, and is made from hardwood strips 9mm thick. The joints are simple halving joints, so the box is very simple to make using only a fine-toothed saw, a plane, and a chisel. The top plate is made from double-sided glass-fibre pcb material so that the valveholders can be soldered directly to the lower surface. This means that no screws are visible on the top surface. I have added tubular metal supports for the valves, soldered to the top surface, but although these look nice, and steady the valves well, they need a lathe to make them, and they are not really necessary.
|Quantity||RS Part #||Part description|
|1||249-9222||100k log pot|
|1||148-203||62R 1W resistor|
|9||348-5432||1N 4937 Diode|
|1||286-8757||DC Supply Socket|
|1||288-137||Common Mode Choke|
1 DG7-6 Cathode Ray Tube
1 B7G Valveholder
1 B9G Valveholder
1 EF91 Valve
1 EF80 Valve
1 EF184 Valve
1 Ferrite Transformer Core
1 Wooden Box
1 2M2 Linear Potentiometer
2 2SD1723 Transistor