Fullscreen HTML5Med Res (??MB)HTML4

replacement bipolar transistor units Can a very simple circuit replace a mechanical vibrator in a vintage radio? Could early germanium power transistors from the 1960s be used in a design that would have been economical then? The answers to both questions are yes; here is how it would work. Part 3: by Dr Hugo Holden T his is the fourth & final full vibrator replacement design I’m presenting; the other three were described in the June and July 2023 issues and were based on Mosfets or Darlington transistors. This one is based on bipolar transistors; while it has an extremely elegant circuit, it’s the most difficult to build as it involves a custom-wound transformer and custom housing. As with the other designs, two rectifiers formed from four diodes replace the secondary switching contacts. Power switching circuits that use bipolar junction transistors (BJTs) without driver transformers have large energy losses in the base bias resistors. Fig.1: the vibrator primary replacement circuit comprises just two germanium PNP transistors, two resistors and a transformer. The transformer converts the 24V peak-to-peak output to a much lower voltage signal for driving the transistor bases and limits the base current, while the external transformer controls the oscillation. 94 Silicon Chip The required transistor base-­emitter current is in the order of 0.21A (210mA) because the maximum collector current (the primary side switching current) in this application is in the order of 2.1A (with my ZC1 Mk2 in transmit mode), and the transistors must be operated in saturated switching mode. As a rule of thumb, a 1:10 ratio of base current to collector current is required to ensure saturation. Here we can see one of the significant advantages of Mosfets in such a role, with their high-impedance (capacitive) gates. If the 0.21A base current is sourced from the fellow transistor’s collector, which is transformed up to 24V in use, the power dissipation is around 5W total in the two bias resistors. More efficient transfer of power to the transistor bases involves using a feedback transformer, as shown in the circuit diagram, Fig.1. The ASZ17 germanium PNP transistors I’m using have a collector-emitter saturation voltage drop of only 0.15V at 2A, which is favourable compared to its silicon transistor counterparts like the 2N3055, with a C-E drop of around 0.3V. Modern silicon power transistors can do a little better than this, but the ASZ17s are pretty close and undoubtedly impressive for their time. The transformer is a small ‘feedback transformer’ that fits inside a similar housing to the original vibrator. The configuration is a version of the Royer Oscillator. The feedback transformer transfers the appropriate amount of drive current to each transistor base on consecutive half-cycles from a potential that is stepped down from the 24V peak collector voltage to about 3.6V. So the total transistor base power for the two transistors is about 800mW. The power loss in the 680W bias resistors is about another 850mW (425mW each). The transistor losses are about 0.3W due to their low collector-­emitter saturation voltages. Fig.2: this shows how the vibrator replacement (including the four BY448 diodes for the secondary) connects to the external transformer. This is important to understand since the properties of that transformer are responsible for causing oscillation and determining the operating frequency. Australia's electronics magazine siliconchip.com.au The power losses in the four HT rectifiers (in transmit mode output current around 80mA) are about 200mW. So the total power loss is only about 2W, which, coincidentally, is practically identical to the original mechanical vibrator. Notice how pin 4 of the socket, the 12V power supply connection, is not used. The circuit is powered by the ZC1 unit’s main primary power transformer connections. No DC voltage is applied across this small coupling transformer’s primary, even if the oscillations stop due to an extreme overload. The transformer wire lead colours are also shown in Fig.1 since they match those on the physical transformer. Fig.2 shows the electrical configuration when the unit is plugged into the ZC1 Mk2 radio’s power supply. Starting from the premise that one transistor is conducting, the circuit oscillates because, as time passes, the main power transformer’s primary current begins to magnetically saturate the transformer’s core, suddenly increasing the transistor’s collector current. The induced voltage is proportional to the current’s rate of change with time or dI/dt, and this rate of change falls away with core saturation. Therefore, the voltage via the feedback transformer directed to the conducting transistor’s base drops rapidly, along with the base current, as magnetic saturation begins. This process is accelerated via positive feedback, and the transistor rapidly comes out of conduction. The drive voltages at the base-emitter junctions reverse polarity, and the other transistor is driven hard into saturation. The process repeats for another half cycle. On switch-on, due to the inexact matching of the transistors, the asymmetry in the current encourages initial small sinusoidal oscillations, which rapidly grow to establish stable saturated switching in less than half a second. The switching frequency is determined by the magnetic saturation properties of the main transformer core and works out to about 60Hz. That is a little slower than the original V6295 vibrator, which ran at about 100Hz. This does not matter, provided the 10µF filter electrolytic capacitors in the radio’s power supply circuit are in good condition. siliconchip.com.au While Fig.1 might appear to show the load being driven by the emitters, from an electrical perspective, the load is actually in the collector circuit with the power supply circuit acting in series. This is because the drive voltage is applied by the feedback transformer directly and independently to the transistor base/emitter connections. Some people have become confused, thinking that the transistors are being used as emitter-followers and therefore could not act as saturated switches. An actual emitter follower circuit is unsuited to saturated switching or for use in a Royer-style DC/DC converter application. Regarding diodes D1-D4, it is necessary to have a very high PIV diode rating. That’s in case the unit is plugged in and out while running (or has a bad connection to one of its socket pins). In that case, the undamped collapsing field of the main vibrator transformer can produce a peak voltage high enough to break down and destroy a single 1N4007 rated at 1kV. Two series 1N4007s are required to prevent this. BY448s are 1.5kV rectifiers for modern switch-mode power supply applications and are even better. Construction This transformer-based version is the most challenging vibrator replacement for the home constructor to build. The easiest to make is the self-­ oscillating Mosfet version described previously. The first step is manufacturing the tools required to make a UX7 base. This is done with two solid aluminium cylinders. I traced the original UX7 base pattern from a scan to create a template to mark the position of the pins. There are two fat pins and five thin pins. The tool makes both a carrier for a disc of circuit board material and a template to mark the holes. This can be rotated in the lathe to set its outer diameter to 36mm – see Photo 1. I made another aluminium piece to support the pins while I pressed them into the PCB discs (Photo 2). I set the hole for the fat pins at 3.95mm and 3.1mm for the thin pins. The pins are pressed as an interference fit into the PCB using the drill press and the carrier, with a small socket to do the pressing – see Photos 3 & 4. It is not necessary to rivet the Australia's electronics magazine Photo 1: I cut and etched these PCBs as a starting point for the 7-pin bases. The tool above them masks the areas where copper is to be preserved during etching. Carrier to support 7 pins Photo 2: I made this tool to press the pins onto the etched PCB disc using a drill press. Photo 3: the pins being pressed in. Note the socket mounted in the drill press chuck for that job. Photo 4: the completed custom UX7 bases. August 2023  95 ◀ Photo 6: a small lathe with an RPM indicator and revolution counter is a handy aid in winding transformers. Photo 5 (left): the BY448 diodes have been soldered in series across the appropriate pairs of pins, and the three extra wires (tinned copper wire surrounded by silicone insulation) have also been soldered in place. pins in as the press fit and soldering to the copper laminate on the PCB material impart the required strength. One reason I didn’t rivet the pins is that it can split the thin brass material they are made from. The above method creates a very stable and reliable UX7 base into which the BY448 diodes can be fitted (Photo 5). Only three wires pass from the base up into the unit, made from 0.71mm tinned copper with silicone rubber insulation. One is the Earth connection, while the other two go to the transistor emitters. You might be wondering why I didn’t use a prefabricated base like the Amphenol UX7 base I used in my previous vibrator replacement designs. The Amphenol bases are pretty thick, and there was a limit to how tall the unit could be and still fit in my ZC1 Mk2 transceiver. The space needed inside the canister to fit the transformer makes this more difficult. The Amphenol base could probably be made to work for a taller unit. The housing would need to be adjusted to be the right size to accept such a base. transformer core must be well away from magnetic saturation. It must also have a precise DC secondary resistance to avoid the need for additional resistors in the transistor’s base circuit. It must fit inside the machined aluminium housing (34mm internal diameter) that replaces the original V6295 vibrator. The transformer must also provide a good base drive current to the transistors’ bases to ensure they are saturated with a 2A collector current. This base current is around 150-250mA, a typical value being 210mA. A suitably-­ sized core is 1cm2 inside the bobbin with grain-oriented steel laminations. In this operating mode, the feedback transformer’s secondaries are effectively shorted out on each half cycle by the base-emitter voltage of about 0.45V. The DC load resistance is of the transformer wire itself. The electrical equivalent circuit for this somewhat unusual arrangement is shown in Fig.3. This indicates that the transformer naturally limits the base current to around 227mA. For this calculation, the primary value DC resistance is reflected onto the secondary winding by the impedance ratio, which is the square of the turns ratio. The drive voltage for the feedback transformer during operation is a 24V square wave at 60Hz. The diodes with forward voltages of 0.45V represent the base-emitter junctions of the ASZ17 transistors. The RMS current in each secondary winding is about 160mA, which is over the upper limit for the current carrying capacity of 32AWG wire (using the 500 circular mils per amp specification of 126mA for 32AWG wire). However, in this case, the total power dissipated in each winding is only about 270mW. Also, because of its physical size and external location on the bobbin, the winding barely gets warm, and there is no threat to the grade-2 enamel insulation. The generally accepted flux density (Webers/m2 or Teslas) for iron-cored low-frequency transformers is in the vicinity of 1T. The higher this value, the greater the chance of pushing the iron core into magnetic saturation. Transformer requirements and design The transformer must have specific properties. It must have an iron core due to the low operating frequency and a primary winding designed for a low core flux density. This is because the core saturation properties of the main power transformer determine the operating frequency, not the driver transformer. During each half of the squarewave cycle (about 8.3ms), the driver 96 Silicon Chip Fig.3: Rp, Rp’ and Rs are resistances inherent to the driver transformer; Rp is the primary winding resistance, Rp’ is that resistance reflected into the secondary and Rs is the secondary winding resistance. These limit the current into the transistor bases (shown as diodes) to about 227mA per the calculations. Australia's electronics magazine siliconchip.com.au Estimating transformer winding resistances ◀ Photo 7: the completed windings on the bobbins with clear Kapton tape over the top. This also depends on the magnetic properties of the iron core; some materials saturate before others. As noted, the feedback transformer mustn’t come anywhere near saturation. By selecting a modest value of 0.5T, we ensure that the core is well below saturation. I performed some calculations to verify this would be the case, but they are a bit long and complicated to present here. I also won’t go into other aspects of transformer design here, like leakage reactance, core losses, winding capacitances etc. Making the transformer Improved wire enamels and factors of economy have meant that the configuration of the typical power transformer has changed over the last century. Until the mid-1960s, even those transformers with very fine wire and thousands of turns were wound in perfect layers, with very thin rice paper like insulation between each layer. This had disadvantages as residual salts in the paper could, in conjunction with water vapour, cause corrosion of the copper wire. They also had higher inter-winding capacitances. Still, one can’t help but admire the winding perfection seen in these vintage transformers. Such windings are still used in oil-filled car ignition coils. The primary winding is wound onto the bobbin first with 2000 turns of 36AWG (0.125mm or 0.127mm diameter) enamelled copper wire. Then the secondaries are wound on bifilar, ensuring they have identical DC resistances of about 10.6W. This means that enough DC bias can be developed, in conjunction with the 680W resistors, for self-starting and to limit the base current to the correct value. The wire sizes and numbers of turns siliconchip.com.au You can estimate transformer winding DC resistances from the number of turns and the geometry of the bobbin. The number of turns per layer is closely approximated by the diameter of the wire (including its enamel) divided into the bobbin width. Dividing this number into the total number of turns gives us the number of layers, which is then multiplied again by the wire diameter to calculate the winding height. Once that is known, it is simple to calculate the average length of a turn bisecting the centre of the windings, assuming 90° turns (ie, a square bobbin). We can then multiply this value by the number of turns to calculate the length of the wire, then multiply that by the resistance per length for the wire used to get the actual resistance. Let’s go through this exercise for the primary winding of the feedback transformer. The bobbin is 16.55mm wide (measured) and the 36AWG wire diameter is 0.135mm, including its enamel (measured with a micrometer), so there are 122.6 turns per layer (16.55mm ÷ 0.135mm). A 2000 turn winding is 16.31 layers high, or close to 2.20mm (16.31 × 0.135mm). The inner bobbin, where the winding starts, measures 11.35 × 11.35 mm. Therefore, with a 2.2mm high winding, we have the geometry shown in Fig.4. The average turn length is 54.2mm (13.55mm × 4) and with 2000 turns, the wire length is 108.4m. 36AWG wire has a resistance of 1.361W/m, so the expected primary resistance is 147.5W (108.4 × 1.361W). The measured resistance of the actual wound transformer primary is very close, at 144W. So this method of estimation was within 3% of the actual value. Let’s apply the same principles to the two secondaries, which total 600 turns (two bifilar-wound 300-turn windings). The 32AWG wire on the micrometer measures 0.23mm in diameter. There are 71.95 turns per layer (16.55mm ÷ 0.23mm) and 8.34 layers (600 ÷ 71.95), for a thickness of 1.92mm (8.34 × 0.23mm). Adding this on top of 0.1mm insulation tape on top of the primary gives the geometry shown in Fig.5. The average turn length is therefore 71.48mm (17.87mm × 4), and there are 600 turns total, making the wire length 42.9m. 32AWG wire has a resistance of 0.5383W/m, so the total secondary resistance is expected to be 23W (0.5383W/m x 42.9m). This makes the calculated DC resistance of one 300t winding 11.5W, compared to a measured value of 10.6W, within 8.5%. The calculations slightly overestimate the DC resistance, more so on the secondary, because the windings are modelled as rectangular. In practice, the corners become more rounded as the winding height increases, shortening the wire length of each turn. Figs.4 & 5 show the total height of the windings as 4.22mm (2.2mm + 0.1mm + 1.92mm). The plastic bobbin is about 5.75mm high, so there is enough room for the outer coat of insulation seen in the photos. Fig.4: we can estimate the winding thickness and average turn length by assuming the primary windings are square. are such that the full bobbin volume is used with just enough room for the required insulation. I used a small lathe with an added turns counter and RPM meter (Photo 6) to wind the transformer. With practice, Australia's electronics magazine Fig.5: we assume the secondary windings are square and stacked on top of the primary and insulation, allowing us to estimate their thickness and average turn length. it is possible to make the windings very even, as shown in Photo 7. The 2000-turn primaries have been wound on, and two layers of polyimide (Kapton) tape have been applied. In general, when winding transformers, it August 2023  97 Photo 8: fibreglass tape makes connecting flying leads to the fine wire of the windings much easier. Photo 9: after adding more wires and fibreglass tape, the bobbins are complete and ready for the cores. Photo 10: Another layer of fibreglass tape covers the soldered wire connections. is important to keep the windings as regular and orderly as possible. The secondaries are then wound on bifilar and again, two layers of Kapton tape. Then add some special fibreglass tape (Scotch number 27, made by 3M and available from Hayman’s Electrical) to assist in terminating the wires to their flying leads, as shown in Photos 8 & 9. This fibreglass tape is also used to finish the bobbin as it is far superior to the usual yellow plastic transformer tape. The 32AWG secondary wire used here is insulated with nonself-fluxing tough grade 2 enamel that must be carefully scraped before soldering. The 36AWG primary wire has self-fluxing enamel. Photo 10 shows some finished bobbins. The bobbins can then be stacked with their laminations, the edges of which are lightly painted with Fertan organic rust converter. This deactivates any surface rust crystals on the cut lamination edges. I prepared transformer brackets to allow them to be mounted inside a 34mm diameter cylinder, made from ¼in-wide, 0.8mm-thick brass strip and ½in-wide, 0.6mm-thick brass strip (stocked in model shops). I folded the brass and soldered it to create the brackets shown in Photo 12. The transformer stack is a firm press-fit into the bracket and is also effectively glued to it by the varnishing process. Photo 13 shows the transformers ready for vacuum varnishing. While the transformers could simply be dipped in varnish, it is better to apply a vacuum. A full vacuum removing most of the ‘standard’ air pressure (1013hPa) is good, but it requires a pump. A vacuum of about two-thirds that can be attained with a simple syringe, a strong arm and a jam jar, as shown in Photo 14. This shows one of the transformers inside the jam jar full of polyurethane varnish, subjected to a partial vacuum. This causes the air to exit the small spaces in the transformer windings and the varnish to pass in. Pulling the syringe upwards expands a tiny air bubble into a large volume. As it is hard to hold it there for long, you can use a brass rod to lock the syringe plunger and allow 15 minutes for the multitude of fine air bubbles to exit the transformer. Finally, I hung the transformers up to air dry (Photo 15). This process could be sped up with an oven; however, I simply left them for one week. which is very close to 1mm in diameter and has a springy quality. If wound around a 22mm diameter cylinder, it springs back to about 42mm (Photo 17) and fits into the 0.5mm-deep groove in the housing. The top cover attaches with four countersunk 1/2in-long 1/8in BSW screws. Photo 18 shows the holes I drilled and tapped for the TO-3 (ASZ17) transistors and transformer brackets. The transformer mounting holes are tapped for 1/8in BSW and countersunk. The transistor collectors connect to the case and ground (negative), so there is no need for any insulating washers. Photo 11: the E-cores have now been slipped into the bobbins after coating them with rust converter. Photo 12: I fabricated the transformer brackets from brass strips of two different sizes (12.7 × 0.8mm and 6.35 × 0.6mm). 98 Silicon Chip Aluminium housings UP-Machining in Shenzhen, China, made the high-quality housings based on my drawings (Photo 16 & Figs.6-10). The UX7 base is retained by a wire clip made from #17 piano string wire, Australia's electronics magazine Assembly The 7-pin base is retained in the housing by the spring clip. As it is such a close fit, after applying polyurethane varnish on its edges and over the clip, it is extremely strong and impossible to rotate the base in the housing. The varnish could still be dissolved one day if disassembly was required. The base must be rotated to the correct position before the varnish dries to accommodate the rectangular top of the housing when plugged into the radio – see Photo 19. Photo 20 is a view into the unit before the transformer is inserted. Only three wires rise out of the base. The transformer is retained in the housing by two 1/2in-long 1/8in BSW Photo 13: some of the completed transformers, ready to be varnishimpregnated. siliconchip.com.au Parts List – Bipolar Vibrator Replacement 1 UX7 base (see text) 1 machined housing with hardware (see text) 1 custom-wound transformer (see below) 2 ASZ17 60V 10A PNP germanium transistors, TO-3 2 680W 1W resistors 4 BY448 1.5kV 2A axial diodes 1 300mm length of 0.7mm diameter tinned copper wire 1 300mm length of 1-2mm diameter heatshrink or spaghetti tubing 1 200mm length of #17 piano string wire (~1mm diameter spring wire) 4 ⅛in BSW × 10mm or ⅜in panhead machine screws 4 ⅛in x ½in BSW or 12mm countersunk head machine screws 2 10mm lengths of 1-2mm diameter green heatshrink tubing 2 10mm lengths of 1-2mm diameter blue heatshrink tubing 2 solder lugs various lengths of light-duty hookup wire Photo 14: drawing a partial vacuum on a transformer dipped in varnish allows the varnish to fill in all the gaps. Note the brass rod used to keep the plunger up against the force of the vacuum pulling it down. Transformer parts 1 EI-core transformer bobbin and lamination set, initial winding size 11.35 × 11.35 × 16.5mm 1 110m length of 0.125mm (36AWG) diameter enamelled copper wire 2 22m lengths of 0.2mm (32AWG) diameter enamelled copper wire 1 30cm length of ¼in (6.35mm) wide, 0.8mm-thick brass strip 1 30cm length of ½in (12.7mm) wide, 0.6mm-thick brass strip 2 ⅛in BSW × 10mm or ⅜in countersunk head machine screws and hex nuts 1 small roll of 0.1mm thick polyimide (Kapton) insulating tape 1 small roll of Scotch number 27 fibreglass tape 1 small tin of polyurethane varnish Photo 18: I drilled holes for mounting the TO-3 transistors, the transistor leads and the transformer mounting holes in the cases. The transformer mounting holes are countersunk. Photo 15: the transformers were hung for a week to let the varnish fully cure. 42 mm Photo 16: the aluminium housings and lids, ready to accept the electronic components. siliconchip.com.au Photo 17: after bending 1mm diameter piano wires around a 22mm cylindrical former, they spring back to around 42mm in diameter. They can then be recompressed to fit into the groove in the housing and will expand to prevent the base from falling out. Australia's electronics magazine Photo 19: after placing the UX7 base that I made and inserting the spring clip, I applied varnish and let it cure so the clip couldn’t be accidentally knocked loose. August 2023  99 Photo 20: an inside view of the housing with the plug in place. slot head countersunk screws with nuts and spring washers. Solder lugs are placed between the transformer mounts and the inside of the aluminium housing as the solder tie points for the two 680W 1W resistors and ground, and the black ground wire from pin 7 on the base. The transistors can then be screwed to the case with 3/8in-long 1/8in BSW panhead screws. The transistor base and emitter leads have a protective silicone rubber insulating sleeve applied, green for the base and blue for the emitters. The emitters connect to the blue wires leading to pins 1 and 6 in the base, as shown in Photo 21. It is best to use a 1W resistor for reliability, as the dissipation in each resistor is 426mW, and then taking into consideration the enclosed space they operate in. The top cover can then be fitted, as shown in Photos 22 & 23. Photo 24 shows the unit working in a ZC1 Mk2 communications receiver. It looks the part and suits the rugged character of the radio. Performance Scope 1 is a dual-trace recording of the emitter waveforms of the two ASZ17s (ie, the ZC1’s primary transformer connections) with the unit running in receive mode. It oscillates at close to 60Hz, with a very clean switching waveform. The 12.4V across half of the transformer primary plus the 12.4V supply voltage results in about 24.8V appearing on one transistor’s emitter while the other is conducting. After a time, due to the magnetic saturation of the ZC1’s transformer core, the induced voltage suddenly starts to fall. This takes the conducting transistor out of conduction, and the other goes into conduction for the next half-cycle. The base drive current for each ASZ17 transistor is around 210mA and the collector current in receive mode is about 1A. To see how well the Photo 21: the electronic components are now in place; only a few junctions need to be soldered. One end of each resistor goes to ground via a transformer mounting screw to the case (along with the ground lead), and the transistor collectors are in intimate contact with the case. Six solder joints are required, four on the transistor base and emitter leads. Photo 23: the completed bipolar transistor vibrator units look rugged, with the two TO-3 package germanium PNP transistors mounted on the outside of a machined aluminium case. Photo 22: the completed vibrator replacement ready for testing and use. Photo 24: the industrial look of the vibrator replacement unit suits the appearance of my ZC1 Mk2 communications receiver very well! 100 Silicon Chip Australia's electronics magazine siliconchip.com.au Fig.8: isometric view of the machined housing. Fig.6: a side view of the machined aluminium housing for the vibrator replacement. The holes drilled into the sides for mounting the TO-3 transistors and transformer are not shown. Fig.9: plan view of the lid for the machined housing. Fig.10: details of the grooves in the base of the housing. The square inner grooves are for the UX7 base, while the rounded outer groove engages clips in the radio to retain the unit. Fig.7: a top view of the machined housing. siliconchip.com.au Australia's electronics magazine August 2023  101 +25V Emitter Voltage ASZ17 (1) +100mV 0V 0V ASZ17 C-E saturation voltage Collector current 1A, Receive mode ZC1 XFMR Core Saturation begins +25V Emitter Voltage ASZ17 (2) +100mV 0V 0V Transistor Collector – Emitter – Saturation Scope 1: the transistor emitter (external transformer primary) voltages during operation. The switching frequency is measured as 60.4Hz. transistors were saturating, I wound the scope gain up to 100mV/div on DC, giving the result shown in Scope 2. This shows the very low collector-­ emitter saturation voltage of the ASZ17 germanium power transistors. In transmit mode, the collector current is about doubled to 2A, and the saturation voltage increases slightly to 150mV (Scope 3). If these were Mosfets, that would correspond to an RDS(on) of 75mW. The oscillation frequency slows a little bit due to the additional loading. In transmit mode, the power loss in each transistor is about 0.3W (2A × 0.15V). The base-emitter power is 0.0945W (0.21A × 0.45V), so the dissipation in each transistor is only about 400-600mW (there are some additional losses during the switching transitions). So the whole assembly runs very cool on account of the size of the metal housing. The waveform in Scope 4 was taken with an isolated scope across Scope 2: by increasing the sensitivity of the oscilloscope compared to Scope 1, we can see the transistor collectoremitter saturation voltages are just over 100mV at just over 1A. That’s good for an obsolete germanium transistor. the coupling transformer primary, between pins 1 and 6 of the device. It is a 48V peak-to-peak rectangular wave. The radio’s HT measures +243V DC with only 70mV of ripple (see Scope 5). My radio has been upgraded with 25µF filter capacitors, so with the original 10µF capacitors, the ripple would be a little higher. Still, this is a very low figure for this type of power supply. The electronic vibrator replacement gives an HT of about 10V or 4% higher than the original V6295 vibrator in receive mode (with the sender switch on). This is to be expected, as the mechanical unit can’t quite reach a full 50% duty cycle due to its contact gaps and the time that neither contact is closed. In transmit mode, the output voltage from the electronic unit is about 14-15% higher than the original unit. So this electronic unit is superior overall to the electromechanical V6295. RECEIVE MODE VOLTAGES WITH ELECTRONIC V6295: +244.6V DC AC Ripple, 120Hz Approx. 3Vpp NOTE: -68V rail is ZERO in transmit mode and main output voltage at junction of L9B and L20A is +288V ELECTRONIC V6295 L20A +12.1V 0V +243V DC AC Ripple, Approx. 70mVpp Scope 4: connecting an isolated ‘scope across the two emitters, we see that they are generating a relatively clean 48V peak-to-peak square wave. 102 Silicon Chip -68.3V DC AC Ripple, Approx. 100mVpp Scope 5: three views of the ripple out of the transceiver’s power supply with the vibrator replacement operating. The amplitude is low and will not interfere with the set’s operation. Australia's electronics magazine siliconchip.com.au +150mV ASZ17 C-E saturation voltage drop, transmit mode, Collector current 2A 0V VIBRATOR TRANSFORMER 3/IT/9 47W 5W +150mV 12V 1.5W 400μF 2N3055 Scope 3: the same scenario as Scope 2 but with the ZC1 Mk2 in transmit mode, where the transistor collector current is a little over 2A. The saturation voltages have increased to a little over 150mV. Note that the 470nF tuning capacitor used in the oscillator-driven Mosfet-­ based vibrator replacement presented last month is not required here. Scopes 6 & 7 show the switching transients with this unit. Likely, because the transistors in the self-oscillating version do not switch-on as abruptly, or switch-off as quickly, as the oscillator-driven versions, there is more damping during the change-over time, suppressing the switching transients on the transformer primary. Also, should the oscillation stop for some reason (perhaps due to an overload), the base and collector currents Another BJT-based vibrator replacement Fig.11 shows a circuit for a 2N3055 silicon bipolar transistor-based vibrator replacement, originally published in Electronics Australia magazine, October 1975 (pages 58-61). As presented then, it was built on tag strips mounted on a large metal plate – much bigger than the original vibrator, making it a bit impractical. Notice the R-C snubber networks on the transistor collectors. Without these, because of the high transition frequency of the silicon transistor EM401 150W 16μF (compared to a germanium transistor), the circuit is unstable and bursts into oscillation at a high frequency. However, those snubber networks can be omitted if each 2N3055 has a 100nF collector-to-base feedback capacitor. Since the base drive is acquired from the opposite transistor’s collector, the dissipation in the 47W resistors is very high at around 5W and only just below the resistor ratings. So it is substantially less efficient at acquiring the transistor’s base drive than the ASZ17 circuit and much less efficient overall. This is why I did not use the EA design, but came up with SC my own. ASZ17/TRANSFORMER UNIT Scope 6: even without a tuning capacitor across the radio’s transformer primary, overshoot and ringing are well under control thanks to the gentle transition characteristics of the ASZ17 transistors in this configuration. siliconchip.com.au 150W 1.5W Fig.11: the EA October 1975 Solid-State Vibrator circuit. It works but is very inefficient, with each base resistor dissipating almost 5W. This shows why the transformer is necessary for my version. are too low to cause any trouble. ASZ17/TRANSFORMER UNIT 47W 5W 2N3055 EM401 16μF 0V 400μF Scope 7: a close-up of Scope 6 with a faster timebase showing the transition in detail. The overshoot is only a few volts and dampens out after just a couple of cycles. Australia's electronics magazine August 2023  103