Fullscreen HTML5Med Res (??MB)HTML4

Vintage Radio Philips Philips BX205 BX205 B-01 B-01 superhet superhet radio radio This 1950s valve radio is switchable between AM broadcast band and shortwave reception. Strangely, it uses battery valves but does not have a battery compartment, and it also has no internal antenna. Nor does it have any stations marked on the dial. It’s a bit of a head-scratcher! I bought this radio a couple of years ago on eBay. It didn’t work, and as I couldn’t immediately figure out why, I got bored with it. So it sat in a corner (metaphorically speaking) for quite a while. With the previous lockdown in Melbourne, “one of these days” finally arrived, so I decided to resurrect it. Its tuning covers two bands: the usual medium-wave band from 530kHz to 1600kHz, plus a shortwave band from about 5MHz to 16MHz. It uses four battery valves with 1.5V filaments, but there is no battery compartment. It came with a cord attached, but no plug on the end. Presumably, the idea was that you wired it up to 102 Silicon Chip a pair of batteries hidden away in a nearby cabinet. As it has no internal loop or ferrite rod antenna, it requires an external antenna. It doesn’t seem to be a model made specifically for Australia as the dial does not show radio station names, just a rough indication of frequency and wavelength. When I got it, the radio was in reasonable condition, with only minor scratches on the Bakelite case. To remove the chassis required removal of the rear heavy cardboard cover, two screws that held the chassis in place, and the knobs. The loudspeaker looked rather moth-eaten with a couple of holes, but seemed workable. Australia’s electronics magazine By Charles Kosina The speaker transformer is in an unusual large cylinder at top right, visible in the top view of the chassis. The bottom view shows the messy wiring which is typical of radios of that era. It makes modifications somewhat tricky. The circuit diagram (Fig.1) shows that it is a fairly standard design. The copy I managed to download did not have very readable lettering, but with the aid of Photoshop, I cleaned it up. I also added the component values and pin numbers for the valves. That made circuit tracing much easier. One of the banana sockets on the back of the set is for a ground connection and the other two are the antenna inputs. The top one connects directly to the input coil (S1 or S3) via the band selection switch. The second connection is via 100kW resistor R14 and is marked for LOCAL stations. I think that the station would have to be awfully close to get through that much attenuation. The input transformer secondaries (S2 or S4) are applied to grid 3 of B1, the DK92/1C2 pentagrid valve, again via the band selection switch. The local oscillator uses grids 1 and 2. The tuning capacitor is a two gang unit, C4 and C5. Band changing The switching between the two bands is rather complex, and interpreting the diagram is no mean feat! On the antenna coil side, it is essentially a 4-pole, 2-position switch. Two poles are used for switching the antenna between the mediumwave, S3 coil and the short wave S1 coil. The other two poles switch grid 3 of B1 between the tuned secondary coils, S2 and S4. The local oscillator gets a bit more complicated. The medium-wave tuning range is 985kHz to 2050kHz, ie, 450 kHz above the tuned input frequency. The padding capacitor C14 (476pF) is siliconchip.com.au Fig.1: I added the component values to this original circuit for the BX205 B-01. Note the switched (++) and unswitched (+) supply connections and the somewhat complicated band-switching arrangement. A single wafer switch is used to select between two sets of antenna coils and oscillator coils. effectively in series with tuning capacitor C5 for reasonable tracking with the signal input frequency. There are three coils on the shortwave oscillator, with two of them connected by 120pF capacitor C11. This appears to be an alternative way of tracking the oscillator with the input signal. Switching between the two bands is again by a four-pole, two-position switch in the same assembly as the others. The difference frequency of 450kHz passes through a double-tuned IF transformer (S11-S14) and is then amplified by variable mu pentode B2, a DF91 or 1T4. This is followed by another double-tuned IF transformer (S15/S16) feeding the diode in B3, a DAF91/1S5. As well as the envelope detection for recovering the audio, the filtered negative DC component is used to provide AGC to the two previous valves via 1.5MW resistor R4. The audio is then amplified by the pentode section of B3, and feeds into the grid of “power amplifier” B4, a DL94 or 3V4. A transformer (S17/ S18) couples this to the loudspeaker. The gain of these battery valves is not that high, so it can’t be wasted by having any negative feedback in the audio stages. Not shown on the circuit diagram is a connection to two screw terminals siliconchip.com.au on the side of the case. These connect to either end of the volume control R6, and are provided for external audio input. The audio signal from the radio is applied to R6 via 56kW resistor R15, so the external source should easily be able to ‘short out’ the audio from the radio (which presumably would be tuned off-station). Power supply Note that the power supplies do not have a common earth. The 90V negative goes via 560W resistor R13 to chassis Earth, resulting in a grid bias voltage of about -1.8V for the DL94. The 90V supply is connected directly to the anode circuits of B2 and anode and screen grid of B3, not via the switch. I’m not sure of the reason for this, but perhaps it keeps some capacitors charged up, preventing a thump from the speaker on turning the power on. Restoration Coupling capacitors are likely to be leaky after all this time, so I replaced C22 and C24 with modern high-voltage types, and also increased their values to 220nF. I fitted a suitable plug on the power cord and connected it to a mains supply that can deliver 90V and 1.5V. There was no sound at all from the speaker, so out came the chassis. The first thing I did was to test the continuity of the filaments in all the valve. Sadly, the DAF91 had an open filament. I decided to work backwards; connecting a signal genera- A close-up of the Philip BX205’s dial. Australia’s electronics magazine February 2021  103 tor to the grid of the DL94 provided a clean tone in the loudspeaker. At least this proved that the output valve and speaker transformer worked. The next problem was the defunct DAF91. Searching various websites, I found that this type is available, but at prices ranging from $26 to over $80, more than I paid for the entire radio! Valve substitution Fig.2: this is the circuit that I ‘juryrigged’ up to replace the open-circuit DAF91 diode pentode valve. It uses a JFET to perform a similar role to the pentode, plus a schottky diode for demodulation. I fitted this to the underside of the chassis and left the defunct valve plugged in for the sake of appearance. Fig.3: another cobbled together fix, this time for an open-circuit antenna coupling transformer. It’s made up of four separate chokes and relies on coupling through proximity; while it may seem crude, it works just fine. I did not want to hold up getting the radio working, so I decided on a workaround. My approach will no doubt offend the purists! How many of you are old enough to have heard of Fetrons? Teledyne Semiconductors made plug-in solid-state replacements for a number of different valve types. Editor’s note: In next month’s issue of Silicon Chip we’ll have a detailed article on Fetrons. They consist of two N-channel JFETs connected such that they have similar characteristics to a pentode valve. They are no longer available, and never were for this valve. But I thought I could whip up something similar. I decided on a simplified approach of using just one JFET and used the only type that I have in stock, a J310 (2N5484) to replace the pentode section. The arrangement that I came up with is shown in Fig.2. The 1MW resistor (R10) in the radio circuit is far too high for a drain load of the FET, so I reduced this to 33kW. This resulted in a drain voltage of 13V, well within the maximum rating of 25V. If we compare the performance of the JFET configured thus with the valve, they are surprisingly similar. The DAF91 has a transconductance of around 720µ℧ (or microsiemens, if you prefer). The load resistance is the parallel of R10, R12 and Ra (the plate resistance) which comes to 250kW. Hence, its voltage gain is 180 times (0.72µ℧ × 250kW). Doing the same calculation with the JFET, the current through it is about 1.9mA. This gives a Yfs of about 8500µ℧ and Yos of around 20µ℧, or 50kW. The effective load resistance is the 33kW in parallel with the 50kW, ie, about 20kW, resulting in a gain of about 170; not far short of the pentode. I left the defunct DAF91 valve plugged in as it does nothing; it’s just for show now. The JFET circuit plus the schottky 1N5711 diode replace its functions. Now I had the audio stages working, but injecting a signal into the antenna terminals still produced nothing. Putting a scope on the oscillator coils showed that the local oscillator was not working on either band (medium or shortwave). Faulty transformer Rather than trying to analyse what was at fault, I decided to replace all the capacitors in the oscillator section, and sure enough, the oscillator fired up The DAF91 diode pentode valve (B3) was open-circuit and therefore replaced with a circuit based around a J310 JFET shown in Fig.2. This is shown at the base of B3 which is circled in white below. 104 Silicon Chip Australia’s electronics magazine siliconchip.com.au The BX205 B-01 could be considered a portable version of the previous BX205U /00/ 35 & BX205U (a 5-valve mains powered superhet). The circuits between the BX205 and U-series are somewhat similar with some changes to account for one less valve and the use of an A & B battery instead mains. on both bands. However, I still could not receive anything from the antenna input on MW. Injecting a signal directly into grid 3 of the DK92 worked. This led me to suspect the input coil, and sure enough, the S4 winding was open-circuit. Taking apart the input transformer required much care, as the aluminium case is just pressed into place, and I had to prise it apart. The damage was then apparent. The HF input coils appeared intact, but the MW winding had loose, thin wires hanging off it. These are extremely thin wires, and after some attempts at repairing it, I decided it was just not possible. The thin wires would not accept solder at all. This presented something of a dilemma, so I came up with an alternative, shown in Fig.3. I used my collection of inductors to cobble up a suitable substitute that would fit in the case. The input from the antenna is ap- The chassis of the BX205 B-01 was rusted and the speaker grille had started to disintegrate. The non-working DAF91 valve (B3) was left in place as it has no impact on the rest of the radio. C1/2 S17 S10 S8 S11-14 B1: DK92 S2 S18 B2: DF91 B3: DAF91 B4: DL94 C4/C5 S4 siliconchip.com.au Australia’s electronics magazine February 2021  105 105 On the left is the short wave input transformer S1/S2, which is intact. The faulty S3/S4 was replaced by fixed inductors L1-L4. plied across a 10µH coil (L1). This is placed alongside a 100µH coil (L2). The side-be-side arrangement results in good coupling between the coils. I then added a 220µH coil in series with L2. The resulting total of 320µH was a bit too high for the tuning range of C4, so I added a 1000µH inductor, L4, in parallel which resulted in an effective value of 242µH. This may not be the exact value needed, but it was close enough so as not to adversely affect the tracking and performance. Alignment The standard alignment procedure is to set the receiver near the top of the frequency range, say 1500kHz, and adjust trimmer C7 for maximum output. Then the receiver is set to the low end, say 600kHz, and the inductor is trimmed. Obviously, I could only make the top-end adjustment, and as it turned out, the sensitivity at the low end was comparable, which meant that my inductance value must have been close enough. Fig.4: the set’s frequency response is down by 3dB at 60Hz and 3.3kHz. heavily polluted by hash from all the electronics inside. More accurate measurements with a signal generator showed that it requires about 10µV for something useable, but more like 100µV for a decent sound. This did not vary much over the range of either the MW or SW bands. It could probably be slightly improved with tuning the various coils, but quite frankly, I dared not touch them as by now they could be awfully brittle. I did a frequency response graph from the antenna to the speaker, shown in Fig.4. While the response at 50Hz is only down by 3.9dB, the waveform is extremely distorted, and the sound from the small speaker is minimal. The primary inductance of the speaker transformer is obviously not high enough for this frequency. The high-frequency -3dB point is about 3.3kHz, and by the time we get to 5kHz, the response is well down. This is primarily determined by the intermediate frequency bandwidth of the set. Without negative feedback, there is noticeable even harmonic distortion in the Class-A audio output stage. This is evident in Fig.5, which is a scope grab of the output just before clipping sets in. Unlike odd harmonic distortion, even harmonic distortion is not particularly objectionable, so the sound with a strong station is acceptable. The maximum power output is about 250mW, quite adequate for this sort of radio. SC Performance I decided it was time to install a proper outdoor antenna. I ran about 10m of wire between a 5m-tall mast at my back fence and a short mast on the metal roof. I connected the shield of the coaxial cable lead-in to the roof. The results were amazing; all the Melbourne stations came through cleanly with little noise between stations, and on shortwave, there were many stations with strong signals in the evening. By comparison, using a piece of wire indoors gave a signal 106 Silicon Chip Fig.5: as the set lacks any feedback around the output stage, there is plenty of second-order harmonic distortion in the output waveform. At least it is more pleasant-sounding than odd-order distortion! Australia’s electronics magazine siliconchip.com.au