Some consumer valve gear

I have had a few items of valve "consumer electronics" come my way over recent years. The items I have worked on I have listed (or linked) below together with my thoughts about them.

back to index

Sansui 1000A hi-fi receiver

The first time I saw the Sansui I did a double take. I was surprised to see what appeared to be a piece of 1970's transistor hi-fi being sold at the NVCF, but as I looked away from the front panel and over the chassis I saw that this receiver was valve and not transistor.

Completed item, the Sansui 1000A valve receiver

I was at the "National Vintage Communications Fair" in Birmingham (UK). This show is for vintage communications equipment and one can normally find old and vintage radios (both broadcast and ham band), old telephones, televisions & hi-fi together a selection of interesting components.

The Sansui had no top cover and the air of a "project in progress" about it although the front panel was unmarked and the valve complement seemed complete. After some unsuccessful haggling I ended up handing over the asking price of £65 (about $100) cash. I figured even if it was in some way totally beyond repair it would be worth that for parts. The unit is certainly "built", it weighs around 45 lbs/20 kg!

I'd already spotted the four 1940's era RCA 'metal' 6L6 beam tetrodes, obviously in the output part of the circuit. Closer inspection revealed that the two output transformers were not identical, one of the originals had failed (or been damaged by another fault) and so had been replaced. The dropper resistor in the power suppy also had an open circuit section that had been repaired by the addition of extra power resistors, soldered onto the original tags.

Overall the repair was reasonable mechanically but I decided to fully check the wiring out. I found two schematics online for this receiver: here and here. A quick look at the schematic revealed that the output tubes should have been 7591A valves, and not the 6L6 valves fitted. Although both are beam tetrodes the pinout for the two types is not identical and in fact the characteristics are not particularly similar either!

One of the WW2 vintage RCA 'metal' 6L6 beam tetrodes

I decided to ensure that the unit worked, or get the unit working, using the 6L6s before seeing if I could (or wanted to) buy some 7591As. The octal bases had already been rewired to accommodate the 6L6s but the wiring was not easy to follow and I did not particularly like some of the soldering - call me fussy :)

I spent a few hours satisfying myself that the unit would not come to harm when I plugged it in, making good various connections and substituting some different coloured wire in places to make it easier to see what was going on. I added (un-bypassed) 3.3Ω resistors in each of the 6L6 cathode connections so that I could monitor the current in each valve. The bias supply circuitry has a potentiometer for each valve to allow the standing current to be individually adjusted.

I replaced the original 2 core mains lead with a 3 core lead (thus earthing the chassis), changed the unit's original back panel 4A fuse with a 2A fuse and disconnected the two US style 2 pin power sockets on the back of the unit. These would have otherwise been live with 240 V mains power when the unit was on. These features suggest a unit primarily made for the US & Japanese markets but also imported (without change :) into the UK. The transformer operating voltage is selected using a four position plug providing 110, 117, 220 and 240 volt operation.

top view with 7591As fitted

One of my friends was surprised I was so cautious before I powered the unit up for the first time but I didn't want to damage it with a silly problem I could have foreseen with a bit more effort. I was soon glad of this approach as it turned out although I'm not normally quite so organised ...

When I was ready to power the unit up I connected the Sansui's mains input to the output of my variac via a multimeter set to read AC current. This way I could slowly bring up the input voltage and check the unit for signs of excess current and heating of components etc. For this sort of exercise your nose is quite valuable as it will tell you when things are "smelling" hot.

As I slowly increased the voltage the valve heaters began to glow slightly, and the input current rose quite rapidly as the HT circuit became loaded. I didn't try it for more than a few seconds but at about 200 V input the unit was taking more than 700 mA - definitely too much.

An inspection of the 7591A and 6L6 data (obtained online) showed that the 7591A was designed for quite a high screen grid voltage, perhaps 450 volts whereas the 6L6 (and I don't mean the 6L6GC) really ought to have around 250 volts on the screen. The Sansui circuit diagram showed a nominal 420 V screen voltage confirming the 7591A data. This high screen voltage was causing excessive current to be taken by the 6L6 output tubes, even at the reduced supply voltage obtained from the variac.

classical voltage doubler power supply

In the Sansui the HT voltage is generated by a 180 V RMS secondary winding followed by the classic "full wave" voltage doubler circuit. This uses the two alternate halves of the mains cycle to charge two capacitors wired in series providing (up to) twice the peak voltage of the input, in this case around 520V DC. Lower voltages used in this set are derived from a multi stage resistive divider, (often known as a dropper) with decoupling capacitors spread down the line to minimise the hum in the small signal stages.

By a simple modification I was able to connect the 6L6 screen supply to the mid point of the voltage doubler capacitors providing them 260V - just about ideal. I powered the unit up again and the input current was much more sensible, the exact figure escapes me. With a CD player connected to the AUX inputs and speakers hitched up I was in business!

The unit has a front panel speaker switch which will connect an 8Ω resistive load to each of the output transformers and disconnect the ground connection to the speaker connector when switched to the "off" position. During the initial debug I used these internal loads rather than endure too many pops and bangs from real speakers as I probed round the rest of the circuit. The schematics show connections for 4,8 and 16Ω speakers but my unit supports only 4 or 8Ω speakers, selectable via a back panel switch rather than by changing connections. The repalcement transformer only has an 8Ω output and so I wired the impedance selection switch on this channel to select this in both positions. My speakers are a nominal 8Ω anyway.

Another switch allows the phase of one channel to be reversed so if you wired up one of the speakers the opposite polarity from the other you can fix it easily. Because the circuit was not the same as the online schematics in this area it took a little while to work out what was going on :) If anyone is interested I could redraw this portion of the schematic and put it online or email it.

underside view, it's quite busy

The unit uses negative feedback from the 16Ω transformer taping back to the cathode of the 6AN8 driver pentode. Because the replacement transformer only has an 8Ω winding I had to use this for the feedback and so altered the value of the series feedback resistor on that channel accordingly (to try and keep the two channel's gain as similar as possible).

Like all things related to tone control the Sansui is complex and a "presence" switch optionally connects in an extra resistor and capacitor into this feedback loop to tailor the frequency response. I decided not to bother trying to correct this configuration also. The two channels do not sound identical anyway but because of the individual potentiometers on the bass and treble controls it's possible to adjust the amp to sound similar both channels. This may not sound like a real "hi-fi" approach but it's not a perfect piece of gear so there we go ... If anyway has a spare output transformer then perhaps they can drop me a line :)

I've no idea what the official standing current figure is for this set but I decided around 10 mA would be OK for the 6L6s, giving an anode dissipation (no signal) of approximately 5 watts per tube. I found that the four 6L6s were pretty evenly matched, at least for the standing current, and ended up with three grids set to -26 volts and one at -24.

A little mains hum was evident at the output with this configuration and I replaced the (non original and leaking) 50 μF HT smoothing capacitors with some good 330 μF units I had to hand; the circuit diagrams show 200 μF capacitors and so the 50 μF units were much too small. This change fixed the hum problem.

front view with the 7591As fitted

With the now much larger smoothing capacitors I decided to replace the two original rectifier diodes with new 3A 1kV devices as a precaution. Blown semiconductor rectifiers in power supplies can cause the total destruction of the transformer very quickly because they often fail short.

After some further checks the unit was "deployed" into the living room and used as a part of our otherwise aged but currently all solid state hi-fi setup. As this point there was still no lid for the unit and to keep hands away from the high voltage points at the back of the unit I put it in a wooden cabinet we have whose dimensions made it hard to get one's hand in.

A short period of use ended when I noticed a moderately loud hum in one channel and also some distortion as well. A lengthy period of debugging back in the office ensued. Eventually I found that the hum was caused by one of the 6AN8 triode/pentode driver valves which presumably had developed and intermittent cathode/heater path. Someone more experienced with valve audio equipment would have found this much quicker I think although I learned something in the process and so it was time well spent.

The distortion, coincidently also on the same channel, turned out to be caused by a slightly leaky coupling capacitor in the tone control stage upsetting the biasing. The tone control circuitry uses a 12AX7A double triode for each channel together with a complex arrangement of RC filtering. Minimalist (tube) audio fans would not be happy with this circuit I think! Purist tube audio fans would probably not be keen on this circuit in any case as the preamplifier circuit that feeds the tone control circuit uses transistors :(

Russian made 7591A audio tube

As a precaution I replaced 2 coupling capacitors in each channel and the distortion disappeared. A check of the DC voltages indicated the two channels were back in step again.

At this point more listening tests indicated the unit was working well enough to use as a part of our regular hi-fi system. The 7591A has a a greater power rating, a slightly lower heater power and lower distortion than the 6L6 and so I decided to fit these to the unit. I bought four 7591As from Watford Valves.

I rewired the octal bases to suit the 7591As, double checked and then powered up, no variac this time :) I had already arranged the tube's bias to be set for minimum current as a precaution and after warming the unit up set the tubes for about 15 mA each, no signal. I've readjusted the bias after some hours of use because the currents had changed slightly from the 15 mA I'd originally set.

With unit connected up in the living room again I can honestly say it sounds absolutely great. I've not seriously investigated the radio section but it appears to work fine with plenty of sensitivity on both the AM and FM bands and successful FM stereo de-multiplex. The stereo indicator is a neon, rather than an incandescent bulb or LED.

One of my well informed audio friends has suggested I could, or even should, use a higher standing current in the interests of fidelity but to my ears it sounds nice as it is. Because it would run hotter with more current I've left it at 15 mA per tube for now. If someone with better ears than me heard it they'd probably suggest the same increase I guess, we'll see.

rear view showing the connections and the new case top

I made a top cover for it so it was safe for all to use. This cover was not as easy to make as I'd hoped because replacement transformer is taller that the original type and so the top surface of the new cover has to be proud of the top of the front panel. I made a folded back section to allow the top cover to sit squarely on the top portion of the dial cover which sits behind (and lower) than the front panel. In the end I think it looks ok and allows the glow from the tubes inside to be seen as well because I used perforated sheet. I hoped the perforations would help with the cooling although the region over the right channel o/p tubes (located on the l/h side of the chassis) and dropper resistors still gets pretty warm.

In summary I'm most pleased I was able to get this receiver working properly. It's certainly not a perfect specimen but is works well enough to be part of the household and not just a curio. It's actually a pretty feature rich piece of equipment and has integrated into my sound system with ease. A little online research suggested that my model was likely to have been made around 1968 and so it's 36 years old this year!

Recently I found the "The Best of SANSUI" website located here.

back to top

1920's style one tube receiver

A few years back I bought a small 1920's style radio at a vintage radio rally. I'd always thought items of this age were likely to be very expensive but this one cost me £20 (about $30), not unreasonable I thought.

VR21 regen, front view

The unit was scruffy but did use original 1920's components. One thought I had was that it had been constructed much more recently than its components suggested although the level of grime on the coils indicated that they at least had been wound a long time ago. Some of the soldering was recent; I'm not sure if the last owner thought that it worked well or not but it appears he/she made good some of the connections.

TThe unit was a single valve regenerative short wave receiver covering very roughly 5 to 15 MHz. In total, 1 fixed and 3 variable capacitors are used together with 1 resistor, 1 switch, two coils (on the same former) and of course the triode valve. The antenna was connected via a series variable capacitor and crocodile clip which taped into the tuned circuit.

The valve was a "VR21" which is a directly heated triode with a 2 volt filament. The early radios used a single lead acid accumulator cell to power the filament(s), the High Tension being provided by large dry cells. Periodically the owner would take the exhausted accumulator to a local garage which had facilities for recharging it. The dry cells would last longer as the current consumption from the HT line was generally quite low.

Later, by perhaps the early or mid 1930's, domestic electrical supply became more commonplace and radios evolved into much larger and more sophisticated units that were powered from the mains supply. Generally, these mains powered radios were of the super-heterodyne type, rather than the regenerative and "tuned radio frequency" types that typified the early battery operated valve radio sets.

VR21 regen, top view

I verified the unit was basically working and made some sensitivity measurements using a signal generator. I did not have a suitable 2 volt power supply and rather than use a resistor from a higher voltage supply I decided to use a 1.5 volt 'D' size dry cell. This only made the filament glow a dull red, just visible in a darkened room, but the unit would operate with this heater voltage.

I used a 50 volt DC power supply to provide the HT and the receiver took around 1 mA at this voltage. The unit would regenerate at 7 MHz but as I decreased the frequency the unit dropped out of regeneration, even with the control set to the maximum.

No audio output transformer was included and so I simply connected a pair of old high impedance headphones in series with the anode circuit. Using these headphones I was able to obtain the following results for receiver sensitivity at 7 MHz. I measured my "Minimum Discernable Signal" level by ear, it was the lowest level that I could just hear a CW tone. On this basis the unit was too deaf for conventional ham radio use.

HT voltage:   50 V
HT current:   1 mA
LT voltage:   1.5 V (should be 2 V)
LT current:   85 mA (would be 100 mA at 2V)
Frequency:    7000 kHz
Mode:         CW
Output:       Genuine 1920's headphones, in series with HT and anode RFC. 
MDS:          -90 dBm or roughly S7 based on S9 = 50 uV and 6 dB per S point.
Min usable:   about -80 dBm or pretty nearly S9.
For reference -90 dBm is 1 pico watt or about 14 uV RMS into 50Ω.

In an attempt to get the unit to regenerate below 7 MHz I tried a higher supply voltage as well as a 2 volt heater voltage but neither of these made any difference and didn't noticeably increase the sensitivity either. To extend the regeneration down to the lowest frequency covered by the main tuning control the feedback winding would have required modification or a fixed capacitor added in parallel with the regeneration control.

I did try using an audio transformer and sensitive (but low impedance) modern headphones but they were not as good as the old 'phones although they were a lot more comfortable! To put the sensitivity figure into context my twin double-triode regenerative 80m receiver has an MDS of better than -110 dBm when driving a loud speaker but this set does employ 4 triodes rather than 1.

The rheostat visible in some pictures is a 350Ω unit and despite appearing to be intact was open circuit at one end and had only one circuit connection to it as a whole and so had no effect on the radio's operation. Although the wiper contact still worked (in relation to the one good end connection) I couldn't work out how it should be incorporated into the radio and so I left unconnected. I have often seen rheostats in early radio filament circuits but I think this particular one had too high a resistance to be used in this application.

VR21 regen, back view

I'm interested in CW (Morse) and Single Side Band reception on the Amateur radio bands and the reception of these modes requires the set to be oscillating both in order to demodulate the incoming signal and also to provide a very high gain from the otherwise low gain valve.

The addition of another stage of audio amplification and perhaps an RF stage to help isolate the detector from the antenna will make a massive difference to a single stage regen receiver. In this instance there was no room on the chassis and I would have ended up with a "from scratch" project. I have too many projects as it is and I don't have any sort of stock of vintage valves and components either.

Some people have successfully used similar sets to make contacts but generally use at least one more amplifying stage. I decided it was not for me and passed it onto a friend. With a little tweaking the unit would make a reasonable broadcast receiver for the patient vintage enthusiast.

back to index