Yes it does by modern standards but it still sounded good and I don't recall any comments on thermal stability following publication. I did a home made PCB based on the design in the mag but did not use the preamp section.
This is a sim of it from a long time ago. All defaults models, click and run.
"You could do that with an MJE340."
Not in the UK with typical valve voltages at 240-300V. Maybe in America with 110V supplies.
I toyed with the idea of a BU108 (think it was that) because of the higher voltage rating. With a transformer output it has to withstand up to 700V.
Maybe a low power HV MOSFET would work today, but the need for replacing valves with transistors directly has pretty much gone. People either use actual valves or design tranny ones.
Not in the UK with typical valve voltages at 240-300V. Maybe in America with 110V supplies.
I toyed with the idea of a BU108 (think it was that) because of the higher voltage rating. With a transformer output it has to withstand up to 700V.
Maybe a low power HV MOSFET would work today, but the need for replacing valves with transistors directly has pretty much gone. People either use actual valves or design tranny ones.
If you were actually going to build it as an experiment you would have a transformer producing 110 to 170 volts DC regardless of what country you’re in. Not supposed to build line operated amps anymore, but people did back then. The MJE340 and TIP50 ran off the same supplies they used to use for SE 50C5’s, and made the same watts. Just under 2W at 110, about 3 at 170. Or less, if a higher Z primary is used. Couldn’t run 50C5s directly off rectified 240 or equivalent either. At least not without dropping it down to the 200V range. Tubelab George would have, but not in SE. Want to build something “bigger”, the 340 isn’t really big enough - and neither are many HOTs. A transistor that I’ve found recently thats actually good for the purpose is MJE18004 (BUL45). 200V or less supplies, of course. But the DC SOA is impressive, even among non-switching TO-220 types. Full power rating to 75 volts. That will eat similar size mosfets or IGBTs for breakfast in terms of linear power handling capability.
A new design could have any power supply voltage which was suitable, of course. I was looking to replace output valves (at the time) and used the available output transformer, rather than having to rebuild that for a different voltage. Some old 1940's and 1950's radios did run directly from mains with dropper resistors for the heaters, but other, better designs used transformers and still gave around 240V output.
In the main, BU108/208 types although handling ~700V or more could still only manage up to about 20mA for DC operation at ~250V in a transformer class A stage. And the gain was pathetic, needed driver transistor even at low currents.
Maybe the 2SC3153 could manage a little more - not sure if it is still current, ON semi bought out Sanyo and I can't see horizontal OP transistors (line output transistors here) being marketable, but, perhaps rebadged/marketed for SMPS.
I have a few MJE18004's - for SMPS application, but probably won't try them for audio. Hard to get proper GOSS iron for transformers these days, could do so a decade ago or so. Expensive if order from companies who do still build transformers.
In the main, BU108/208 types although handling ~700V or more could still only manage up to about 20mA for DC operation at ~250V in a transformer class A stage. And the gain was pathetic, needed driver transistor even at low currents.
Maybe the 2SC3153 could manage a little more - not sure if it is still current, ON semi bought out Sanyo and I can't see horizontal OP transistors (line output transistors here) being marketable, but, perhaps rebadged/marketed for SMPS.
I have a few MJE18004's - for SMPS application, but probably won't try them for audio. Hard to get proper GOSS iron for transformers these days, could do so a decade ago or so. Expensive if order from companies who do still build transformers.
The 3153 was in a Mitsubishi PLC SMPS, and a common failure.
When I checked it out, it was a fairly easily available part for TVs, and IIRC good for max. 900 V, bit more than the ones being discussed.
I changed to a new PLC in 2018, no need to keep track, it was a common HOT / SMPS part in TVs.
But now? CRTs are being scrapped, and I have no idea what transistors are used in LCDs...
When I checked it out, it was a fairly easily available part for TVs, and IIRC good for max. 900 V, bit more than the ones being discussed.
I changed to a new PLC in 2018, no need to keep track, it was a common HOT / SMPS part in TVs.
But now? CRTs are being scrapped, and I have no idea what transistors are used in LCDs...
I take it you‘ve never needed a 350 volt VAS before. “Darlingtoned” with something smaller, of course. No transformer needed.
I don’t even try to find transformer materials anymore (except magnet wire, which you can get from people who make wind turbines). It’s either buy or salvage. Salvage includes rewinding - but I limit that to power trafos.
I've not built anything needing voltages higher than about +/-60V in audio except when I made a 20W valve unit which ran at 425V - and needed a power transformer with a higher output voltage than the 240V going in. I suppose you could consider the ECC82 as a sort of high voltage VAS.
I no longer try to find transformer iron either now, but may still rework older ones should the need arise and I have something to hand.
I no longer try to find transformer iron either now, but may still rework older ones should the need arise and I have something to hand.
The 50C5 will eat 340 volts on the plate forever if the screen is held at 125 volts or so. A pair will make 20 watts in push pull AB1 under these conditions. They will make a couple watts in SE on 340 volts into a 5K OPT using a bunch of local feedback to get triode like curves. I ran them at about 30 mA for 10 watts of plate dissipation which is about twice the spec, but not an issue. There are however better tubes for that job like the octal 50L6 and all it's many relatives which have bigger plates. All were derived from the 6W6. The trick European tube for this use is the 45B5 / UL84 which is a 6CW5 with a 45 volt 100 mA heater.
I have been digging through boxes of parts that have not been opened in many years eliminating stuff that I will never use in what's left of my life. I found some Motorola BJT's from 2001 that are rated for use as "coil drivers" at up to 1000 volts. The term HOT is not used in the data sheet, but many of the breakdown type failure modes are discussed. The number is MJH16006A.
I also found a couple others that might work as 50C5 stand ins. Maybe I'll "test" a few.
When they used those 50C5’s in radios and record players, they got lazy (or cheap) and just ran the screen off the same B+ as the plate - which works on rectified 110, but not 220. You‘ve got to run that screen down to 150 or poof. And they didn’t have the UNSET technique back then. “Easier“ to use something else with s 300 volt screen rating on a 340 volt rail. The typical application called for a couple of watts and to basically last “forever”, while being cost effective. I had plenty of those old pieces and never had to replace the original 50C5’s or 50EH5’s. If someone built something like that today and sold it, the bean counters would have them taken out and shot for making it run for 20 years straight.
MJE340’s were used as 50C5 stand-ins, on 110-170 volt supplies, where a watt or two was required. I had table radios with those in them too. To duplicate what you did maxing a pair out, however, takes a bit more doing. Assuming you can get a 330 volt swing, you need 85 mA of current. Drive a resistive load and the instantaneous power dissipation would reach a peak half way up the rail at 42.5 mA, at just over 7 miserable watts. The average is much lower, so it sounds easy enough. But real loudspeakers can draw up to half the peak current at FULL voltage due to the elliptical load line (a 60 degree phase angle will do this - cosine 60 is 0.5). What’s required of the transistor to be safe is 42.5 mA at 340 volts. MJH16006 won’t cut it - despite the higher 25 C low voltage Pdiss rating. The switching SOA is square as it can be, but audio waveforms linger for much longer than 10 microseconds. The 2SC3317 fares much better. Tubes do not have that inherent limitation, as long as screen ratings are respected. And even if not, runaway takes more time than it does with a transistor. You may get a few hours or minutes running a 50C5 or sweep tube in triode at 300V before the current runs away, or even at 50 watts for a few seconds with a 120V screen. Second breakdown can happen in milliseconds. The MJE1800x is about the best I have found for DC and 10ms FBSOA. MY application is in the VAS of a transistor amplifier running off similar voltages. 20 mA of VAS current, and with no load you can get the full 350 volts of swing on it. Set the current limiting for the stage (during fault conditions at 175V) to 50 mA, and we’re talking about similar requirements. Not just any transistor will hold up. Of course, I’m stuck requiring a PNP that can do it too. Only one will - MJE5852. At least with NPN only you have a couple of choices, even if you also need 800V Vceo.
MJE340’s were used as 50C5 stand-ins, on 110-170 volt supplies, where a watt or two was required. I had table radios with those in them too. To duplicate what you did maxing a pair out, however, takes a bit more doing. Assuming you can get a 330 volt swing, you need 85 mA of current. Drive a resistive load and the instantaneous power dissipation would reach a peak half way up the rail at 42.5 mA, at just over 7 miserable watts. The average is much lower, so it sounds easy enough. But real loudspeakers can draw up to half the peak current at FULL voltage due to the elliptical load line (a 60 degree phase angle will do this - cosine 60 is 0.5). What’s required of the transistor to be safe is 42.5 mA at 340 volts. MJH16006 won’t cut it - despite the higher 25 C low voltage Pdiss rating. The switching SOA is square as it can be, but audio waveforms linger for much longer than 10 microseconds. The 2SC3317 fares much better. Tubes do not have that inherent limitation, as long as screen ratings are respected. And even if not, runaway takes more time than it does with a transistor. You may get a few hours or minutes running a 50C5 or sweep tube in triode at 300V before the current runs away, or even at 50 watts for a few seconds with a 120V screen. Second breakdown can happen in milliseconds. The MJE1800x is about the best I have found for DC and 10ms FBSOA. MY application is in the VAS of a transistor amplifier running off similar voltages. 20 mA of VAS current, and with no load you can get the full 350 volts of swing on it. Set the current limiting for the stage (during fault conditions at 175V) to 50 mA, and we’re talking about similar requirements. Not just any transistor will hold up. Of course, I’m stuck requiring a PNP that can do it too. Only one will - MJE5852. At least with NPN only you have a couple of choices, even if you also need 800V Vceo.