13E1 Valve Bass or HiFi Power Amplifier

This is a stereo amp I made using 13E1 output valves. I built it initially thinking it would make a good bass amp project (I like true stereo chorus and Leslie effects, or the two channels can be bridged for double power in mono). The 13E1 valve is a very rugged industrial type low impedance beam tetrode that has been found to have good audio qualities along with its high power handling capabilities (three times that of the classic EL34 valve shown for scale resting on the right hand output transformer).

However at the moment it is set up as a pure Class A amp with no global feedback loop, and it sounds so good in my hifi that I’m not sure it will get to play bass much! It can take a CD input directly (it needs something like the F2B described elsewhere on this site in front to adequately amplify passive guitar pickup signal levels, or active pickups will work as direct input), so doesn’t need a preamp for hifi, which is maybe part of the reason for the excellent sound. But I think not using global feedback is probably the key to the magic here, though without the damping effect this provides it does need speakers that are an easy load to drive – I used BBC LS5/9 monitors. I’ve not hung out in hi-fi demonstration rooms much, but the airiness around the stereo imaging had me digging out long not listened to CDs, and the bass and treble qualities suit all from dub to folk and world. I don’t really listen to much classical music, but I tried Rodrigo’s Concierto de Aranjuez (first movement rocks in a way that always takes me by surprise, years ahead of its time), and with opera the vocals were, well, reliably opera-like…

The actual amp itself is not so heavy for a large valve amp, but its separate power supply chassis weighs nearly as much on its own as a whole Fender Twin Reverb, so in reality it’s probably not very practical as an instrument amp. It has 17 valves of 5 different and slightly exotic types so the amp’s first aid kit might be a bit unwieldy too. The weight issue and large power supply meant building it as a head was inconceivable, so it had to be designed for rack mounting (allowing for cooling space the amp takes up 6U, the power supply another 4U), but I still wouldn’t want to have to lug it around much.

Because of the weight, the fairly high voltages present, and significant heat dissipation issues, the construction of suitable chassis and ensuring safe and quiet electrical connection between them was the main challenge when building this amp, and it felt more like mechanical engineering than electronics at times. I guess the chassis are best described as functional rather than elegant, but of course when installed in 19″ racking only the front panels are visible.

As with all the best ships, below decks appears somewhat chaotic, but the fully point-to-point wiring on tag strips does give lead dress with direct clean signal paths without coupling, and there is no hum present at all when the amp is in place on top of its power supply.

Technical description

The circuits for each channel are identical, and somewhat similar to the old classic Williamson hi-fi design, employing 6J5 valves for the input stage, further gain stage, and direct coupled concertina phase invertor; and paralleled 6N7s as drivers for the push-pull 13E1 Ultralinear connected output. There are also active Baxandall treble (+/- 18dB @ 10kHz) and bass (+15/-13 dB @ 50Hz) tone control circuits for each channel after the input stage which employ local feedback from the second gain stage. With the global feedback loop from the speaker output to input stage cathodes disconnected, obviously there was initially way too much gain in the amp overall, but I found replacing the 6J5s with 6P5 valves was an easy way of reducing this without having to entirely re-work the whole circuit (they have about half the amplification of 6J5s).

The power supply is heavy because it actually contains a total of four transformers and two chokes to run the two HT circuits and six individual LT heater filament supply circuits (two of 26V needed for the 13E1s, and four of 6.3V for the rest). The output valves’ HT is provided by a pair of paralled GZ34 rectifiers feeding a large choke input filter; this supplies 550mA at 460V with good regulation. These rectifiers have UF4007 silicon diodes in series with the anodes to ensure PIV is not exceeded. A GZ32 for the small signal valves in to CLC filtering is adequate for the 125mA at 400V requirement, with a further CLC filter for each channel built in to the amp chassis itself, and further CR decoupling filters at each small signal stage. ¬†Even though the main HT supplies are shared, sound-wise channel separation is very good. The ac heater filament circuits however are separate for each channel, and the four for the small signal valves and drivers each have a hum balance pot with the centre taps elevated to one of the 13E1’s cathode potential (about 75V).

The valve rectifiers provide a soft start at turn on, so there is no stand by switch required; it takes about 20 seconds for the 13E1’s HT to come up, by which time the valves are fully warmed and able to conduct. Even so, I included a Zener clamp circuit to protect the main filter cap in case the amp is ever run with no load (ie if the output valves have been removed or the connecting umbilical cables are unplugged), as the voltage would then rise to about 750V. (Bleed resistors in the power supply do discharge the caps fully within a couple of minutes, but they don’t draw enough current by themselves to put the choke supply in to its proper regulation.)

The 13E1s are biased using non-inductive 20W cathode resistors (individually fully bypassed), putting the cathodes at about -75V relative to ground, with another 20V negative bias additionally provided at the grids by individually adjustable networks for each 13E1; bias test points in each cathode circuit are provided on the back panel of the amp. This is important for output transformer idle current balance, purchasing nicely matched quads of 13E1s being simply not an option! The 13E1s are run at 125mA quiescent current so their dissipation is around 55W each when idling. Even though this means they are significantly under-stressed considering they are rated for up to 90W, I was deliberately conservative with the current levels as I didn’t have detailed specs for the used power transformer and choke of unknown provenance that were used. Actually, in the end I just compared the power transformer physical core size to that from an Ampeg SVT amp; they were about the same so I figured that was good enough! In use neither component does get too hot or buzzes unacceptably.

The chassis was constructed using aluminium offcuts principally for ease of machining, for cheapness, and to minimise magnetic coupling from all the various transformers and chokes, but using this material has also significantly helped in dissipating the large amounts of heat generated by the 13E1s, their power transformer and chassis-mounted cathode resistors. The 4mm thick front and back panels do radiate out of the amp a lot of the heat transferred to them, nevertheless, forced air convection using a fan is necessary when rack mounted.

I didn’t use any specifically audiophile grade parts anywhere in this amp, although I did avoid using any really trashy capacitors (the final filtering sections of the small signal HT supply use large MKP caps rather than electros); and the input stage and 13E1 valve sockets are teflon rather than ceramic, to reduce stray capacitance between pins.

The amp did work pretty much out of the blocks; other than the excessive gain, there were only two debugging issues that arose during testing. I had one power supply voltage dropping resistor burn (resulting from an untrimmed wire connection shorting to ground). The other was an interesting but beware aspect: paralleling two GZ34 rectifiers to cope with the required HT current isn’t complicated (eg Mesa/Boogie do it with the Triple Rectifier circuits), but in this case their heater filaments also had to be paralleled across two separate 6.3V windings on the power transformer, with eight high current (3 amp) diodes used to reduce the voltage down to the 5V the rectifiers need. Paralleling transformer windings in this way must have the magnetic polarities for correct phase observed, otherwise seriously large (maybe 100 amps) currents can flow. But as one can’t tell from inspection, without a proper detailed transformer internal winding spec it’s down to pot luck as to which way round the connections are. It was fifty-fifty, and I chose wrong, but the heat generated by that large current for the approx two seconds the power was on while I was checking resulted in the circuit having the most perfect, strongest re-flowed solder joints I will ever see. Fortunately the voltage dropping diodes used are rated for 150 amp surge…

Both of these issues failed to blow any of the internal conservatively-rated fuses, or the 2 amp mains plug fuse – which just goes to prove the old adage that you should assume fuses are there to protect the supply wiring, not the equipment itself!