Tuesday, October 28, 2008

Building a Single Ended EL156 Tube amplifier

This post is from my original Jenison Audio website which is now sadly extinct. If you're interested in making your own amplifiers and have never worked with high voltage before, be sure to follow correct safety procedures. I am not liable for your death or dismemberment.


Introduction: The articles you read in the stereo rags usually start with some clever comparison between the author's Porsche, gardening, or the different flavor of wines from Tuscany. Well, I'll skip all that and jump right to the point - I've built a few low powered SE amplifiers in my time using the 2A3 and 300B, but I've always found their real world performance disappointing. Even with fairly efficient speakers, I could hear the output clip when listening to some solo acoustic guitar! I like listening at levels that attempt to give some sense of the actual event. For example, throw on some AC/DC and the 300B amplifier would quickly fall apart. I'm not ready to invest in some mojo-efficient horn speakers yet and since my music tastes vary, I needed a single-ended amplifier that could at least drive some fairly efficient speakers, pump out at least 20Ws (which seems to be the bare minimum for real world amplification) and have some bass extension. This limited me to three choices - parallel 6550s/EL34s/KT88s/6L6GCs, a single 845/211s, or a bigger pentode.

Paralleling tubes was out since the extra cost, matching and fiddly bias adjustments would be a headache. The 211/845 was out because I have a morbid fear of working on anything over 600Vs. High voltage has a way of playing tricks and jumping if you're not careful - there's a reason to use ceramic sockets. So I decided on the 50W plate dissipation Shuguang EL156. I picked up a quad of these before the value of the dollar dropped like a rock and was impressed by their construction and massive plate. I also really like Ultralinear operation since it strikes a decent balance between the power of pentode and the delicacy of triode.

A quick word on the EL156 - this was originally a 10-pin European socketed pentode designed by Telefunken to be used in cutter amplifiers and has a big 50W plate dissipation with a 450V/8W screen. The new Shuguang version uses an octal socket and uses the same design/layout as the original Telefunken. Pin out is the same as an EL34 - meaning it is a true pentode and you will have to tie pins 1&8 together. Rumor is that Telefunken even had this tube made in China towards the last days of major tube production and even provided the cathode emulsion. The EL156 is one mighty tube and takes 6.3V/1.9A to light up, however it doesn't need big plate voltage like the 211/845 tubes. I grew interested in this tube as a replacement for the 6550/KT88s in my Dynaco Mark IIIs - but had problems getting it to bias without modifying the stock circuit. I was however convinced by my experiment that the EL156 deserved some further attention. Specs are available here. These tubes are available on Ebay or from Super TNT.


Note - I built these as monoblocks, so two will be required for stereo. eg - each schematic is for one channel. Total cost was just over $1000.


Schematics:



Power Supply is very basic - started off with a choke input power supply that was 'tuned' with a 1.5uF capacitor to get the voltage I needed. Even with the slow-warm rectification of the 5AR4, I would suggest using 600V (or two 350V in series) capacitors as the voltage does initially rise above 500V before settling down to ~475V as the EL156 filament takes some time to come on line. In my case I used two 350V/100uF capacitors in series where each capacitor had a 470k/2W resistor in parallel.

I got away with AC on the filaments of the EF86 too - careful lead dressing plus the negative feedback helped in that regard. Note with the AC from my wall, filament voltage was a touch high a 6.7VAC - I added two .10 ohm dropping resistors to get this more in line with the standard 6.3VAC. Hammond transformers are known to run a little high.


Amplifier circuit is simple. You may have seen this same type of input stage used on several other amplifiers - a basic EF86 pentode RC coupled to the next stage. You could also try a 6SJ7/6J7/EF37/etc in this position. Pentode drivers are not favored by many people, but I like them for their detail and snap - plus it only requires one stage to get the desirable gain. Of course the minus is the higher output impedance which can be an issue when driving some tubes. Luckily the EL156 is an easy tube to drive and doesn't require the extra complexity of a cathode follower or mosfet to drive it. I used high quality but not crazy expensive parts - KOA carbon film resistors, Russian Teflon FT-3 coupling capacitors, old Allen Bradley carbon-composition grid-stoppers, PEC mono volume controls and teflon-coated wiring. Cathode bypasses are just Nichicon electrolytics. The output stage cathode has a potentiometer with one leg tied to the wiper. Using the current meter allows me to dial in bias for the output tube. You can play with different bias points to find the best mix of current/voltage. It should be noted that the voltages are approximate as the power supply will slightly increase the B+ if the current is dropped. The zener diode on the screen tap provides a bit of regulation and also drops the g2 voltage down to a safer region. I got this idea from Bill Kleronomos in his Sound Practices #4 article "The Classic Williamson 1993 Style" where he noted a decrease in THD when using zeners on the screens.

If you wish to use other output tubes than the EL156, for example the KT88 or 6550, you will have to increase the 150 ohm output tube cathode resistor or use a larger potentiometer in the cathode circuit. This is to get the right bias range and you may have to change the B+ via the first PS capacitor to your favored operating point. An optional 25K linear potentiometer with one leg tied to the wiper could be added before the 8.2k feedback resistor - this will allow you to adjust the amount of feedback, though amplifier sensitivity will change as the amount of feedback is varied.





Initial Listening Impressions: Listening was on a pair of budget KEF Q30 speakers, a 'Bride of Zen' preamplifier, and a RAKK DAC. The amps are extremely quick with a fast rise and settle time. I'm not sure if it is the Teflon capacitor I'm hearing, but I'm mighty impressed when listening to the Mobile Fidelity release of Bela Fleck. Fast acoustic music never sounded smeared or has the notes run together. It's like a really good solid-state amplifier but without any of that slight grit/glare you hear in the upper-mids. Instruments are well fleshed out and voice is very natural. Listening to a needledrop of Willy Nelson's Stardust, I heard lots of little details (reverb effect, drum brush, etc) that were just a touch more apparent. Maybe one of the better amps I've heard, I'm curious to see if it holds out on something more hi-def like my UREI loudspeakers. In a short word (and I hate using this word mind you), the amps appear to be very neutral. They don't sound like 'tube amps' in the classic sense, but they don't sound like solid-state either. Amazing considering the pedestrian Hammond iron, electrolytic power supply and overall budget.


Further Impressions: My UREI 813A speakers are still in storage so I gave these a listen with a pair of 90dB efficient 4 ohm PSB Stratus Bronze speakers. The amps are breaking in nicely - there is enough power to drive these speakers and there is very good bass definition and control too. If anything the massive Hammond iron can pass some bass. Very nice midrange that has a tight grip on the music, but yet I don't feel a lack of detail. The sound has some of the smoothness of a SE triode amplifier, but to my ears they have more speed and control. These amplifiers are now a permanent part of my system and will be useful for future bi-amping.




A small video of a single amp being tested with some R.L. Burnside.

1 comment:

AnnieO said...

...but I like comparisons to Porsches, wine, etc. It shows how arcane and yet sophisticated things can be...