Mar 23, 2016

EL84 Push Pull amplifier, the little giant

The EL84 is NOT a DHT, but still a fascinating tube for many triodes lovers (I belong to this group) who recognize its sonic qualities.
Great manufacturers like Dynaco, Scott, Leak, Loyez just to name a few, sold thousands of amplifiers using this cute little pentode and for many serious listeners of that time these amps were by far superior for music reproduction to their more powerfull concurrents using EL34, 6550 or KT88.
Yes this tube is very involving, it as the speed ( that usually lacks to the bigger tetrodes and pentodes) the naturalness and ease so many people do like in the very best triodes. It reminds me the 10, the E130 or the RS242, it makes music living like and it's power limitation is it's only drawback. Remember that in the 60's there were not so many speakers with 97/99dB sensitivity to perfectly match a 10 watt amplifier.

In the past 20 years I made half a dozen amplifiers using this tube, always with great pleasure, and that's why I decided to make a post about it.

Also because I always keep an EL84 amp to drive my beloved Wharfedale Super 12 RS/DD. Magnificent full range speakers especially on open baffle.

The amp I will talk about is not a design of my own, it belongs to Harman Kardon. I just made some modifications for safer use and better sound reproduction (I built one exactly like the original for comparison).
Why the HK amplifier ? I found during all these passed years making electronic stuff that simplicity is usually the best way to fidelity and the HK20 design is really simple and very smart.

It operates with a single 6CG7 as driver and phase splitter, two EL84 and a rectifier.
You can't do more simple and when you add the qualities of the 6SN7 family tubes you are right on the way for a good sounding amp.

HK 20 original schematic

As you can see, very simple with some tricks to prevent the use of a decoupling cap in the first stage. V1a bias is taken through R10 in the cathode of the phase inverter V1b. The current of the two sections flows in this resistor, the result is a smaller value that it would have been if directly grounded the usual way. This helps to lower the effect of degenerative feedback thus giving a higher gain to the first stage. Elegant, efficient and easy to calculate if you want to change the tube (I tried the E80CC instead but prefered the 6CG7 after a long listening comparison). For a perfect cathodyne balance a single R11 resistor in parallel with R8_R9_R10 will give the exact value of R7.

To calculate R11 with (R8 + R9 + R10) = Req   use formula   R11 = Rreq X R7/Req - R7

My own HK20

I was not comfortable with EL84 operating voltages.
370V on anode and 350V on screen make the tube work very hot and very hard. I am not sure the usual EL84 will last very long under such a fire.
Even the professional and super strong Mazda 7320 reach their limits, furthermore when bread boarding an amp for tests I had some hum that I could not cancel. So I decided to modify the whole amplifier supply and to pay great attention to time constants, I also preferred the EL84 UL to the pentode mode, I reduced NFB to get decent THD (1,5% @ 10W, mainly third) and a more mellow sound. I suppressed R1 potentiometer (always a soucre of trouble) and replaced it by a voltage divider to match the output of my line preamp.
Now with such a supply and mods the amp is dead silent, punchy, open with a wide soundstage.
Very detailed heights, lush mids, rock solid bass make this amp an all purpose companion.

Power supply revisited.

Inside this little giant, AB resistors and Aerovox V161 caps give a vintage look without compromising sound qualities.
Reliable Siemens capacitors to smooth out the ripple, way better than the usual electrolytics (MKV and MP J/S are my preferred) and Hashimoto OPT.

Philips Holland killer tubes, Mazda GZ32 and shielded 6CG7

The extra socket was for a buffer stage intended to be used with a 600 ohm symmetrical input transformer. I finally did not populate.

Personal EL84 review

I am lucky enough to have different brands in stock to make a comparison.
I wont write a lot just give my impressions, for an extensive test read  Vacuum Tube Valley issue 8

Mullard soft and round sound, not my taste.
Philips RTC the very best with this amp, very well balanced.
Philips ( Amperex ) Holland top of the line with Philips RTC.
Mazda 7320 strong and powerfull sound, can be tiring.

TFK the worst nothing else to say.
Tungsram excellent performer, lacks a bit of punch in low end, the best on voices.
Russian Reflector 6P14P/K (and only this suffix, means vibrations proof. NOT EB or ER) a real surprise, very good sounding, on par with Philips RTC.

Some great CD's to listen with

Mar 9, 2016

VT25, VT25A Push Pull amplifier Part 2

All Tango amp in white slim chassis...

and fed with good looking and good sounding tubes.

Sylvania VT25 bright light

Magic eye for fun

Philips 5R4GYS is a perfect companion for the Sylvania VT25

Sylvania VT25A soft glow...


The amp puts out a reasonable 5.8 Watt before clipping ( 6.5 Vrms/8 ohm) and THD @ 1000Hz do not exceed 2.60% , not that bad. Second and third harmonics are barely visible.

With VT25A I got 6.9 Watt and same distortion figure. Nice little amp.


I have used this amp for years and it is an all around music maker with lots of punch and refinement. Compared to my MC1/60 I would say the VT25 PP is at its best on vocals, small jazz and classical formations, guitar and piano. It is an amp for people seeking for naturalness and accuracy in music transcription.

Some CD's and LP's that I like to listen

VT25, VT25A Push Pull amplifier Part 1

The 10 is a very old triode and probably one of the first mass production power tube. It's linearity and sound qualities brought this tube a favourite among many DIY's. Dozens of amp and preamp using this tube and it's relatives (10Y, 801A, VT25) can be found on the net. The only and main drawback is it's flea power.
Long time ago I decided to build an amp that could drive speakers with 95/98 dB efficiency and the logical solution was push pull operation.
In that way I began to modelize an AB1 operation push pull to reach 2 goals, higher output power (about 4/5 watt) and better amplitude distortion figure ( Push pull operation helps to compensate characteristics curvature). For that purpose I had on hand a pair of Tango FX50/16G, one of the finest transformers of Mr Hirata company.

Sylvania VT25 in "modern" ST shape

To get the very best of these tubes a good driver with high output voltage and low distortion is mandatory. Furthermore the perfect way to split signal in two is by interstage transformer. The main difficulty is to find one that accepts some primary current with large bandwidth, I finally selected the Tango NC14 for its very good reputation and its current handling capability.

To provide enough voltage a two stage driver was necessary. After several tests with different dual triodes my choice remained on the 5687 tube.
However due to driver limitations and NC14 step down voltage ratio (1/0,7) I can't go very deep in AB. Operating point is defined by plate voltage ( 415V ) and cathode resistor (3,3 Kohm) giving a steady plate current of 12 mA @ Vg -38V. The load (8 Kohm) of each tube give me the maximum plate voltage swing, here 275V peak.

5687 operating points have to be carefully chosen to give a large voltage with minimum distortion. It is especially critical for the 2nd stage as anode current must remain in the 15/18mA to keep good bandwidth in low frequencies from the NC14. To drive the VT25 full power we need to swing grid ± 38V peak, it means ± 55 V anode peak for the 5687 due to transformer voltage ratio.

Amplifier circuit

Power supply circuit
Two independent 7.5V supplies for VT25/VT25A tubes (just one shown on schematic)

Construction details

To check power tubes bias voltage I used a 6AF6 magic eye, although not accurate it gives an original look to the amplifier.

There is a bonus, I can use VT25A tubes without any change in amplifier circuit.

Sylvania VT25A big plate triode and oxide coated filament. Power rating seems to be the same but gives a bit more output mainly because of its lower internal resistance.