EL84 Amp completed

The EL84 amp is complete and has been delivered to its new owner, who has compared it favourably to his 60wpc Harmon Kardon Solid State amplifier.

The aesthetics of this one are much more favourable than the previous build. In testament to this, the new owner reports a high Spousal Acceptance Factor 🙂 

Looking good next to the turntable

In the end the chassis was about 5mm too narrow to fit the power transformer and the output transformers across the back in the usual configuration. So it had to be non-symmetrical

Making a virtue out of necessity: The non-symmetrical theme carried through from the transformers to the placement of the valves, and the controls on the front panel.

The front three small-signal valves form the RIAA Phono stage, the rear two are the line-level voltage gain and phase splitter.
The Bias test points sit between the EL84 output valves, with recessed trimmers to adjust, and test points to measure the voltage across the 10Ohm cathode shunt.

The Slovak-made JJ EL84 output valves on this one have quite a pleasing amount of light-leakage from the filaments and cathodes. Unlike the Russian Electro-Harmonix small-signal valves which are hard to see any filament glow from at all

Inside, the amp is crowded. Point to point wiring inside a tight working space.
Polyethylene Film capacitors are used for inter-stage coupling, and also in the RIAA stage, which as at top left in this photo. The power supply board sits under the output transformers, The power supply board has my usual 555 timer-based circuit to delay the B+ turn-on be 30sec giving the valves plenty of time to warm up first. By use of this circuit, combined with heater elevation for the small-signal valves, I can get away with avoiding the diode on the cathode follower in the RIAA stage, which is DC-coupled to the previous gain stage.

6 PCBs inside this case, including 5 home-made ones

Listening tests

The sound from this one is clean, detailed and very pleasing. The main amp stage is based on the well-regarded Fisher X100, with the 12AX7 gain and 12AU7 phase splitter, though this design runs that 12AU7 closer to its 5mA sweet spot for linearity from a 300V B+ than the Fisher does.
The output stage is EL84 in fixed bias ultralinear, biased to 8.4W quiescent dissipation (70% of rated maximum)
The RIAA stage works well. The Cathode follower is needed to drive the volume control which represents a 50K load across the input. The noise floor is low, hum is non-existent, and distortion does not occur even on the loudest passages. 

Lessons learned

During the construction of this amp several lessons were learned…

1) Hum was a constant problem

The amp was built backwards, with the output stages and transformers being wired up first, powered on to test, then the preceding stage, right back to the RIAA stage.
The 12AU7 Phase Splitter was putting a nasty hum into the output. After chasing that down and much testing with the oscilloscope etc, it was determined that the hum was on the anode but not the cathode. Many solutions to this common problem are available on the internet, in the end I opted for an additional level of decoupling in the power supply with a 10K resistor and 220µF capacitor, this fixed the problem. The lesson is that Phase Splitters have NO PSRR on the anode side. That power needs to have no trace of ripple on it

2) Grounding needs close attention in a RIAA stage

You can read as many books as you like but it’s only when you build an RIAA stage that you truly get to appreciate how to ground the incoming signal… and how not to. This one had a nasty hum which was coming in through the Earth side, it would only manifest when there was a source plugged into the phono input. If the jacks were empty, the stage did not hum. But plug any source in, the hum appeared … after bridging the input with a 1K2 resistor to simulate the cartridge, it was observed the hum was injecting into the live from the earth through the source. After moving all the signal Earth to a common point – which was the star earth off the first valve in the phono stage – suddenly it went dead quiet.

3) Take care with design and placement

There were a few too many near-misses with things fitting much tighter than planned, or almost overlapping other parts, 
There are also some minor changes I’ll be making to the design of my bias boards and power supply boards for the next build, which is not currently planned.
Next post… I’ll put up the as-implemented circuit schematic.

Leave a Reply

Your email address will not be published. Required fields are marked *