A few months ago I started a thread asking for suggestions on a single ended amplifier build. I've decided to build the mullard 3-3, as I have the appropriate hardware for it, but my el34s are 6L6s in my valve box were looking at me alluringly and I couldn't resist. I do still intend to build the 3-3 as my next project.
This is my very first single ended build from scratch (aluminium plate and wood from B and Q). Its going to my sister in law (if it is allowed on the plane), as she asked me for one after hearing the armstrong 220 and mullard 5-10. It will be paired up with a Raspberry pi 3a + allo boss DAC pre amp for digital streaming which I will provide, and god knows what speakers which she has yet to purchase.
The build is not perfect and there is still some room for improvement. Not being particularly experienced, I botched out the layout slightly, which caused me to chase around for hum for a while. The position of the output transformers was determined using the headphone test, no issues there. The choke is however too close to the OPTs and the pre-amp valves - my mistake - and the el34s should have been closer to the OPTs. In any case, every day is a school day. I'll do better next time.
The iron was all bought from china - cheaply - and the idea was to surpass the chinese commercial offerings in the 300-400 price range in terms of performance, given they they use the same (or broadly similar) iron. The measurements for the Reisong a10 which are shown here:
https://www.audiosciencereview.com/f...-review.19972/
As you can see the Reisong measures horribly - 4.39% THD at 0.78w, with massive 60Hz and 120Hz spikes, and a lot of HF rolloff on the frequency response. Nevertheless, people seem to rate them really highly for some inexplicable reason.
some pictures of the completed build are shown.
my main build requirements were:
1. stability - especially since the load it will be driving is still unknown
2. input sensitivity of about 0.8vrms - although the DAC is capable of 2v, digital recording volume varies quite a bit. This can be "normalised" , but this process can degrade the audio quality
3. sufficient power to be able to drive potentially modern(-ish) speakers
My first circuit was a straightforward 6sn7 cascade driving an el34/6l6, with 18db of global negative feedback. This worked very well, with 0.3% THD at 1w, negligible hum and noise, a very flat frequency response and a good square wave. If I build this amp again in the future, for myself, this is what I will do (with a better layout and probably slightly less feedback). However, I struggled with lead and lag compensation and the amplifier went into oscillation with low-ish capacitive loads across the output. Given that I don't know what load this will be driving, and I don't want to risk blowing up any expensive speakers, I wasn't happy with the overall stability and ended up scrapping this circuit and starting again
As the EL34 is running in UL mode (although the transformer manufacturer does not specify the % of the SG tap!) it requires greater voltage at the grid to drive it c/w pentode mode. The el34 anode voltage is 340v, with a 300ohm cathode resistor (el34 being run quite conservatively), meaning 22-23v on the grid for max drive. The problem was finding an octal valve that could give this amount of drive with a <1v input signal and allow some headroom for feedback.
Google led me to the 6sl7 srpp circuit, which although optimised for a low impedance load drive has been used to successfully drive 300b triodes. It should be sufficient for an el34 in ultralinear (or triode mode)
LT spice calculated the voltage gain of the 6sl7 srpp at 50 using the parameters shown (see picture), meaning that this should allow for almost 10db feedback which is a 3x drop in input sensitivity (i think!). Use of feedback was deemed essential as without it the FR looked like the top of one of Kim Yung Un's missiles. Incidentally, the voltage gain of a 6sl7 in parallel was lower than that of the srpp. 135v on the cathode of the second section of the 6sl7 required elevating the heaters (virtual centre tapped with a humbucking pot), which I did (to 56v). I should have saved myself some grief and installed a noval socket instead, and used an ecc81, but then no plan survives first contact with the enemy.
The final circuit is shown below, along with performance parameters. The 100k grid leak input is actually a concentric volume control. As you can see, it far exceeds the Reisong A10 in terms of measured performance. Hum and noise is virtually non existent. Frequency response between 20-20,000Hz is very good given the cheapness of the OPTs. The only disappointing parameter is the THD (1.1% at 1W), which at least is almost all 2nd harmonic (see picture). This rises to 4.0% at full power, which is 6w per channel. el34 anode plate voltage is 340v and the transformer primary impedance is 3.5k. The % of the ultralinear tap remains unknown. Feedback is around 8db. The amplifier is now completely stable into pure capacitive loads up to 1uf which is the largest non electrolytic cap I've got, so it should be able to handle any speaker/electrostatic tweeter and any length of speaker cable (I hope). the 1kHZ square wave is good, and it sounds good on my mission speakers.
I mentioned before that I had an issue with hum initially. This was due to the choke (magnetically?) coupling with either the OPT or the input valve. It took me a while to figure this out as the hum only appeared once the valves warmed up. In any case, switching from a CLCRC to a CRCLC to reduce the voltage swing in the choke sorted this out.
Any comments and suggestions welcome. Whilst it is unlikely that I will be doing any further changes to this one unless I've done something horribly wrong, I still have a pair of 6l6s and will probably be going down this route again in future.
Note: all measurements done using REW and ARTA on a laptop via a 20db attenuator. i.e. the 50HZ spike at -110db is actually -90db. the 150 Hz spike if from the laptop.
I failed to mention that I'm not actually using 6sl7 valves but their russian copies. I have yet to run it alongside the push pull amps to see if I can hear a difference.
regards,
gabriel