Bonsai SE Amps: 6BQ5 & 6V6 in Triode

It's been an unusually bitter cold winter. Typically, I could build a chassis or do simple woodworking in the garage but the frigid temperatures this season were prohibitive. So I postponed a stereo amp project and decided to revisit the 5687 and WE396A. These are tubes which have a loyal following but failed to excite me musically in the past.

Top: pullout from a Silvertone RTR 
Bottom: stock Magnavox 8601-20

The challenge was to find out if either tube could subjectively improve the performance of my two pre-existing amplifiers equipped with the tiny output transformers pictured above, especially at the frequency extremes.

!!!WARNING!!! 

The voltages in this circuit are potentially lethal! Build at your own risk!

5687 + 6BQ5

I featured the triode'd Magnavox 8601 amp last year with a 6CG7 driver stage. I'm using it for this experiment because the power transformer is quite hefty and can support the 1A @ 6.3V heater requirement of the 5687. As shown below, the filament winding with brown leads light up the 6CA4 rectifier tube while the green leads heat up the amplifier tubes.

 I've been very familiar with the 5687 since the 90s and even prototyped an SE2A3 amp with an R-C coupled driver stage. This tube has a lot more low frequency grunt than the 6SN7 and touted to be just as linear. However, to my ears, it could never match the organic midrange of the 6SN7 family of tubes (37, 76, 6P5, 6J5, 6CG7, etc.).

100 Hz, 1kHz and 10kHz square waves

5687>6BQ5 + Magnavox output transformer 

The superior specs of the 5687 wasn't significantly reflected in the above square waves compared to those produced by the 6CG7 driven 8601 with stock output transformers (click here).

Sensitivity = 0.38 V rms
Power output = 2.12 V rms into 4Ω 
or 1.1 W

This amp produced a very clean 1 kHz sine wave until the bottom phase flattens at clipping point.  

I listened to this amp extensively via 12" Altec 414Zs + horns, Altec 755As and 755Cs in Silbatone boxes. For a proof of concept video, let's transfer to our living room with the amp  driving Altec 405As.😊

With the 5687 driver stage, the Magnavox amp produced deeper, punchier and articulate bass. However, there was a hardness in the upper midrange, which I was only able to mitigate by reverting back to the original Magnavox output transformers. Only then was I able to appreciate the nicely focused midrange. Overall, I yearned for greater harmonic richness and air. But I was quite surprised that the 5687 made this Lilliputian amp sound bigger and more powerful than it actually is!

WE396A + 6V6

Keen-eyed readers will note that I featured this chassis as a PSE 6SN7 headphone amplifier during lockdown. Since I never became a dyed-in-the-wool headphone aficionado, this amp has spent more time in the closet. 

I played around with the WE396A/2C51/5670 while I was researching on how to improve my digital audio setup. Plugging a Western Electric 396A into the output buffer section of my Xiang Sheng DAC 01a transformed its sound. OTOH, I was quite disappointed when I tried the WE396A in a line stage ala Bruce Berman's 76 as well as when I grafted it at the output stage of my nosTDA1541tube DAC.

 

 I pulled the headphone amp from storage and rewired it as an SE 6V6 triode connected amp with a WE396A front end and installed the surplus pair of Silvertone output transformers. The feedback loop from the 6V6 plate to the WE396A cathode is crucial to this circuit (marked in red). I'll talk more about it below.

100 Hz, 1kHz and 10kHz square waves
WE396A>6V6T + Silvertone OPTs

Again, despite the superior specs of the WE396A/2C51/5670 versus the 6CG7, the above square waves are virtually indistinguishable from the square wave performance of the amp when a 6CG7 was in the driver stage position.

I've established in the previous 8601 upload that the Silvertone output transformers are technically superior to the stock Magnavox units. Perhaps what we're seeing here is mainly the behavior of the output transformer rather than the driver tube/circuit.

Sensitivity = 0.45V rms
Power output = 2.02V rms into 4Ω or 1W

The WE396A has the highest amplification factor (mu=35, 5687=17 and 6CG7=20) in this lot, but the amp has lower sensitivity due to the local feedback loop marked in red in the schematic.  1W output per channel is available but pushing the 1kHz sine wave beyond clipping severely distorts the wave form compared to the 5687. 

This amp produces deeper and more articulate bass notes compared to the 6CG7 although it's not quite as impressive as the low frequency performance of the 5687. The midrange is definitely more three dimensional with richer harmonics, if not quite as organic as the 6CG7. The top end is expansive with lots of airy overtones!

Now if the plate to cathode feedback loop is removed, all those virtues will disappear and the sound turns into mush. I got this idea from studying Western Electric preamp and amp schematics.

Simple Triode 6V6

After all is said and done, I still believe my Simple Triode 6V6 amp is the closest I got to emulating the sound of a true SE-DHT amp driving reasonably sensitive speakers. All you need is a set of irons pulled out of an out of commission Akai/Roberts or Sony tube RTR deck. IME, the output transformers from these equipment have better bandwidth and offer better sound quality than those from Magnavox and Silvertone.

Gotta get back to the WE131A/133A homage amp on my work bench. I'm excited to hear this!😊