GOLDEN VOICE 2A3C
About a year ago, I ordered a matched pair of Golden Voice 2A3Cs from AliExpress priced a little below $100, shipped. Within two weeks of receipt, one tube started arcing. I sent the above TV7 test video to the seller.
About a year ago, I ordered a matched pair of Golden Voice 2A3Cs from AliExpress priced a little below $100, shipped. Within two weeks of receipt, one tube started arcing. I sent the above TV7 test video to the seller.
Over two decades ago, I ventured into single channel music reproduction to trace the roots of high fidelity. From that whole experience, I developed an aesthetic quest for a certain purity and honesty in sound, which runs parallel with my other hobby of black and white film photography! Sadly, I had to leave that particular mono set up in the attic in 2017.
During a recent visit to Steve and Ray at VAS Audio, my attention was caught by an amplifier I built in the early 2000s. This is one of the very few I released into the wild.
My dad would've turned 97 today. To celebrate his birthday, the video clip above features a recording of his favorite violinist, Nathan Milstein, playing Presto from Bach's Sonata No. 1 in G minor for unaccompanied violin. While he was doing post-graduate studies at The Juilliard School in the mid 50s, he was fortunate to have attended concerts of the great violin virtuosos of that era - Jascha Heifetz, David Oistrakh, Nathan Milstein, etc.
He would later recount that Nathan Milstein's performance of the Glazunov violin concerto was the most unforgettable! Even if he never got a chance to add this violin concerto to his repertoire, he conducted the Manila Chamber Orchestra with me playing the Glazunov violin concerto in 1987. The music of the slide clip above is the middle movement from our performance.
Before I bore everyone reminiscing about my dad and our life as musicians, let's get on with the subject at hand. The amp featured in this upload uses a directly heated triode power tube that was introduced around the time my dad was born.
![]() |
Canon P + Canon 50mm f1.8 LTM Foma 100 in Rodinal |
Inspired by my plate choke-loaded 76 line stage preamp adventures, I was curious how a similar driver circuit configuration would sound in an SE2A3 amp. So I pulled out my Radiotron Designer's Handbook, 3rd Edition to gain a better understanding of choke-capacitance coupling (aka plate choke loading) vis-à-vis resistance coupling.
Below are highlights of relevant information I gathered.
To celebrate the holiday season I've prepared a short video. Please note that given the dynamic compression and bandwidth limitations imposed on sound quality by YouTube, I upload videos for the visual aspect rather than their sonic merits.😊
Back in the late 90s, I dismissed the hi-mu/low-gm SRPP input/driver because I found that the circuit had a tendency to sound tiring over extended listening. However, my memorable encounter with the Morikawa SE2A3 amp owned by my buddy Jose was haunting. I had to revisit this circuit just to make sure I left no stones unturned.
Earlier this year, Steve of Angela Instruments shipped me the "shop mule" SRPP SE2A3 amp for a check up. After replacing the leaky coupling caps, the amp was as good as when I delivered it to the shop over two decades ago.
My hotter operating point = slightly more power compared to Prof. Morikawa's SRPP 2A3 amp.
![]() |
Use a Hammond 302AX power transformer instead of T1 + T2 A&B |
![]() |
Click here to see the "cleaner" 10kHz square wave from the Morikawa SE2A3 amp |
In spite of the 10 kHz square wave spike from the budget Hammond 125ESE output transformers, it was a very refreshing musical reunion with the Angela SRPP 2A3 amp.
During its residency in the man cave, the amp powered my back up Altec 2-way system above, Altec 755As and Altec 755Cs in Silbatone cabs below. It was totally devoid of the listener fatigue I previously ascribed to the circuit. Even with the brightest and harshest software and/or hardware thrown into the listening chain, I couldn't induce the unpleasant quality which bothered me in the past.
Then it dawned on me that I started doubting the SRPP's musical abilities when my Altec 2-way project commenced. Sorry SRPP, it wasn't your fault...😟
Joe Roberts' piece, "I Never Met a 2A3 Amp That I Didn't Like," from Sound Practices Issue 15, is very persuasive. I built jc morrison’s micro 2A3 amp soon after I received the issue via snail mail fresh from the printers in the mid '90s. After rereading the article in ‘07, I got inspired to make a Radiotron SE2A3 variant.
![]() |
jc morrison's micro |
Since my Altec 755As in open baffles required more power, or so I thought then, the power and output transformers were expropriated for the classic JE Labs SE300B project.
![]() |
Fi "X" by Don Garber from a Sound Practices ad |
In 2017, I had wanted to build a 6SF5>2A3 amplifier tribute to the Fi "X" in memory of Don Garber 😢 but a busy schedule got in the way.
During the March 2020 lockdown, I was motivated to review my audio DIY to-do list which led to a long-overdue parts bin inventory. I turned up a 700 VCT @ 150 mA power transformer pulled from a junked Scott 99D mono integrated. This iron is not a stranger since I've used a pair in my PP2A3 mono-blocks in the olden days. I dug up a couple of 6C8G grid capped/octal based twin triodes as well.
6SL7 vs. 6C8G
transconductance (gm)= 1600 umhos (both)
gain (mu) = 70 vs. 38
plate resistance (Rp) = 22,500 vs. 44,000
max. plate current (each plate) = 2.3mA vs. 3.2mA
Despite the equal transconductance, the 6C8G isn't a drop-in equivalent of the 6SL7. Aside from the different pinout, the mu and Rp are lower. When the triode sections are paralleled, the Rp is closer to a medium mu/gm triode like a 76.
![]() |
morrison micro, reimagined |
This 2-stage direct coupled circuit is a reimagined morrison micro using a 6C8G front end. The paralleled triode sections draw about 3.4 mA and the 510 ohm cathode bias resistor is bypassed with a 100uf @ 16V cap to maximize input sensitivity.
Each 2A3 is AC heated by a Hammond 166M2 (2.5V @ 3A) filament transformer. The operating points are well within the modest specs of the Scott 99D power transformer.
The fullest potential of an amplifier circuit can't be realized if it's inhibited by a mediocre output transformer.
![]() |
Hashimoto H203S |
![]() |
James 6123HS |
![]() |
Nature Sound NS70SE |
![]() |
Tamura F7002 Permalloy |
![]() |
Tango/Hirata NY15S |
![]() |
Budget Output Transformers Noguchi PMF-10WS, Edcor GXSE15-16-3.5K, Noguchi PMF-15WS |
![]() |
Noguchi (now G-trans.co.jp) Finemet |
![]() |
Kasuga KA6625ST Square Waves L > R: 100Hz, 1kHz, 10kHz |
![]() |
4.34V rms squared/8 ohms = 2.35W |
Spring and Fall are the best seasons to enjoy food and drinks on our deck. To complement the beautiful atmosphere, I assembled an e-z to set up and dismantle hifi system!
This stack consists of a Raspberry Pi 3B loaded with Volumio music player/streamer tethered to a Signstek USB to SPDiF converter which connects to a Muse TDA1543x4 Non-OverSampling DAC.
Volumio offers a wide choice of radio stations streaming MP3 files up to 320 Kbps. The Radio Paradise channel has even started broadcasting FLAC files. 👍
In the late 80s, I built my first amplifier from scratch using transformers from a cosmetically challenged Dyna SCA35. I used this paraphase circuit from the Dynaco Transformer Catalog with a 6SL7 driving a pair of push-pull 6V6s. Musically, I preferred this amp to my then highly modified Dyna ST70 and pair of MKIVs.
![]() |
with 6F6Gs courtesy of John Piro |
Since I always had reservations about the 1626 triode's musical abilities, I converted the JEL Darling Style amp to use triode connected 6V6s. Wow!👍
![]() |
Input sensitivity = 0.3V, no need for a preamp This amp also does a good job of driving my collection of orthodynamic headphones via adapter. |
Triode connected 6V6s mimic directly heated triodes much better! 🎻🎸
This pair of Electro-Voice Sonocasters came from the flea market. I guesstimate the sensitivity to be in the mid 90s dB @ 1W @ 1M. Driven by 1.4 triode watts per channel from the SE6V6 amp, I sure won't be annoying the neighbors. 🤣
The 8" drivers are unmarked but look very much like MC-8s - not quite an Altec 755C or Pioneer PIM8L in terms of refinement or realism, but not too shabby either + a lot more rugged!
Ever since I grafted a 6SN7 on the line stage section of my hacked Dyna PAS preamp, I've been hooked on the "octal sound." Could it be the higher heater current draw boiling more electrons and/or larger stature that give it a 3-D quality, richer tonal color, detail, dynamics + more, compared to their 9-pin/noval siblings?
Ideally, the 3850 ohm plate resistance of the two triode sections of a 6SN7 connected in parallel should drive a 12K primary Z. I considered getting a pair of Hammond 125BSE output transformers, which have the flexibility of reflecting back a 2.5K, 5K or 10K primary Z but I already had a pair of these 5.5K:4 ohm SE OPTs in my parts bin that I was sure would fit under the chassis.
The circuit is a simple 2-stage cascade - 1 triode section of a 6CG7 functions as the input/driver stage R-C coupled to paralleled 6SN7 triode sections.
![]() |
Square waves at 100Hz, 1 kHz and 10 kHz Top trace = audio generator Bottom trace = amplifier output |