6AM4 Darling

Copyright 2005-2009  Industrial Valve Audio
All Rights Reserved.
Here is my latest spin on 1 watt tube amplification.

The previous iteration was super-dandy, but it had two limitations:  The output transformers were starting to saturate at sustained maximum drive, and the 12SL7GT preamp gave a slightly dry sound.

This incarnation addresses the output transformer issue directly with the use of factory matched pair of James JS-6112HS output transformers. 

As for the preamp, I went away from the usual suspect (5963, 6SL7GT, 12AX7, 5965) twin-triode solutions, and instead went to a dedicated single triode per channel using the 6AM4.   I have to say that the 6AM4 is the best sounding preamp triode that you have never heard of;  Gain characteristics of the 12AX7, combined with the high plate current capability of the medium mu triodes like the 6SN7 and 12AU7, all stuffed  into a small bottle ‘flat-pack’,  all make for fantastic sound and a unique look.   The dedicated single triode assy also makes good for low noise and cross-talk isolation.  Best of all they are dirt cheap – about $5 each or less right now.  On the downside, transconductance is all over the map; I had to go through about 8 tubes to finally get a perfectly matched pair on my tester.

I kept the loadline for the 1626 the same as before with a B+ just north of 400 volts.  The 6AM4 was also setup to run directly off the common rail HV source – something I have been doing for a while, and I really like the sound.  Standard practice is to drop the preamp source voltage thru a resistor and add some downstream capacitor filtration to de-couple the stages, but you start picking up some odd-order harmonics in the process, which takes away from that single ended mojo.

The power supply is tube rectified via a 5V4G and choke loaded with a ‘Mojo Triple 7’ 5H choke.  Capacitance was intentionally kept light to deliver quick sound. 

All tube filaments are rectified and filtered DC for low noise.  I’ve found that AC filaments themselves aren’t noisy, but no matter how carefully you route them, and no matter how tightly you twist the lead pairs, some AC always bleeds over into the signal path somewhere.   For DHTs it might be a different story, but for in-directly heated cathodes, DC is the only way to go for filament power.

Looking inside, you can see that all the AC circuitry is isolated on one side of the amp, with only DC crossing over into the amplifier side of the chassis, again, to minimize noise.

Line power is delivered via a modern IEC connector, and is of the grounded three-prong variety for safety and noise suppression. 

I had gain to spare with the 6AM4s, so I did allow local feedback at that stage only by not bypassing the cathode resistors.   I also took the extra step to add a ferrite choke and grid-stopper resistor to kill off any RF hash that had the potential to set of ultrasonic oscillations.  I did not do any before or after comparisons, but as configured, I don’t see any oscillation (audio spectrum all the way out to 1MHz) on my oscilloscope. 

Coupling between stages is accomplished via 0.1uF Audience Auricaps paralled with some vintage NOS Sangamo Paper-in-Oil 0.047uFs.  I’ve always felt less is more when it comes to signal capacitors, and as you can see from the frequency response plots, that is the case here.

The 1626s are run hotter than called for on the datasheet for Continuous Commercial Service (CCS), but are still below the limits allowed for Intermittent Mobile Service (IMS).   I much prefer the sound of the 1626 when you kick the voltage up a bit to get it into a more linear operating range, as compared to the ‘textbook’ Darling circuit that leaves you down in the mud, so to speak.  You can see difference on the graph of load lines below – The solid blue line is where my amp is operating; the dotted green is the loadline recommended in the datasheet and the one used in the original Darling circuit and its variants.

The cathodes on the 1626s were bypassed to prevent any local AC feedback and to increase power output.  I used a pair of EPCOS Sikorel capacitors for this, and am very happy with the sound (or more correctly, the lack thereof).  They did take a while to break-in, but are now completely transparent and clear.

No global feedback is used.  I good friend and wise man once told me: ‘Global feedback is a cheap Band-Aid used to cover up any multitude of circuit atrocities’.  I couldn’t agree more.

Performance numbers are tabbed out and graphed below.  These are all from real measured data on calibrated equipment, not wishful estimates or shady seller BS.

The mechanical assembly includes my standard industrial fare:  A heavy steel chassis which is boxed in on all sides, oven cured epoxy paint, all stainless steel fastening hardware and washers, Mil-Spec stake bonding, solid copper ground bus, and appropriate fusing and chassis grounding.  All components are new, the exception being the power transformer, which is a vintage copper shielded Stancor unit.  

The faceplate is nicely figured and oil finished Ovankol Shedua;  re-claimed from support timbers from an abandoned South African Diamond mine. 

This amp ships completely turn-key and ready to go.  Includes all tubes pictured (1 matched pair RCA 1626 triodes, 1 matched pair RCA orange label 6AM4s, and 1 RCA 5V4G rectifier).  Also includes a new IEC 120VAC line power cord.  Everything is cathode biased, so there are no adjustments or tube biasing required.

This amp was engineered and hand built by myself.  There are lots of ‘Darling’ variants out there, but this one I am proud to call my own.  I am quite certain it will give you a lifetime of faithful service and enjoyment.


THD @ 1 Watt / Channel; Frequency Sweep 100Hz - 10KHz
THD ; Level Sweep, -10dB to Rated Output (1 Watt)
THD @ 1 Watt / Channel @ 1KHz
1626 Loadline
6AM4 Loadline