1W Push/Pull 12AU7 Amp

This is another conversion. This time it began life as a Heathkit CB-1 Transceiver. The name of this amp, Lunch Box, comes from the fact that the CB-1 was in a case affectionately called the “lunchbox”, because it looks like, well…a lunch box. I found this one at Urban Ore in Berkeley for $5 (although the chances of finding a deal like that in 2019 are pretty much nil - there's not much tube equipment left to salvage, and what there is generally goes for a lot more now). It only had 2 of the orginal 4 tubes, but one of those is a Mullard 12AX7. I now understand the craze for these tubes. I popped it into the Micro Champ and it beat the pants off of new Russian tubes.

I haven’t been able to find much about the CB-1 on the net, but several other models in the same case have a spot at the Heathkit Virtual Museum. There are also decent pictures of the case here, and here.

This amp uses the 12AU7 dual triode tube in Class AB push/pull output configuration with a long-tailed PI. Doug Hammond’s Firefly also uses a 12AU7 as an output tube, but in self-split push/pull configuration, as does Mark Lavelle’s Decimator.

Point to Point Construction

I used old-school point-to-point construction for this amp. With the pre-existing holes in the chassis in weird places, I couldn't easily fit a turret board in, so there are simply a bunch of terminals with component soldered directly to pots and sockets and switches, etc...

It's a mess.

And works perfectly.

There's basically no noise whatsoever, which I attribute primarily to two things - the very direct audio signal path, which hardly ever crosses high-voltage lines, and the tightly twisted heater wires, which are kept far away from the audio signal.

Tone Stack

The amp has a Baxandall Tone Stack, driven by a MOSFET source follower. The stack is also bypassable by a push/pull DPDT switch – without the tone stack it’s exactly like Revision 1a. The source follower doesn’t add any coloration of it’s own, but it might impart a nice glassiness to the top end by allowing high frequencies to pass through the tone stack at lower impedance. At any rate, I think it sounds quite good. I added one to the Moonverb in the same position, but took it out again because I felt there was too much treble going on. The 12AU7 seems to be a bit more mid-rangey than the 6SN7 as an output tube.

The Bax stack is particularly useful when driving the internal speaker. The internal speaker is an old 3” alnico radio speaker – naturally, it doesn’t like big bass signals very much. With no tone stack, I was basically limited to playing the top 3 strings. With the Bax stack, I can dial out enough bass that the speaker is happy. This speaker, by the way, sounds surprisingly good! I don’t have an SPL meter, but I’d guess that its efficiency is somewhere around 80dB/m/W. The perceived SPL seems roughly equivalent to the same amp through my 94dB/m/W 8” Jensen with 15dB of attenuation.

Naturally, with the tone stack there’s less overall gain, but I really like the sound. With the mid-cut from the tone stack and an NOS GE 7025 in the phase inverter, it gets a sort of SRV-ish vibe that’s very nice. Alternatively, using a 12AU7 in the PI gets a very fat, warm clean sound. With the tone stack engaged, full volume will drive it into mild to moderate crunch, and it cleans up nicely with the guitar volume knob. The tone stack needs to be bypassed to get into serious overdrive.

Internal L-Pad

One more small addition – naturally, I added a switchable 6 dB of attenuation to the internal speaker. The L-pad is built right onto the switch which enables it, on the rear panel.

Operation with Alternative Output Tubes

12BH7

As drawn above, this circuit can also be used with a 12BH7 in the output position. The 12BH7 is very similar to the 12AU7, but it's rater for a higher max plate dissipation, so it can handle higher plate currents. It also has slightly higher transconductance, and slightly lower plate resistance.

To operate it with a 12BH7, the output staged should be biased up to about 10mA or 11mA per triode, whereas the 12AU7 is happiest at about 6mA per triode.

After doing comparison recordings, I’ve decided that there is essentially no sonic difference between the 12AU7 and 12BH7 in this amp, at least not for the set of tubes which I have. At first I thought the 12BH7 sounded more aggressive, but it turns out that’s really just because it’s slightly louder.

Heater Switching

I built one more feature which doesn’t appear on the main schematic, and that’s a DPDT switch to switch the heater wiring to support either the standard 9A basing (for 12AU7, 12BH7, etc…) or 9AJ, which is the basing used for the 6CG7 and the Russian 6N1P tube. The 6CG7 is a 9-pin version of the octal 6SN7 tube found in the output stage of the Moonlight.

9A and 9AJ are identical except for the heater. 9AJ supports only a 6.3V heater (pins 4 & 5), with pin 9 connecting to an internal shield between the two triode sections.

With this switch & adjustable cathode bias, the output section supports quite a range of output tubes. Here are some of the tubes which could be used as output tubes:

9A
- 12AU7
- 12BH7
9AJ
- 6N1P
- 6CG7

Revision 2

Revision 2 came about because of my natural inability to leave well enough alone. I love tremolo, but it’s an effect which, to my ear, sounds best with a good clean tone. And while this amp has a decent clean tone, it just doesn’t have much headroom. My Moonlight, on the other hand, has great clean tone. And with my Moonlight having reverb too, it was just calling out for the addition of tremolo (that’s a project awaiting some time for planning). I figured it would suit each amp better if I added tremolo to the Moonverb and gave the LunchBox another raucous crunch option.

So revision 2 has the following changes:

Tremolo removed
Bright switch removed (to reuse the switch for the channels, below)
6AU6 converted to a gain stage, as a second channel in parallel with the 5879 channel. The amplification factor of the 6AU6 is about twice that of the 5879, so this channel should get pretty raucous.
The bright switch was converted to a switch to use both channels in parallel through either input jack. This leaves the inputs flexible — with the switch open you can use each channel independently, or pan between them if you happen to have one of them cool Ernie Ball panning volume pedals. With the switch closed, you get both in parallel.
The power supply connections for nodes V2 and V3 have been swapped. In R1, V3 fed the PI and V2 fed the 6AU6. With moving the 6AU6 to an input stage, I wanted for filtering on its B+ supply, so I swapped it with V3.

For the 6AU6 channel I’m using the Marshall 18Watt tone control.

Schematics & Sound Samples

Projects Lunchbox Amp

Table of Contents