top of page

Marshall Combo Gut Rehab (Hybrid of AX84 Lead Preamp, Fender Harvard 6V6 Power Amp, 2017)

I had this solid-state Marshall amplifier as a teenager. It was 80 watts and had a great aggressive tone and I used it for everything, even jazz band in high school. Unfortunately, the circuit crapped out and it languished for years in the basement.















I always had the idea that I would go back and build a tube circuit in that cabinet, even before I learned anything about tubes. The cabinet itself is good plywood, the chassis is bent aluminum and the speaker is a great Celestion rated for 80+ watts (I think it’s a re-branded G12-75T) and produces a tough, scooped sound with no speaker breakup. And there is of course the sentimental value of re-purposing an old amp.  


Recently, I decided to follow through on the build. After a truly heroic amount of havering and indecisiveness about what circuit to build (Marshall 18 watt? Trainwreck clone?), I settled on a mashup: the “lead” preamp from combined with the power amp section of the 5F10 Fender Harvard (schematic widely available online). The lead preamp is basically a Marshall “Plexi” preamp with an extra gain stage, and there are many clips online of this preamp paired with power section designs also from AX84. They all sound great. The Fender Harvard power section was chosen for sentimental reasons (my alma mater), but also because it has just about the right power to hang with a drummer while still having some power-tube breakup. Online clips of the Fender Harvard show a great Fender clean tone (the circuit has high negative feedback) and a really cool breakup when pushed. The power amp is unusual in using a concertina phase inverter and fixed bias, which is more common in higher-powered models. I liked the idea of a tweed-era vintage power amp paired with a modern, high-gain preamp.


To add flexibility in the preamp, I added a small DPDT switch to completely bypass the second gain stage (this is a bit different than the bypass shown on the schematic). With this stage bypassed, the amp is effectively a Marshall Plexi preamp going into the Harvard power section. Because the preamp was designed for a higher B+ voltage than I would be using, I replaced all 100k resistors with 82k to allow for more current through the 12AX7s. To power the preamp, I tapped the B+ voltage directly after the rectifier tube using a 60H Hammond choke (model: 155C).


Additionally, I added a tube-buffered effects loop, using a 12AU7 cathode follower directly after the tone stack (using a voltage divider to reduce gain by a factor of 40), and a gain stage at the effects return (replacing the 6AT6 of the Harvard). The effects loop is wired with switching jacks, so that there is a direct connection if there are no effects cables plugged into the amp.

Construction of the amp was pretty straightforward. The only difficult part was the metalwork (since I only have a hand drill and not a proper press).






The power and output transformers are from






I tried to follow best wiring practices, but sometimes things got a little messy.

















Fortunately, I only touched the B+ voltage when building this amp. It worked well with no serious debugging required, not counting the horrible squealing the first time I powered on (caused by wiring the negative feedback from the wrong transformer tap, causing it to become positive feedback).


I could call this amp a complete success. Although I live in an apartment and can’t crank it up very often, this amp sounds great at low volume or high. It hasn’t given me any issues in the year since it was first built. It sounds exactly the way I wanted: a focused, midrange-heavy overdrive sound, and a great clean tone. The Celestion speaker gives it a nice thump in the lower end. Even without the extra gain stage engaged (and with some gain lost in the buffered effects loop), the “clean” channel produces an awesome, saturated grind when turned up.


Here is the amp on the “clean” channel, slightly cranked.








Here is the amp on the “dirty” channel (extra gain stage engaged), turned up. Note that this guitar is the same one from I had as a teenager, but fixed up with a new neck, pickups, and accouterment. Although this was a quick, informal video, I think it shows the nice balance of preamp and power amp distortion.



































This amp formed the basis for a deep learning project – trying to use deep neural nets to emulate a guitar amp.

bottom of page