Sunday, 19 February 2012

Guitar amp modding

Sometimes that buy off Kijiji wasn't all that it was cracked up to be, and you wind up with a $40 footrest.  Sometimes, though, you will not take this laying down, and will do your damnedest to polish a turd.  This is one of those times.  Enter, the Nova:

 Fig 1: A piece of junk funk?

This thing boasted an acceptable clean channel, but the "Overdrive" on it sounded something like Inuit throat singing being piped through a colostomy bag.  This thing sucked out loud, even for $40.... that's saying something.
So, instead of shelling out for a distortion pedal, or even being sensible enough to throw out this shabby thing, I decided to take the geekiest route possible and modify it into something acceptable.

I first started out by looking at the signal being delivered to the speaker, on full distortion:

Fig 2: Inuit throat singing, colostomy bag filter applied

If Leo Fender himself tried to tell me this was distortion, I'd have to snap a strat over his face.  The next step was to poke around inside and see if I can identify any actual distortion circuitry:

Fig 3: A promising candidate?

VD09 and VD10 in Fig 3 look like they're in anti-parallel - this is a common means of "Clipping" a signal to distort it.  But, this wasn't what it was cracked up to be when I checked out the connections.  Poking around further unveiled a few sets of diodes like these, but none of them were actually connected for distortion... but, how does this thing sound so horrible?  At last I found the ones I sought:

Fig 4: Geek skillz != photography skillz
The two diodes back-to-back (In the shadow of the large capacitor) are indeed connected to provide clipping! Checking out this PCB's backside confirmed this:

Fig 5: VD08, VD07 connected in (anti-)parallel

So they're applying a common technique, and getting a garbage result.  The scale of this error is like putting a pizza pop in the microwave, but getting a decomposing fetus out of it.  It just doesn't add up.  
When faced with such nonsense it's usually best to write it off as someones careless mistake, but I figured I'd dig deeper and try to figure out why.  Not surprisingly the answer was a bunch of bullshit:

Fig 6: A bunch of BS.

The designer(s?) here was on the right track with the antiparallel diodes, but managed to do it in some stupid-assed way that made it suck.  The solution is to cut through the bullshit and keep it down to essentials:

  Fig 7: Now with 100% less suck

This was the redesign I settled on (Fig. 7).  Notice that it's way the hell simpler - and the way the PCB was laid out, very simple to implement. The capacitor was left in to remove any DC bias, as I'm certain a circuit this shitty wasn't designed to handle direct-coupling.  The diodes directly in the feedback path is something Ive seen in many tubescreamer-type distortions, this is a fairly simple version here.  Check out this link for more info.

So, on with the build!
 Fig 7: The refuckulator

As it turns out, soldering this thing up was a breeze.  In Fig7 above you can see my drop-in replacement for a resistor in the original schematic.  Here's it soldered into place:

Fig 8: Tweaks ahoy!
Fig 9: This is what amp surgery looks like
After implementing my tweaks, I managed to get some garbley-looking output, which is exactly what I was looking for:

Fig 10: this squiggle sounds better than the other squiggle.

And that's all it took! Granted, it took an hour or two of staring at the PCB to decipher it, but in the grand scheme of things, it only took me a couple hours to improve the quality of an otherwise sucky amp.

Note: I wouldn't recommend going through the trouble to mod any amp of actual quality.  I'm pretty sure this thing is a 20 year-old K-Mart amp, which is why it was designed so poorly to begin with.


  1. Shoulda Before and Aftered it... and correct me if I'm wrong, but isnt the "BS" a lowpass filter?

  2. Mitch - it is certainly some kind of filter, but, this amp sounded pretty bad. This was more a quick hack to get a better sound than an exercise in circuit analysis.

  3. FYI, some analysis of the circuit.
    The "BS"'s step response is:

    U_out = (C1+C2)/C1 H(t) +(e^(-t/C1R1) -1)*(2C1+C2)(C1R1+1)^2/C1

    C1 being the capacitor in the feedback.
    So, the filter is effectively a highpass (note the e^xx term, same as discharging a capacitor through a resistor), which is then decoupled through another highpass. Below the voltage threshold the highpass' only load
    is the output voltage divider. Above, the diodes draw most current out of it, effectively moving its cutoff frequency up.

    So what you got is a highpass (maybe intended as some kind of treble boost?), followed by a circuit which clips and while doing so adds some more high frequencies.
    Depending on the values of the parts not a bad approach per se, but i guess this analys goes far beyond what the designers did.

  4. Can you provide the original values for the resistors and caps for the BS circuit. It might be interesting to perform a spice analysis...