Repairing a Vox Mini3 guitar amplifier

I was asked to repair one of these by a friend of a friend, and it languished on the pile for over a year since it was virtually unrepairable, could cost more than it's worth to repair, and I had no clear instructions as to what to do about it.

I'd call this a practice amp, rather a real guitar amplifier.  It's tiny, only about the size of an electric toaster, the speaker is probably about 4 inches wide, and only produces about 3 watts of sound from the speaker.  The kind of thing you might use while wearing headphones, or noodling around with your guitar while sitting in your lounge.  It has a few effects built into it, I didn't think that much of their sound quality, though Hi-Fi sound and guitar amps are usually two very opposite desires.  Mostly, electric guitar players are trying to deliberately create some kind of distortion.

It suffered a common problem that the speaker driver IC had failed (with an internal short that made it run very hot).  The trouble is that IC is approximately a 1 mm square ball grid array (BGA) surface mount device (SMD) that's meant for robot assembly, not human handwork, and has virtually no surface area to dissipate any of its operating heat.  If not for the fact that it had cooked itself so much that it crumbled apart while poking around it, it would have been very difficult even just to remove it.  It had no identifiable part number on it (if it did, it had burnt it off long before I ever saw it), if you could even obtain one knowing what it was.

The owner was faced with a few choices:  Abandon it.  Take it back as a headphone-only practice device (the headphone output is driven by a different amplifier than the speaker).  Use it as an oversized effects pedal (I think there are better, and far more convenient, guitar effect pedals around to make it not worth using it as one of them, and its only output connector is a 3.5mm headphone socket).  Or, I could try to bodge in an external amplifier circuit to replace the broken IC.

In the end I went down the route of replacing the amplifier IC with something else, fitted externally to the circuit board, but still inside the box.

It should go without saying that the original faulty IC needs removing, but I will explicitly say so, just to be clear.  If you leave faulty parts in-situ, it's just inviting further trouble in the future.  Even parts that might seem to be open-circuit, now, could easily develop other faults, too.

The original IC runs from 5 volts (as a tiny low-voltage high-current device, which was probably its downfall—cooking itself).  It appears that its input stage may be differentially driven (not sure of any advantage of doing that where they've done it), and it may drive the speaker in bridge mode (both outputs drive the speaker in opposition, rather than singled-ended against ground, which would allow for around twice as much voltage across the voice coil, but I'm not sure how noticeable that is with such a small system, and a doubling of a signal voltage isn't as much of an increase in volume as you might think).  It gets it's audio feed via switched contacts on the headphone socket AC-coupled to the IC, muting the speaker when headphones are plugged in, and the speaker is direct coupled to its output.

I opted to build a test board be fitted with a normal-sized IC that runs from around 10 to 22 volts, and power it directly from the other side of the power switch, where it can get the full voltage from the external DC supply socket.  There's an internal battery holder that takes six AA cells, but that will only produce 9 volts, which may not be enough to power the speaker amplifier, particularly as the batteries flatten.

I bodged together a prototype using a LM380 amplifier IC, since I have one to spare, it can run from a single supply rail, and is about the same wattage as the original (so the output volume ought to be similar).  But I have no idea whether it will sound similar, since I never heard the device before it failed.  I don't have a guitar to test it with, so I hooked it up to a portable radio, and it sounds about like what I'd expect all things considered (simple PA IC, basic single-speaker).  I've essentially used the suggested design as per the IC's spec sheet, but fed it power through a diode for reversed-power protection, with some large and small decoupling caps to shunt low- and high-frequency power supply noise to ground, and to help stop self-oscillations.  And, the inverting input is shunted to ground through a small capacitor, also to improve stability, and minimise stray noise pick-up (mentioned in the spec sheet, but no actual example values were given).  The 470 nF bypass cap is per the spec sheet, as is the Zobel RC network on the speaker output (for high-frequency amplifier stability—by compensating, with the opposite effect, for how the speaker's impedance rises with rising frequencies), and so is the large DC-blocking AC-coupling cap to the speaker (this IC's output is biased at half the supply voltage).  The resistor in the Zobel network should probably be one that's able to carry a few watts (the same as the amplifier's output), though that doesn't always seem to be needed.  None of these component values are particularly specific, something close to them is generally fine.

I could use two LM380s to make a bridged amplifier, but, as I mentioned earlier, I'm not sure if that'd make a significant difference, and I have no clear instructions about how far to go in resurrecting the amp.  Also, the internal speaker is 4 ohms, the LM380 really suits 8 ohm speakers, and bridging two amps into one speaker virtually halves the speaker impedance.  So some other IC, which I don't have, would probably be required.  And only having a low-voltage single-ended power source is somewhat of a limitation as to what ICs can be used, too.  And, if I do too much experimentation and design research, it's not financially viable to repair the amp.  So far, it's been around three hours work (finding and clearing the original fault, then building, testing, and fitting the replacement power-amp board), which really is spending too much time on it (though would have been quicker if I still had a workshop set up for this kind of thing), and $13 worth of parts.

Some people will argue that there's benefits in creating an amplifier just using discrete component parts instead of using an IC, but an IC is a tried and tested circuit, has more consistent component tolerances, can have short-circuit protection, operational stability over a range of operating temperatures, over-temperature protection, less distortion, etc, with all of the hard design work already done for you.  Simple two or three transistor amplifiers don't have all that, it's up to you to try and design, and test and debug, a circuit (which becomes increasingly complex) that does.

The prototype was built in a very old-school way, and was eventually fitted as the repair.  Essentially it was just bodge together some spare parts to see if it could do the job, then fit it as-is.

I used a single-sided copper-clad circuit board, but didn't etch any traces nor drill any holes for component legs (just some screw holes to mount the board inside the cabinet).  The IC is directly mounted to it via its ground pins (the middle three pins on each side actually act as heatsink contacts, using the PCB to dissipate some heat (only time will tell if that's enough).  The other IC legs are lifted up, with components soldered to them and the nearby PCB which also acts as a ground-plane, and some components are point-to-point.  I cut some gaps in the copper with a Stanley knife to make a few isolated pads, for positive power in, speaker positive out, and mounting the trimpot (for audio-in, audio-out, with its third leg connecting directly to the ground-plane).

This kind of circuit construction has some advantages:  It's simple to do, it involves very little work in creating the board.  It's not the smallest or neatest, but is easy to make changes and experiment, it's quite robust, and it's clear to see what's what and where it's connected.  The large ground-plane helps reduce the pick-up of external noise, and gives a large surface area for some components to radiate their heat through.

The rest of the amp's boards are similarly exposed to damage from foreign objects rattling around the inside of the amp, so there's little point in encasing it.  It's common enough for guitar amps to not enclose the electronics, and it was very traditional to use point-to-point wiring.

Scoping around the small electrolytic capacitors near to where the original IC was on the main board, I found an AC-coupled audio signal to drive the amplifier with.  Since the two ICs (original and repair) have different gain, I elected to put a thumbwheel trimpot on the external board to be able to adjust the signal level to suit it.  This board has been screwed to the back panel adjacent to the external DC socket, and the thumbwheel can be reached by feeling through the gap in the back of the amp.  It's out of the way, so it's essentially a set-and-forget device, but the owner can tweak it to suit themselves.  Though it might be destructive if they decide to use it as an over-drive control.

It also had grotty contacts on the ¼″ sockets, you could hear hum and popping noises if the amp was bumped.  When there's no jack in the socket, it shunts the input to ground to mute it, and it wasn't doing that properly.  They just needed a clean and light scrub.  That's a very easy fix if yours does that.  It's no surprise that guitar leads and sockets need a clean every now and then, the leads end up on the floor and pick muck up, and people think nothing of putting their fingers all over the conductive contacts of the plug.

Not having a guitar to test it with, I took it up to my friend and got him to try it out.  It's surprisingly good for such a tiny box.  Naturally it can't match the bass output of a big speaker, but it certainly put out a lot of sound for such a small practice amp.  It's easy to fall down the trap of buying amps for their high wattage because big numbers sound impressive, but it only takes a few watts to sound loud in a small room.  And while the effects aren't particularly stunning, they're good enough to take the dullness out of just playing something straight.

And I've, since, heard that the owner likes the repair, and said that it's even louder than it used to be.  So I'd call this LM380 circuit a viable solution for repairing this amp, and certainly easier than working with SMD components.

If I were doing this repair for myself, where time and effort is no object, likewise for any other costs, I can think of some possible improvements:  As it stands, the power ground and the shielding for the audio link between the main board and this one form a loop, albeit a very small one.  I might have tested disconnecting the shielded audio cable's shield at one end to see if it offered any benefits, but the chances are that this Lo-Fi device wouldn't see any significant change.  I might have tried a better audio power amp IC, though the LM380N is doing a better job than I expected.  And it's probably worth replacing the six AA cell battery holder to use a better (rechargable) battery than using AA cells (they don't last long, it can become expensive to keep replacing them, and modern batteries tend to leak badly before they go flat, which was your usual prompt to remove them before they leaked).  I use 12 volt lead acid gel cell batteries with my Yamaha P45 plastic piano, it gives me a long running time, is easy to charge, not expensive (the charger can cost more than the batteries), and they're generally safer to charge than the modern lithium batteries infamous for exploding and causing so many fires.