DIY guitar amp

[PBE6]

Thanks. I will take a look at the Tiny Giant project and see what I can learn.

vigilante397, do you have a link to discussion about your Class D build?  There may be other easier paths, but I'm leaning toward a SMPS and a Class D power strictly for the small/powerful/cuteness factor and it would be great to see a working example.

Although I enjoy building pedals from scratch, I think I will be purchasing the power supply and power amp off the shelf and limiting the perf board activities to the preamp section.  Is there anything obvious I should be looking for spec-wise from these power supplies and power amps?  I think my original idea about the wattage was a bit over-blown, scaling back to 35-50 W average power seems like a more sensible option.

[vigilante397]

do you have a link to discussion about your Class D build?

Nope, never had a discussion about it I built it last summer I think. I just went back and looked at my design though and it looks like I lied, it ran on 24V, not 12. I used a switcher to step a 9V input up to 24V which I fed to the amplifier. It was a scratch build based around the TPA3106D, which I think is the same one EHX uses in the Magnum 44. In the configuration I had (mostly limited by the power supply) it was good for probably 30W or so.

The amplifier worked and it was okay, but I ended up scrapping it because my goal was to have a world-changing portable super powerful amp, and I wasn't able to get it as portable as I wanted. I may go back to it eventually, but I haven't touched it in almost a year.

[samhay]

My very limited experience with the Vox is that Class D is fine as long as you run entirely within its limits--i.e. not as a guitar amp, but rather as a PA amp reproducing guitar. If you expect it to contribute nothing tonally, you won't be disappointed. Of course, YMMV.

This.
I have been using a class D amp powered from a large laptop power supply as my dining room amp for a few years. I initially used it as a practice amp for bass and you'll find almost all contemporary bass amps are now class D affairs. Clean, it works beautifully, but nothing good can be said about over driving a class D amp, so you need to get all your dirt from the preamp, and you need to limit the signal out of the preamp quite carefully.

[PBE6]

The general consensus seems to be that the distortion characteristics of Class D amps are awful, so I will try to run it clean as you've described and leave the fuzz to the preamp.

I went ahead and bought a few pieces last night, a TDA3116D2 board for the power amp (2 actually, in case one is a dud) and a WX-DC2416 power supply.  The power supply comes in a bunch of different flavours, but this one is supposed to be able to deliver 6A which I hope will be sufficient for the power amp which will run off of 24V DC.  If I understand correctly, with an 8 ohm cabinet the maximum average clean power should be Vrms^2/R = [(24/2) / SQRT(2)]^2 / 8 = 9 W, the peak power should be Vpeak^2/R = (24/2)^2 / 8 = 18 W, and the peak current draw would be SQRT(Ppeak/R) = SQRT(18/8) = 1.5 A.  If that's the case, would I be able to run both power amp boards in parallel?  Is it difficult to get 2 power amps with the same design to play nicely with each other when driving the same speaker?  Even if that's the case, maybe I could just send the preamp output to each power amp and give them their own outputs so I could run multiple cabinets.   

[PBE6]

OK my calculation seems right but it seems like it has to be wrong...the EHX 44 Magnum "amp in a pedal" is supposed to give you 44 W into 8 ohms and comes with a 24V DC, 3000 (sic...they must mean "300") W power supply.  If the peak power from a 24V DC supply into 8 ohms is only 18 W, how is EHX getting 44 W?  Would this thing have a voltage multiplier buried in there somewhere?  Some other trick?  I suppose they could be rating it at much higher levels of distortion, but from all accounts the Class D distortion sounds like crap so anyone expecting 44 W of musical noise is going to be incredibly annoyed with their purchase. 

[R.G.]

I suspect that all Class D amps will eventually come to using the old trick that Thomas Organ Vox amps used - inserting an analog clipper/limiter before the power amp so that the power amp never clips on its own, only at the analog clip level. The analog clip level is adjusted to be just under the threshold of the power amp's (much uglier) input clip level. This way, the power amp itself never displays its own clipping level, only that of the analog clipper.

You still have the issues of what amount of its own switching noise and distortion gets through, but the newer Class D amps seem to be getting pretty good at this these days.

[vigilante397]

If the peak power from a 24V DC supply into 8 ohms is only 18 W, how is EHX getting 44 W?

Where did you get those numbers? This comes from the datasheet for the TPA3106D:

[PBE6]

If the peak power from a 24V DC supply into 8 ohms is only 18 W, how is EHX getting 44 W?

Where did you get those numbers?

The 18 W came from my calculation, the other number were listed on the EHX product page for the 44 Magnum under the "Quick Specs" heading:

Quick Specs

44 watts of power into 8 or 16 ohms
Tone switch toggles between a neutral, flat setting and a Musical top end boost
Maximum headroom potential
Can be driven to a natural, true amplifier saturation
Fits in the palm of your hand
Comes with a 24DC-3000 power supply
Dimensions in inches: 2.75 (w) x 4.5 (l) x 2.1 (h)

https://www.ehx.com/products/44-magnum

[PBE6]

I suspect that all Class D amps will eventually come to using the old trick that Thomas Organ Vox amps used - inserting an analog clipper/limiter before the power amp so that the power amp never clips on its own, only at the analog clip level. The analog clip level is adjusted to be just under the threshold of the power amp's (much uglier) input clip level. This way, the power amp itself never displays its own clipping level, only that of the analog clipper.

You still have the issues of what amount of its own switching noise and distortion gets through, but the newer Class D amps seem to be getting pretty good at this these days.

Ahh, interesting trick!  I should look up some limiter designs - would something as simple as an appropriate selection of LEDs to ground do?
(EDIT: Found this, looks like a biased clipper would be more straightforward: https://www.electronics-tutorials.ws/diode/diode-clipping-circuits.html)

I just had another strange idea.  I remember making an LM386-based distortion pedal many years ago, and liking the sound.  Would there be any issues with placing an LM386 at the end of the preamp section to get that particular kind of distortion, possibly buffering that output, and then sending that signal to the power amp?  Would the buffer be required in order to cut down the current?  Would buffering that signal create issues of its own?

[vigilante397]

The 18 W came from my calculation

Sorry I meant, where did you get that equation? Class D amplifiers operate very differently from Class A or Class A/B allowing much more efficient power output.

[PBE6]

I assumed that the peak voltage for a clean sine wave would be limited to half the DC supply (24/2 = 12V), used V = I*R and P = V*I to get P = V^2/R and calculated a peak power of 18W based on an 8 ohm cabinet.  I'm a straight-up novice when it comes to power amp calculations.  If Class D amps can generate more power than this, I am happy to learn the new equations!  Although hopefully I haven't undersized my power supply at 6A. 

[vigilante397]

As different amplifiers will have different efficiency ratings (though most class-D are 90% or higher) there isn't a one-size-fits-all equation, you really have to read the datasheet

[PRR]

> I assumed that the peak voltage for a clean sine wave would be limited to half the DC supply

That's conventional single-ended totem-poles.

Most higher-power lower-voltage power amps "BRIDGE" two amplifiers against each other.

For 24V supply, at one peak one amp goes to +23V the other to +1V. Vice-versa for the other peak. So you do get near 24V peak swing. 72W peak but measured as Sine we get 36W(rms).

Do note: in a Bridged amp *neither* side of the speaker is grounded! You should not use 1/4" plugs because their jacks are normally grounded and their plugs usually expose the "ground" side of the line to touch and accidental shorts. In fact on the simple single-supply bridge-modes both lines also sit at a significant DC voltage which should not be shorted-out. (~~12VDC is not real lethal but can startle you on a damp day.)

[R.G.]

Ahh, interesting trick!  I should look up some limiter designs - would something as simple as an appropriate selection of LEDs to ground do?

What is critical is to match the clipped analog input signal to the size of input signal that is just below the maximum clipped output. One way to do this is to compute the (analog) gain of the power amp; as in it puts out its maximum output peak voltage with X volts peak on the input. Then you make your clipper put out just this amount and no more. Hifi amps tend (with exceptions, to be sure) to have their final, fed-back gain set so that 0.775mV rms input (0dbm) just makes them hit peak output voltage.

775mV is 1.096V peak, so your clipper should have a max output of about 1Vpeak. Too bad ordinary silicons won't do this directly, so you might have to amplify your signal up so that you could use a four-diode (two positive, two negative) clipper and then attenuate it down to just the right size for the power amp input with resistors and/or a pot. Thomas did this a little differently, with a wider range clipper that had adjustable + and - clipping points, and you tweaked the clipping level. But you get the idea.

I just had another strange idea.  I remember making an LM386-based distortion pedal many years ago, and liking the sound.  Would there be any issues with placing an LM386 at the end of the preamp section to get that particular kind of distortion, possibly buffering that output, and then sending that signal to the power amp?  Would the buffer be required in order to cut down the current?  Would buffering that signal create issues of its own?

If you like the way an LM386 sounds, you can freely concoct an LM386 and attenuate its output signal down to whatever makes the just-before-clipping on the power amp output come true. I don't understand your meaning about buffering and current. The input of most class D amps will be in the hifi range of about 10K or so. An LM386 would not notice a 10K resistor as a load, particularly, so it's effectively running unloaded. If you liked the LM386 distortion running effectively unloaded, you're in.

This is all about manipulating voltage levels into what is likely to be about 10K at the input of the Class D amp. Not much current flowing anywhere, necessarily.

[PBE6]

Do note: in a Bridged amp *neither* side of the speaker is grounded! You should not use 1/4" plugs because their jacks are normally grounded and their plugs usually expose the "ground" side of the line to touch and accidental shorts.

Thanks, did not know this!  I think this particular board must be bridged, as it is billed as 100W mono but uses the same TDA3116D2 chip as the very similar-looking 50W+50W dual channel version.  What connectors should I be using instead?   Speakon? The cabinet I was planning on using only has 1/4" inputs, I guess a retrofit will be in order.

[PBE6]

Thanks RG, great advice as always.  With regard to the LM386, I wasn't sure if it put out enough current (the LM386 being a tiny power amp itself) to damage the power amp somehow, glad to hear it will not.  I think the strategy of using a few diodes to get the ceiling above 1V and then attenuating down to 1V will be easy to implement, and if it worked for the pros then I will definitely give it a try.

[bluebunny]

What connectors should I be using instead?   Speakon?

You can use a plastic jack like this:



It makes no connection to the chassis, so nothing will short out.  It's what I used on my Tiny Giant, which IIRC also uses a bridged power amp chip.

[PBE6]

OK, so my power supply and power amp have finally arrived from overseas and I'm ready to start building in earnest.  A few n00b questions about the actual construction:

1. Enclosure

I assume I will need an aluminum enclosure (or equivalent) to keep the noise out and the signal in.  The power components are impressively tiny, so it looks like the enclosure can be as small as 10" x 6" x 2".  I had a bit of trouble finding anything that size made of aluminum on Mammoth Electronics, Mouser and eBay, although there are a few options if I'm willing to forgo even basic nods toward aesthetics and fork over a bit of extra cash.  Does anyone have a favourite enclosure that is roughly this size?  Also, does it seem likely that the enclosure for this 100W Class D amp needs air holes?  I was not planning on putting in a fan, so it would be passive venting only.

2. Arrangement

Are there any rules of thumb or things to keep in mind when arranging the components to minimize interference inside the enclosure?  I was planning on simply placing the preamp, power amp and power supply side-by-side in that order, is that asking for trouble in terms of feedback/interference? Should I be shielding anything separately?

3. Preamp power

My preamp is essentially going to be a glorified stompbox limited to simple buffering, EQ and overdrive sections.  I am currently leaning towards a Fetzer Valve-type input stage to keep loud guitars and overdrive pedals sounding musical, although according to the RunOffGroove website the input will start to distort at +/- Vp which is probably between 0.7V and 2V depending on the JFET used. I'm going to have to limit the signal going to the power amp too (Class D power amp breakup has a terrible reputation). If that's the case, is there any point in trying to take advantage of the large DC voltage provided by the power supply to get more headroom?  Will a 7805 (maybe a 7809) power regulator do just fine for this stage?

[mth5044]

For your enclosure, go to the Hammond small enclosure index:

http://www.hammondmfg.com/scpg.htm


Scroll down and towards the bottom you'll find the steel and aluminum chassis 1441 and 1444. You might be able to find something in there that fits your needs. You can find most or all of the enclosures on mouser and arrow by searching for the part number. There's other interesting options on that list like the slanted stuff, but the chassis are the most basic. Don't forget the lid if you need it!

[printer2]

... is there any point in trying to take advantage of the large DC voltage provided by the power supply to get more headroom?  Will a 7805 (maybe a 7809) power regulator do just fine for this stage?

It would be a good thing to decouple the preamp from the power amp with either a RC network or a regulator IC. If you run a higher supply voltage for the preamp and end up with a greater voltage signal than needed for the power amp you can reduce the signal with a resistive divider, If you cut your signal in half you are also cutting your noise down in half. If you use the LM386 it has a maximum supply voltage of 15V but the LM386N-4 is a 22V max version.

I am going to use a TPA3118 module along with a laptop PS. I am going to try a BSIAB circuit for the preamp with a Fender type tone stack after it. I tried it with a 12V supply and it worked fine. It was a tossup between working on that or some lighting for my indoor garden. I have a lot of amps laying around so the lighting won.