Power Amp in a Box

[jishnudg]

I'm looking to build a power amp in a stompbox sized form factor, with 9v input. What chip should I use (I'm guessing it has to be opamp to maintain simplicity of the circuit and smaller board size, but am open to other options) to get the maximum wattage and output? Onboard voltage doublers and inverters are required,I suppose (lets say, to get +18v and -18v) but I don't think this will suffice for power amp chips like LM3886 etc which will almost certainly need a transformer - which I'm reluctant to work with at this stage. So are there any alternatives I can look at? Maybe a bridged configuration or something?

[Johan]

I think you should look at the tiny giant. Not 9 volt, but close enough
J

[PRR]

> with 9v input

From what kind of supply? Pocket-radio battery? 9V car battery? OneSpot or similar?

LM386 works great, and simple, from 9V. Makes just under a Watt in 8 ohms. Battery life may be poor from pocket-radio batts. Killer 1-Watt with 400mA or more wall power.

Note that if your "9V" is at all weak, and is shared with pedals, the surge and sag from high-power playing will get back into your pedals and make trouble.

All the Car-Radio "12V" chips will work dandy on 9V, at about half car-volt power (8 Watts in 4 ohms). The Tiny Giant PCB kit is a great base. Omit the regulator and feed +9V to reg's OUT pin-hole. Note that bridge-mode takes FOUR times the current, and more if you go low-impedance speaker to suck more power out. It could take ALL the current available from a OneSpot.

There's good reasons to keep POWER amp supply separate from small-stuff power.

[R.G.]

Paul is right. While some results can be obtained from a 9V supply, that's not sufficient for a lot of purposes. Worse, if you're following the common thinking that a power amp is a specific circuit or integrated circuit, you're following the herd's erroneous thinking.

Power amps are, after all, about power. If you don't have enough DC to power it, it simply can't put out the necessary power, whatever that is. And if it can't get rid of the waste heat, it will overheat and shut down (if it is a design with an internal thermal shutdown) or die (if it's crude enough not to have a thermal shutdown).

A power amp is best thought of as a DC power *supply* that has a little extra junk tacked onto it to let the power out to a load in a tightly controlled manner. The circuit is almost moot, affecting the end result only by how much waste heat it puts out.

The proper way to look at power amps is:
1. How much power do you need? 1W? 10W? 100W? Be judicious here. A 10W amp is subjectively "twice as loud" as a 1W, not 10 times. And 100W is about "four times as loud" as a 1W, because of the quirks of the human ear.
2. What speakers will you drive? Speakers put out sound with varying degrees of efficiency in converting electricity to sound. Speaker makers rate them in db of sound pressure level for 1W of drive, generally at 1m distance, on the speaker's axis. Guitar speakers are often rated at 96db to 12db at 1W. That's a lot of racket - er, rock and roll. The big issue on this one is that you have to decide what power you're putting out the speakers, and what impedance the speakers will be. If your speakers all serial/paralleled up as you wire them are 8 ohms, you can relate this back to the size of the signal needed to produce a given SPL. And that signal size tells you something about the power supply you need.

10W into an 8 ohm resistor requires a signal of 8.944Vrms, or 12.64V peak, 25.29V peak to peak. If your power amp does not put out 25Vpeak to peak, it will not drive 10W into 8 ohms. You can change to 4 ohms by paralleling up two speakers. The voltage needed into 4 ohms for 10W is 6.3Vrms, 17.8V peak to peak. Lower speaker impedance need less voltage, but more current.

3. From that kind of thinking, you can work back into the amp circuit. Assume the amp  can only put out a signal 2V less that the + and - supplies. 10W into 8 ohms will need a power supply of about 29V total.  10W into 4 ohms will need about 22V total - and more current!

4. Once you have figured out what power supply is needed, you can then look into what amplifier circuits and/or integrated power circuits will (a) withstand and work correctly with that power supply voltage and (b) put out the desired power. You'll also need to think about how much waste heat comes out of the power devices, and provide a heat sink to get rid of that. Class AB amps generally peak at a waste heat of about 40% of their max power as a very, very crude guess. So a 10W amp will need to get rid of 4W, a 100W amp 40W. At least. 4W can usually be done with Mickey-Mouse sinks like tying the power devices to the chassis thermally. 40W generally cannot.

I did a bunch of this. I got an amplifier with an honest 25W into a 1590BB, and it runs without thermal shutdown. That was not a trivial thing to do.

[brett]

Hi
these questions come along fairly often - the variation is usually something to do with the type of amp: bass, Class A or AB or D or H, for a special speaker, etc...

The common theme, as RG suggests is that people try to get a lot of wattage out of a small power supply (e.g. 9V at 1 amp). The usual reply is that 1W can sound very loud if you've got an efficient speaker (say 95 dB per W), so don't worry about the power too much.

Now I'm going to (slightly) contradict myself and almost everyone else and say that I find some extra satisfaction in building a 5W amp rather than a 1W amp. I don't really know why, but that's what my experience says.

You ask about chips. Amps based on the LM386 have a good reputation. However, bridging them halves the load on each chip, which reduces the output from each half.

My suggestions are to either: 1. build an LM386 amp, enjoy the simplicity, but don't expect to sound like Angus Young.
2. Get another power supply and open up a whole range of designs. The charity shop will have a bucket of power supplies. Inkjet printers need about 25 to 50W to run them (e.g. 32V at 1.5A). These make terrific supplies. Safe voltage, low cost(~ $5) and supports a charity!! Then you might build a low-watt Class A MOSFET amp, or a high-watt discrete AB amp.

There are a lot of choices. Probably too many.
Have fun!
cheers

[jishnudg]

Thank you everyone for clarifying these doubts...
So lets say I take an inkjet printer power supply and use it to make a 3886 power amp (the ratings for these power supplies should be adequate I guess)...can I make a "9v" out in the same pedal to power my other pedals (hence eliminating multiple power supplies/ wall warts) ? How do I go about making the inkjet power supply fit (and safe) for use with other 9vdc and 18vdc stompboxes?
Thanks again!
J.

[R.G.]

So lets say I take an inkjet printer power supply and use it to make a 3886 power amp (the ratings for these power supplies should be adequate I guess)...can I make a "9v" out in the same pedal to power my other pedals (hence eliminating multiple power supplies/ wall warts) ? How do I go about making the inkjet power supply fit (and safe) for use with other 9vdc and 18vdc stompboxes?

It gets tricky.

First, you have to know what the printer power supply puts out, primarily whether it is unipolar (i.e. 0V and ~+30V) or bipolar (i.e. +/-~15V). If it is bipolar you'll have an easier time of it. This is because chips like the LM3886 and also discrete power amps are simpler to work with if they have +/- bipolar power supplies. This allows the ground to be relatively quiet if the wiring is correct. If you have a unipolar power supply, then "ground" is one side of the power supply, and must carry substantial currents, causing some small voltages on it. You also have to put a big (30-50V, 1000-2000uF) capacitor on the chip output to block the 1/2-power-supply DC level that the output of the chip will have. A bipolar supply lets the output sit at 0Vdc and no output capacitor is needed.

Second (and I already referred to it) by making 9Vdc from the same power power output, you run the risk of introducing hum and switching noise on the ground to the pedals. Most pedals have zero rejection of ground noise, and you'll hear it. So the answer is "yes, you can make 9Vdc from the same power supply, but you may not want to."

As for the last, how to make it safe.  Wow. Huge can of worms just opened. One of the huge advantages of prepackaged wall warts is that the maker has already shielded you from the AC power line safety worries. This is a Big Deal.

[Electron Tornado]

3. From that kind of thinking, you can work back into the amp circuit. Assume the amp  can only put out a signal 2V less that the + and - supplies. 10W into 8 ohms will need a power supply of about 29V total.  10W into 4 ohms will need about 22V total - and more current!

R.G., is that assumption above, regarding the 2V less than the + and - supplies, consistent enough among amplifiers that it can be used as a general rule of thumb for design?

[tca]

I've been struggling also to build a small amplifier and tested a few ideas... various class a designs using mosfets or BJTs, various class ab  output stages with different drivers and also a lot of chips amps (LM386, TDA7052, TDA7052A, etc), 9V battery powered , 9V and 18V unipolar power, +-15v, etc...

As I found by experience designing a power amp is much, much .... more difficult than a distortion/OD battery powered circuit.

The question is: what kind of amp do you want? Have you seen one that you like, DIY or commercial? A reference to work with is always good.

9V battery powered amp have small Ws (.5W clean for a limited period of time, I'm being optimistic), almost any booster will make it to distort. A small 9V amp is never very clean, some time in the future you will want to play without distortion, and that kind of amplifiers will give you a very small headroom clean sound and you have to build another one. Maybe that is a good thing! As an example, look for (in)famous 9v battery powered amplifiers, how do they sound? This holds for "hold" class ab BJTs designs and chipamps.

A unipolar powered amplifier will give you more freedom in terms of clean power (before clipping). In this case you can even go to class a but not with 9V (I've tried that also, had the same bad idea of powering the amp with the same power source used to power my stompboxes). In this case if you go for a discrete design you will also find that the "hold" class ab BJTs designs and chipamps will work. But I'm guessing that you will want more power. In this case our life is easy. Just find a chipamp with the power that you want, put a buffer in front of it, an optional tone control, and that's it! I really don't know how scalable this is, I guess this approach will work for 1W to 25W... greater power requires more knowledge. Although this approach will work, you soon will also realize that these kind of amplifiers lacks something... when you do, you will try to build another, and a different one. But that is a completely different story

I'm building a 5W SS amp, it seems to me a nice compromise between power and family/neighbors way of living. I agree with Brett on this and also "I don't really know why, but that's what my experience says.". It does feel right playing with a 5W amp! Why 5 (beautiful prime number, I guess)?

With the size  restriction of a stompbox sized box my suggestion is the TDA7052 (or the TDA7052A if you want a vol pot) for 1W.

The speakers will also play an important role, I've been building these 1W amps with 5.25' speakers with a frequency response of 50-12.000 Hz and a sensitivity of 90 dB - 1 W/1 m works great.

Cheers.

P.S.
Look for similar entries on diystompboxes.com.

[MDK002]

Here's an interesting little amp in a box that may be just right for you...
http://www.beavisaudio.com/projects/NoisyCricket/

[R.G.]

R.G., is that assumption above, regarding the 2V less than the + and - supplies, consistent enough among amplifiers that it can be used as a general rule of thumb for design?

It's better than nothing. It serves as a reminder to you to go calculate the actual output swing once you know the output stage. But it's an observation and a reminder, rather than a product of math.

A high current output stage is limited in how much it can saturate the output devices on the top and bottom sides of the power supply while pouring maximum current into the load. The biggest limitation is how the base is driven. Darlington or double emitter followers are generally 2-3V in saturation at high currents. Complementary-feedback stages (also know as Sziklai pairs) can go better largely because the arrangement of the output driver transistor is better for getting more base current through the final output device. Quasi-complementary stages will have one of each kind, and may have a different saturation voltage for the + and - sides of the output swing.

MOSFET stages generally have 3-6V of losses at the saturation ends of the signal swing, as they saturate to a resistance, not a voltage level.

For chip amps, they usually specify this for you in the datasheet.

Also, there is a chain of interlinked considerations that make it unwise to rely on whether the output can saturate to 1, 2, or 3V and count on that in a design without checking the final design to see if it does what you though. Not least of these is that your power supply probably sags a bit in general audio amp use.

[Electron Tornado]

A high current output stage is limited in how much it can saturate the output devices on the top and bottom sides of the power supply while pouring maximum current into the load. The biggest limitation is how the base is driven. Darlington or double emitter followers are generally 2-3V in saturation at high currents. Complementary-feedback stages (also know as Sziklai pairs) can go better largely because the arrangement of the output driver transistor is better for getting more base current through the final output device. Quasi-complementary stages will have one of each kind, and may have a different saturation voltage for the + and - sides of the output swing.

MOSFET stages generally have 3-6V of losses at the saturation ends of the signal swing, as they saturate to a resistance, not a voltage level.

For chip amps, they usually specify this for you in the datasheet.

How is that presented in a datasheet? Or, what information from the datasheet is used to determine how many volts below the power supply that the chip can put out?

Here is a link to the datasheet for a TDA2009:  http://www.datasheetcatalog.org/datasheets/208/378193_DS.pdf

[R.G.]

How is that presented in a datasheet? Or, what information from the datasheet is used to determine how many volts below the power supply that the chip can put out?

And I should have suspected that the first-up datasheet would have it hidden! 

In that datasheet, it's buried. Look at page 3, Output Power (each channel). There, they say that (1) typically, (2) at a distortion level of 1%, (3) with V2 = 24V,  (4) with a sine wave in (f=1kHz, and the only way to get a single frequency is with a sine wave     ), (4) Rl=4 ohms or Rl=8 ohms (5) output powers are 12.5W and 7W respectively.

The power in a resistor (which they used in that specification, as they said Rl, not Zl) is P = (V**2)/R, where V is the rms signal voltage. So the voltages across the load resistors are
@4 ohms, V = SQRT(P*R), so V = SQRT(12.5*4) = 7.071Vrms. To convert this to peak for a sine wave, we multiply by 1.414 to get 9.998V peak, which is the same as 19.997V peak to peak. They said the power supply was 24V, and it's only swinging 20V, so it's missing 4V from being able to hit the power supply, and ... honest, I didn't contrive this to make my posting look good, I couldn't have figured you'd use this example... we'd guess that it's missing 2V from the top and 2V from the bottom. 

@8 ohms, V=SQRT(7* = 7.483Vrms, Vpeak = 10.58V, V peak to peak = 21.1, and so it can only swing to 2.84V less than the power supply, about 1.42V on each side.

Many datasheets actually either specify directly how close the swing on the output can be, or provide a graph of output swing versus supply voltage. That's where I looked first on this one, but I had to go dig it out of the normal operation specs.

[R.G.]

As I found by experience designing a power amp is much, much .... more difficult than a distortion/OD battery powered circuit.

This is absolutely correct. There are many more factors that simply have to be considered, or it won't work well. And it has the power available to make any mistakes release ugly-smelling smoke.   

[jishnudg]

Wow...never thought there'd be so many issues to consider....I've built a few 386 amps in bridged configuration so was definitely not ready for this So right now I'm looking at the General Guitar Gadgets LM3886 schematic and all it says is 18v bipolar supply. The 3886 datasheet mentions a maximum of 35v bipolar. Supposing I want to just build the amp directly off the GGG schematic, but power it with the 35v bipolar maximum specified in the datasheet,
1. Whats the best power supply solution in terms of voltage and power requirements to milk the maximum output out of the 3886? Laptop supplies are mostly around 17v - definitely not enough...any precedents? has anyone here used a printer's power supply, or any other novel solution to power this chip at its practical maximum?
2. I suppose the other big issue would be heat dissipation....any particular heatsink I'd need, or are there any innovative solutions for this? I've heard of chips being screwed to their enclosures which act as heatsinks, but would that be adequate in this case?
3. Are the ratings specified in the GGG bill of materials safe enough?
Thank you all again for the clarifications - am really learning a lot from this thread (or trying to )
Regards,
J.

[PRR]

> LM3886

Complete LM3886 _kits_ are readily available on eBay for prices lower than you can collect the parts yourself. Many dealers offer suitable heatsinks, power supply boards, and transformers.

But anything more than a modest very-simple power amp leads you to *120V Wall Power wiring* (or 230V if you are so inclined). As R.G. keeps hinting, this is NOT something to fool around with. This is not a dead chip, or a small plume of stink. This can kill you or burn your house down.

Heatsinking a LM3886 to survive 8 ohms at 70V is a heroic job. That's a show-off spec. There are commercial amps which run LM3886 that hard; they had to melt-down a lot of motocycle engines to cast those heatsinks.

In the range over 5W-25W, it really makes sense to haunt the pawnshop and Banjo Center's trade-in room, find a deal on a pre-owned factory amp.

[petey twofinger]

http://compare.ebay.com/like/110932147227?var=lv&ltyp=AllFixedPriceItemTypes&var=sbar#

we have been using these , with modeling type multi effect front ends and 12 volt sla batterys , add a 7809 for powering a few pedals , and maybe a lil booster in between for extra hassles . my wifes rig is the 100 watt board with two 12 volt sla's and a zoom b2 . my kids is a micro cube hacked into a larger box with a 2 x 25 t amp board , mine is the same amp , with a korg pandora front end and ac adapters inside too . we ran them afor 5 hours on sunday and they were all at 12.7 from 13.1 or so . i realize this is a lot different than the original post idea but i just wanted to share , i am extremely happy with these rigs so far . we also have lap top style power supplys and have run these indoors for extended periods with no issues , so far ... i did try a pyle amp once , huge mistake . oh , and i have tweeter style bi axle speakers on mine ... talk about doing it wrong , i should be ashamed ... lol . it sounded like crap with the jenson mods , those are now in the micro cube hack amp . i do have a larger cab with 2 tens and tweters for mine but , it doesnt go to the woods with us , no room in the volvo !

[jishnudg]

What about the LM384? Saw a project called "Sparky - 5w amplifier" http://www.paulinthelab.com/2013/02/sparky-5-watt-amplifier-stripboard.html   So will something like this work from a 9vdc (perhaps doubled with a 1044)? The paulinthelab project doesn't mention any heatsinks or special power supply (aside from 12v).

[Electron Tornado]

And I should have suspected that the first-up datasheet would have it hidden! 

Many datasheets actually either specify directly how close the swing on the output can be, or provide a graph of output swing versus supply voltage. That's where I looked first on this one, but I had to go dig it out of the normal operation specs.

On the other hand, you provided everyone an example they can use when they run into the same situation with a datasheet. Anyone paying attention was taught how to fish.

[R.G.]

... General Guitar Gadgets LM3886 schematic and all it says is 18v bipolar supply. The 3886 datasheet mentions a maximum of 35v bipolar. Supposing I want to just build the amp directly off the GGG schematic, but power it with the 35v bipolar maximum specified in the datasheet,
1. Whats the best power supply solution in terms of voltage and power requirements to milk the maximum output out of the 3886? Laptop supplies are mostly around 17v - definitely not enough...any precedents? has anyone here used a printer's power supply, or any other novel solution to power this chip at its practical maximum?

Since you seem to be open to experimentation:
The 3886 has proven to be about as simple as a significant power amp gets, and reliable if properly powered and cooled. But again, ignore the power amp to start with, and concentrate on how you will power and cool it, at least to develop some idea of the physical build. The physical build will be much more challenging than the circuits, I can promise you.

The 3886 will work OK from +/-10V (as I remember) up to +/-35, literally anywhere in there. A good choice, and easy-ish to find is a 36Vct transformer, giving about +/-25Vdc. You'll get an honest 25W or so out of that.

I recommend you NOT power the chip at its practical maximum. That's an exercise for experts. Too easy to get exploding chips and speakers. Instead, as a radical idea, make one power amp per speaker. [Er... you DO have the speakers for this, yes?] Find a good design for a single 12" speaker cabinet, build that box and put the speaker in it, and only then build an amplifier that fits into the back of that speaker and powers it up to 25W. If you really, really need 50, 100, 200W, build more speaker/amplifier cabs, all the same. It is much, much easier to build a 25W amp than a 50W, and add a few more muches for 200w.

The mechanical part of building an amp is far harder than building the power supply and heat sinks, and these are far harder than building the circuits and PCBs. Do NOT fall into the trap of thinking that if you have the populated PCB for a power amp that you have a working amplifier.

2. I suppose the other big issue would be heat dissipation....any particular heatsink I'd need, or are there any innovative solutions for this? I've heard of chips being screwed to their enclosures which act as heatsinks, but would that be adequate in this case?

Do NOT go innovative on heatsinking. Get a decent heat sink.

THIS is an example of a decent heat sink for a 20-50W class AB power amp, either discrete power transistors or LM3886. http://www.ebay.com/itm/New-Large-Rectangle-Aluminum-Heatsink-with-screw-mount-DIY-POWER-AMPLIFIER-LED-/251219076815?pt=LH_DefaultDomain_0&hash=item3a7dd2e6cf Good price, too.

3. Are the ratings specified in the GGG bill of materials safe enough?

GGG shows out of stock, discontinued, but to be updated.

Again, AC power wiring is not a joke. You can kill yourself, even far in the future once you think it's safe, or kill someone you care for, or burn down the building, with all that entails, if you do it wrong.

One thing I particularly like is buying someone's nightmare of an amplifier very cheaply, throwing out the amplifiers inside and performing a "soul transplant" with good circuits. I bought a nominally 100W combo amp, the Guitar Center cheapo house brand "Rogue" for $40. This had a cabinet, two 12" speakers of dreadful quality, a power supply for 100W solid state, heat sinks, and AC power wiring already done. Oh, yeah. There was some smear of junky circuits inside that I carefully wiped away. I can't build the speaker enclosure for $40 of materials, much less $40 of labor. The speakers were junk. OK - I was buying a cabinet, chassis, power supply and heat sink, already mechanically set up. Kewl.