PCB for Small Power Amp Circuit?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ioplex

Well-known member
Joined
Apr 3, 2013
Messages
48
Does anyone know where I might get a PCB for something like the following circuit?

JF93yjT.jpg


This is just the usual power buffer but with a proper heat sink it could probably put 5W into an 8R load. This might be used to drive a speaker, headphones, a reverb tank, etc. And there's no reason why it couldn't be used as just an overpowered line driver. The extra BD139/BD140 transistors are probably < $1 and they improve the headroom anyway.

Alternatively, if I were to have a board made, would you change the circuit in some way?
 
Depending on what you're expecting from it you could change a few things I guess. Using transistors just a bit bigger to be safe for example, not much or the 2N390X may have a hard time to drive them, also you could add a trimpot in place of R1 to be able to adjust the bias.

You could look around and there are a ton of circuits like what you are looking for, big IC in 220 packages or something alike are available to deliver few watts and pcbs for them as well.

JS
 
joaquins said:
You could look around and there are a ton of circuits like what you are looking for, big IC in 220 packages or something alike are available to deliver few watts and pcbs for them as well.
Can you provide some examples of ICs that are equivalent or better than the above circuit?

From looking at Mouser, I don't see a "ton". And what I do see seem unusually expensive or have significant limitations. I suppose there must be an IC that is effectively an op amp with integrated power transistors but I think making the power transistors separate has it's advantages (mounting above or below, adjusting bias) and it's not surprisingly expensive. But if there is a good through-hole chip that is equivalent or better to the above circuit and costs $5 then I would definitely just use that.
 
There are many, I don't know the limitations you are talking about, the idea of adjusting the bias is to deal with crossover distortion, this chips already have that optimized. In a fast search I got the TDA7388 which is too much for your application (~4x40W) but there are others around and is less than $5 in one off already, I'm not familiar with parts numbers to search for a smaller one but you should be able to find some, which will probably be even cheaper.
http://www.mouser.es/ProductDetail/STMicroelectronics/TDA7388/?qs=sGAEpiMZZMtxdzBvM0rKcUIKWvb%2fnxfryoNj0ukLBVs%3d

I've just found another, LA42102. 10W stereo, about half the price of the other.
http://www.mouser.es/ProductDetail/ON-Semiconductor/LA42102-E/?qs=sGAEpiMZZMtxdzBvM0rKcXkuyYiNwCOvK%252bqo0ZOlcmo%3d

Look harder, there are some solutions out there for what you are looking for. Good luck!

JS
 
It is a *24 Watt* amp with NO protection.

You can easily equal it.

LM3886 is well-regarded and LM3886 PCB kits flood eBay.

LME49990 has fantastic THD specs at 600 Ohms, can deliver 50mA to a load.

LME49724 OPA1632 are full-differential 80mA drivers.

LME49600 LME49610 are HIGH-current buffers.

While it is total over-kill, the LM3886 is SO inexpensive for what you get. And the flood of ready-to-solder PCBs.

Another point: LM3886 is probably more popular than the LM277 which I designed into an institutional interface a long long time ago. Or mis-designed, because they died about one a decade. After the second death, I was scrambling to source replacements for the now obsolete part.
 
No need to hit Google  ;D

Pier offers a great Gainclone PCB in his WM thread.
I built a power amp with it some time ago. Maybe you can replace the LM3886 on it with a  LM3876  for less power...

Best,
Carsten
 
PRR said:
It is a *24 Watt* amp with NO protection.

You can easily equal it.

LM3886 is well-regarded and LM3886 PCB kits flood eBay.

LME49990 has fantastic THD specs at 600 Ohms, can deliver 50mA to a load.

LME49724 OPA1632 are full-differential 80mA drivers.

LME49600 LME49610 are HIGH-current buffers.

While it is total over-kill, the LM3886 is SO inexpensive for what you get. And the flood of ready-to-solder PCBs.

Another point: LM3886 is probably more popular than the LM277 which I designed into an institutional interface a long long time ago. Or mis-designed, because they died about one a decade. After the second death, I was scrambling to source replacements for the now obsolete part.
I still don't see an obvious winner.

LM3886 is way overkill for what I'm thinking. I don't need 30 watts, I need 3 watts (an in most cases more like 0.3 watts). LM3886 has a quiescent current of 50 mA (and that's low compared to the other chips I looked at). I don't know what the quiescent current of the above circuit would be in practice but it could be adjusted to just about as low as I want depending on how much distortion I can tolerate. For something like a reverb tank driver, I don't think anyone is going to notice any distortion below 70dB which is probably around the distortion one would get with the power transistors biased almost to the point of being off. Personally I actually don't care about distortion. At least not in the way most people care about it. Why would anyone care about something that can only be detected with an analyzer? Maybe if you're doing processing that needs to be totally transparent like a mixing console or maybe a HI-FI amp just because you can. But for music gear like guitar amps and effect / filter line drivers and DIs and microphone preamps and such, I care MUCH more about low noise. Incidentally the LM3886 datasheet says the Ein is 2uV (as in micro and not nano). What am I missing there?

Also I don't see how LME49990, LME49724 and those other chips are even applicable to this post. Can they really be used to drive an 8 ohm load?

And of course a lot of the power amp chips are not stable at low gain and actually don't reach the billed performance until they're pushed into the 10's of watts range. These chips are designed to drive speakers *loud* which excludes them from a variety of lower power applications like a line driver that uses a 1:2 transformer to put out +18dBu. For something like a reverb tank driver I would actually want low gain (assuming it can be done without killing noise performance) and low power. But the transducer is 8 ohms (can be anyway).

It just seems to me there is a big gap between driving 600 ohms and driving 8 ohms. In the first case a good op amp is fine. In the later case all of the driver ICs are 30 watts.
 
you are probably looking at a headphone - type circuit. a beefy one maybe. if you want to make it a little simpler in terms of components (number of components and so on) you could try out one of the power opamps or a combination of opamp and buffer driver. this will bring you in the targeted range. you want roughly 2W / 8ohms - that would be around 4V / 500mA.  if you can reduce your requirements a little you could think of a LME49726 that will drive 350mA! you will probably need some cooling though. Of course these all come in surface mounted only.....

on the other hand something like the LM4906 might be the right chip for you. simple, low parts count, looks simple enough.

cheers,
michael
 
not dissing your design but some that perceive the sonics of the Sziklai pair output stage to be a bit acerbic. 
30+ years ago, David Baskind came up with a low cost alternative to the API output stage.
please excuse the hasty attached drawing, but one can get the idea.
it will easily drive a 1:2 output transformer.
 

Attachments

  • buffer.jpg
    buffer.jpg
    155.2 KB · Views: 45
Choose LM1875, plenty of those on eBay as well, and for even less money (VERY compact too). Less power (yeah 30 Watts on huge rail supplies, a lot less on lower rails).

I do have some PCB's (home etched)for a much simpler design (BD139-BD140 following an opamp, with bias using diodes, no other transistors).
Opamp plus follower
Not 3 watt though, but plenty for headphones or reverb tanks (although you need some frequency compensation for the latter).
Decidedly half-baked, but this is DIY :D

But if you already have a design, plonk it in Eagle and order PCB's in China, 15USD for 10 pieces at iTEAD.
 
gridcurrent said:
not dissing your design but some that perceive the sonics of the Sziklai pair output stage to be a bit acerbic. 
30+ years ago, David Baskind came up with a low cost alternative to the API output stage.
please excuse the hasty attached drawing, but one can get the idea.
it will easily drive a 1:2 output transformer.
The above circuit is not my design. It is a fairly standard Sziklai output power amp design that has been simplified by using a conventional op amp.

I just simulated your "David Baskind" circuit and although it does perform noticeably better than the usual output buffer, the "diamond buffer" technique is marginally better still and it's actually fewer parts (and no capacitors).

And I think generally a rubber diode bias is just better. And the Sziklai output gives you wide output swing and can drive 8 ohms no problem if desired.

So if I wanted something really compact, a simple "diamond buffer" is 4 transistors and 2 resistors (not counting the emitter resistors which have to be selected for the application) whereas if I really want to drive the heck out of 8 ohms and have good bias control, the Sziklai output (the circuit in the OP) is 5 transistors and 6 resistors.

If I want to drive speakers at 10's of watts, then yes, the chips mentioned are great.

The problem the above circuit is trying to address is driving less than 600 ohms (possibly 8 ohms) really well but efficiently at  low levels (< 1W).
 
A conclusion of the above is that no one has a PCB for the circuit you posted.
You might consider LM380, it needs very little components, and power is right in your ballpark value. No PCB needed.
 
ioplex said:
I still don't see an obvious winner.
You have already received a lot of good advice. I guess your idea of a winner depends on how you define the design problem. For a few watts the commercial approach is to use one of the sundry IC solutions.  (I used a National Semi headphone driver IC in a kit I published back in 1978).  More recently I have used a small (1-2W) class D IC for a battery powered application where current draw was important.
LM3886 is way overkill for what I'm thinking. I don't need 30 watts, I need 3 watts (an in most cases more like 0.3 watts). LM3886 has a quiescent current of 50 mA (and that's low compared to the other chips I looked at). I don't know what the quiescent current of the above circuit would be in practice but it could be adjusted to just about as low as I want depending on how much distortion I can tolerate.
There are a lot of IC solutions to search through and I suspect you can find one with lower quiescent current. Like I mentioned I am using a class D chip that sips current.
For something like a reverb tank driver, I don't think anyone is going to notice any distortion below 70dB which is probably around the distortion one would get with the power transistors biased almost to the point of being off. Personally I actually don't care about distortion. At least not in the way most people care about it. Why would anyone care about something that can only be detected with an analyzer?
If you are concerned about quiescent current and willing to accept a little more distortion a common op amp buffer circuit connects the op amp output directly to the two output transistor bases ( set as basic emitter followers) with a couple hundred ohm resistor from opamp to final output. This approach has zero extra quiescent current, but suffers some HF crossover distortion at higher current. Using a modern opamp this should be arbitrarily clean.   
Maybe if you're doing processing that needs to be totally transparent like a mixing console or maybe a HI-FI amp just because you can. But for music gear like guitar amps and effect / filter line drivers and DIs and microphone preamps and such, I care MUCH more about low noise. Incidentally the LM3886 datasheet says the Ein is 2uV (as in micro and not nano). What am I missing there?
Late stage noise is generally not an issue. Noise will be dominated by early gain stages so 2 uV of noise is not much for a headphone amp running at modest voltage gain.
Also I don't see how LME49990, LME49724 and those other chips are even applicable to this post. Can they really be used to drive an 8 ohm load?
A better opamp with the simpler buffer I describe can do well. the moving target is are you driving 8 ohms or 600 ohms?
And of course a lot of the power amp chips are not stable at low gain and actually don't reach the billed performance until they're pushed into the 10's of watts range. These chips are designed to drive speakers *loud* which excludes them from a variety of lower power applications like a line driver that uses a 1:2 transformer to put out +18dBu. For something like a reverb tank driver I would actually want low gain (assuming it can be done without killing noise performance) and low power. But the transducer is 8 ohms (can be anyway).
Your posted schematic looks a little skeletal. I can imagine stability issues with the compound darlington outputs that have voltage gain.  I guess the 47 pf around the 5532 is enough with a clean layout.  The 0,5 ohm emitter degeneration resistors are large for driving speaker loads, but smaller Rs there will make the thermal stability of class A bias more touchy.
It just seems to me there is a big gap between driving 600 ohms and driving 8 ohms. In the first case a good op amp is fine. In the later case all of the driver ICs are 30 watts.
You need to search again there are surely more canned solutions for a few watts into 8 ohms.

Note: I designed a pretty robust headphone amp. back in the '80s using a fast opamp (tl074) and some to-220 power transistors. While this product was a headphone amp, it could comfortably drive loud speakers so I had to current limit the power supply so it wouldn't blow the thermal fuse in the 1A wall warts when driving speakers.

JR

PS: A nice thing about your published schematic is the compound darlington will swing close to the rails before saturating, often a concern for low voltage power amps.  That said you will run out of drive current approaching the rails. You can bootstrap the bias string with a couple more resistors and a pair of caps. A cap across the Vbe multiplier will improve drive current in both directions.
 
JohnRoberts said:
ioplex said:
I still don't see an obvious winner.
You have already received a lot of good advice. I guess your idea of a winner depends on how you define the design problem. For a few watts the commercial approach is to use one of the sundry IC solutions.  (I used a National Semi headphone driver IC in a kit I published back in 1978).  More recently I have used a small (1-2W) class D IC for a battery powered application where current draw was important.
I actually don't have a problem with class D. I think class D is highly underrated. Unfortunately they're all designed for digital consumer electronics. So almost all of them are stereo, they have digital inputs for muting and such. They're almost all very small with unusual packages like ball array type things. There's no "pro-audio" class D chip amp that I have heard of. But in terms of just power and performance the class-D chips are closer to what I was thinking. If there were a good class D chip amp that was an old-skool package like DIP or wide SOIC that could do 5 watts mono and maybe use a LC filter on the output, that could be great. Of course the market for such a thing is almost non-existant.
 
ioplex said:
JohnRoberts said:
ioplex said:
I still don't see an obvious winner.
You have already received a lot of good advice. I guess your idea of a winner depends on how you define the design problem. For a few watts the commercial approach is to use one of the sundry IC solutions.  (I used a National Semi headphone driver IC in a kit I published back in 1978).  More recently I have used a small (1-2W) class D IC for a battery powered application where current draw was important.
I actually don't have a problem with class D. I think class D is highly underrated. Unfortunately they're all designed for digital consumer electronics. So almost all of them are stereo, they have digital inputs for muting and such. They're almost all very small with unusual packages like ball array type things. There's no "pro-audio" class D chip amp that I have heard of. But in terms of just power and performance the class-D chips are closer to what I was thinking. If there were a good class D chip amp that was an old-skool package like DIP or wide SOIC that could do 5 watts mono and maybe use a LC filter on the output, that could be great. Of course the market for such a thing is almost non-existant.

I like the PAM8xxx-family (PAM8408, PAM8009 -- there are more). Only 3W, but very easy to integrate. I evaluated the 8009 for a design, ended up going with a smaller device (ON Semi NCP2820, smaller than you'd like).

The older TI TPA312x-series comes in DIP and has higher power output (up to 15W, I believe).

The NXP TDA8920 comes in the familiar power amp SIP package, and claims 2x110W.

These, and many more, can be found through this Digi-Key search query. Most seem to not need an external microcontroller.

JDB.
[mentioning a liking for Class D right off the bat would have gotten you an answer sooner]
 
jdbakker said:
I like the PAM8xxx-family (PAM8408, PAM8009 -- there are more). Only 3W, but very easy to integrate. I evaluated the 8009 for a design, ended up going with a smaller device (ON Semi NCP2820, smaller than you'd like).

The older TI TPA312x-series comes in DIP and has higher power output (up to 15W, I believe).

The NXP TDA8920 comes in the familiar power amp SIP package, and claims 2x110W.

These, and many more, can be found through this Digi-Key search query. Most seem to not need an external microcontroller.

JDB.
[mentioning a liking for Class D right off the bat would have gotten you an answer sooner]
Yeah, but all of class D amps have digital controls that equate to logic gates in the signal path which equates to sprinkling little capacitors everywhere.

But I bet if someone really tested all of these (with the digital inputs "off") and compared them, they would find one that actually performed really well. In practice I bet they're quiet compared to the class AB chip amps (but probably not as quiet as a cold biased Sziklai buffer like the OP circuit.
 
ioplex said:
Yeah, but all of class D amps have digital controls that equate to logic gates in the signal path which equates to sprinkling little capacitors everywhere.
I've read this three times and still don't know what you are trying to say. Why does a digital 'mute' input (present on several Class-AB chip amps as well) lead to 'sprinkling little capacitors everywhere' (and why would that be a bad thing)?

ioplex said:
But I bet if someone really tested all of these (with the digital inputs "off") and compared them, they would find one that actually performed really well. In practice I bet they're quiet compared to the class AB chip amps (but probably not as quiet as a cold biased Sziklai buffer like the OP circuit.
Why would the circuit from the first post (you are the OP, right?) be quieter than most modern chip amps? (What is quiet here, nV/sqrt(Hz) or some other metric? What kind of noise performance do you really need, before it gets swamped by other stages?)

Please clarify,

JD 'Purity of Essence' B.
 
jdbakker said:
ioplex said:
Yeah, but all of class D amps have digital controls that equate to logic gates in the signal path which equates to sprinkling little capacitors everywhere.
I've read this three times and still don't know what you are trying to say. Why does a digital 'mute' input (present on several Class-AB chip amps as well) lead to 'sprinkling little capacitors everywhere' (and why would that be a bad thing)?
The capacitance of the mosfet gates that make up those controls (a digital pot could have a lot of gates) can add up. And they have very limited voltage range so the signal must be attenuated first. In theory maybe this stuff doesn't matter but I don't think I would use a chip amp that has this kind of digital stuff just to use the power amp part.

jdbakker said:
ioplex said:
But I bet if someone really tested all of these (with the digital inputs "off") and compared them, they would find one that actually performed really well. In practice I bet they're quiet compared to the class AB chip amps (but probably not as quiet as a cold biased Sziklai buffer like the OP circuit.
Why would the circuit from the first post (you are the OP, right?) be quieter than most modern chip amps? (What is quiet here, nV/sqrt(Hz) or some other metric? What kind of noise performance do you really need, before it gets swamped by other stages?)

Please clarify,
Yes, wideband noise. Hiss. Created probably mostly by series resistance being amplified because a chip amp has a fixed gain of 50x. The OP circuit can be designed to have low gain (maybe unity) and a cold bias could act as a sort of noise gate (at the expense of crossover distortion of course). As for class D, I really don't know if they are actually quieter. I was just thinking that because they're so efficient, if the input signal is low, they're not going to put as much energy into the load compared to an AB which just amplifies the heck out of the input but strangles it at the same time using feedback. But again, I don't really know if or how D is different from AB wrt noise performance.

At any rate, I don't think I would use a class D chip amp that runs on 5 volts and has digital controls just to use the power amp part of it.

If there were a class D chip amp that same simple input / output / supply as op amp (minus the differential input since presumably the gain would have to be fixed) and it could efficiently put 3 watts into 8 ohms, that would actually be ideal.
 
I am with JD in not understanding your concerns but for audio-phile class D maybe look at Bruno Putzey's designs. I don't know if he makes anything as small as 3 W but they are adequately clean.

You may be investing too much concern about a simple buffer/driver.

JR
 

Latest posts

Back
Top