Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

December 01, 2022, 06:42:05 PM

Login with username, password and session length

Recent Posts


Simple discrete poweramp designs

Started by Rutger, August 18, 2011, 07:31:25 AM

Previous topic - Next topic



I'm looking for simple but effective discrete poweramp designs that can be build easily and perform wel in a guitaramp. I did a view diy-builds but I've never build a poweramp myself, most schematics of poweramps seem way too complicated to start with.

Googling around I find schematics like this:

It comes from an 18W audio amp. But since I haven't much theoretical knowledge, I can't tell wether a design like this perfoms well or is stable enough?
Can you give me some advice?


The most simplest of designs are not often very good performers. This is one example.

For example, in this case what you gain in simplicity in the power amp stage you loose in the power supply: Simple, single-ended amp like this is very prone to power supply noise, hum etc. Basically, it will need a completely regulated power supply to start performing decently - or at least comparably to just a tiny bit more complex designs with differential input and three-stage design with input stage, voltage amp stage and current amp stage. So, you build a simple power amp, that by all measures is not a too good performer to begin with, and in addition you must built a regulator circuit made to withstand quite high currents the whole power amp will be pulling. This will practically double your costs... and for what?

It's a great beginner's project but I have a hunch you'll be tearing it down for parts in just few months once it's finished. That's at least to what I did to about equivalent amp. Torn down just as soon as my thrill of building something that worked and amplified signal wore off (in this case it was about two or three days). Overall it was just an extremely lousy and trouble-prone power amp and likely would have failed sooner or later in more serious use.


Overall, I'd suggest you start with something similar to this, no less:

If you look around you'll find this architecture to be very typical to most MOSFET amps.

In the end, whether you build an extremely simple circuit or something such as this, which is only a tiny bit more complex, your biggest problems will be in the layout, noding, and other similar details that are still equally important for both simple and complex designs. So what you gain by simplicity is really not that much. On the other hand, what you loose by it is. With minor additions circuit like this can be easily scaled from 20 watts to 400, perhaps even higher, which is can not be said about the simple circuit. Even more likely, if you put proper effort into it, the same driver board layout will work in pretty much all versions. I wouldn't even be surprised if one or two board designs for this type of circuit could be found with some persistent Google searching as this architecture is extremely typical for basic MOSFET amps.


Thanks alot for your help and explanation teemuk!
This schematic seems simple enough to handle. Like you said, it looks familiar. And since I like mosfets it's a nice starting point. :)


Googling around I find that Marshall made a couple of nice ss-amps with not too complicated mosfet poweramps :)
Schematics can be found at drtube.com

I could go with one of these, the poweramp of the 5203 for example.
Any tips what would be a great design to start with?


Hi Rutger,
              This may interest you also.


I don't know how good this really is but comes from ESP and Rod E is no mug.
It can run from +/-25V to +/-42V in the low power version.

The hick-up might be the cost of 2SK1058 and 2SJ162 they don't come cheap. :'(

Just throwing it in for observation.

J M Fahey

I would skip the "discrete" specification and go for a tried and true chipamp, LM3886 suggested.
It's simpler than a *real* discrete amplifier, by the way.
(the one you posted originally is even simpler, just it doesn't work ;) )
And you have access to boards, power transformers, heatsinks and a host of people who built them to ask advice from, since it's a very popular amp (for good reason).
If you still want to build a discrete one, good, but it will be definitely more complex, and probably you will have to design and etch your own PCB.
PS: if you need lower power, say 15 to 30W, the TDA2030/2050/LM1875 line  is very good, even simpler, and very tried and true too.
Also check Welleman kits,they offer the PCB plus all parts and good building instructions.


I have to agree that discrete circuits are kinda weak solution for anything below 50 watts. Unless the thing is intented to be a learning project about specifically building discrete power amps I'd just skip to the integrated chips and be happy with that choice.


Thanks guys. I would like to try a discrete design as a learning project, but like everything else it should have a satisfying endresult :)

I've allready built an amp with a LM1875 gainclone-kit. It's a nice little amp and I would like to compare it to a discrete design. Would the difference in quality be great? I mean, the 80's marshall ss-line shouldn't be that great if it hadn't a good design, doesn't it?

J M Fahey

Quoteit should have a satisfying endresult
Then go straight for the chipamps. ;)
Since you have already successfully built an LM1875 amp, your next step should be an LM3886. Go for it.
I've allready built an amp with a LM1875 gainclone-kit. It's a nice little amp and I would like to compare it to a discrete design. Would the difference in quality be great?
No. In fact chipamps are *so* incredibly good, that a guy figured that and became millionaire (well, at least he made *good* money and started a trend) by selling very well made and packaged .... LM3886 based power amps.
And it was no snake oil, they performed incredibly well, rivalling the best designs at Hi Fi shops.
Even corksniffers had to recognize that.
Just google "gainclone" and any reference to the original one.

QuoteI mean, the 80's marshall ss-line shouldn't be that great if it hadn't a good design, doesn't it?
Agree, it did .... but in the 80's chipamps still were at the 15W stage; now Marshall uses chipamps even in its 480W heads ... although they overstretch them too much and the end result was not good.
But if respected Line6, Vox, SWR, etc. use chipamps in their 60 to 100 W amps .... that means something.

I am still using discrete TIP142/147 based 100W amps because I don't trust the TDA7294 which is the popular choice at that power level, and because I sell to faraway places in Argentina and surrounding Countries, so ease of repair by a local Tech is a must, but that will change soon, when chipamps reach the next step.
I trust them up to 70W , definitely not 100W, their case is too small to dissipate so much heat.
When they make a 150W "Hi Fi" rated chipamp, I will trust them for 100W "guitar" amps, simple as that.
For 50/60W go the LM3886 route without a problem.


Okay thanks, thats some good advice. I'll stick to chipamps. :)

It's funny, this forum is probably the only one on which chipamps are adviced over discrete amps. Why is that?


Well, I see only few reasons why not and most of them don't matter a lot in case of building a lowish power amp without too many concerns about its specific architecture.

What is inside that integrated circuit is basically a quite typical discrete circuit but in that integrated form you have much less hassle with components occupying board area / layout, thermal tracking of bias, or just implementing handy features like current limiters, low voltage detectors etc. All which would bloat an ordinary discrete designs massively but fit to a small package in IC form. Basically you get a lot more feature-packed and reliably working amp in a very convenient, small package. What's not good in that?


- You can't tamper with the actual architecture inside the IC. Often this does not really matter, most power amp chips are basic "building blocks" with a rather standard function and things can be altered a lot with the external circuitry. And I mean a lot. Just consider something like OpAmps: The basic architecture and function of them is more or less identical but with proper external circuitry those little things can have thousands and thousands of different applications. You could build discrete OpAmps too but most people admit that in about 99% of applications it wouldn't really make too much sense.

Obviously you can't really make up choices of things like what types of output devices are used eg. MOSFET vs. BJT (aside from specific chip choices that is) but in most cases all that doesn't really matter. But of course, if it does then a chip is not an option.

- Power handling may not be all too impressive. Output power ratings of chips tend to be in the max. 100W ratings and most of those are a quite far stretch, especially if your amplifying highly compressed signals at the level at which the amp wastes the most power (which is usually NOT while the amp is cranked but more or less just average settings). The die area that needs to transfer heat to sink and thus dissipate it is also considerably smaller than what it would be when using many output transistors in parallel for the same task. In other words, this IC stuff may not take the harshest abuse and may not be the best choice for amps with power ratings that should exceed about 80 watts. Again, if you're building something with just moderate output power this does not matter too much, even less if you take a bit of overkill approach to chip selection. For example, I've seen something like LM3886 often used for 30W amps though the chip could easily muster out about a twofold higher output power. In the end, the price difference to a chip rated 30W max. isn't all too big.


Well, there are chips that handle greater power: Class D and Class T designs, T-amps from a company named Tripath. Even the audio snobs have good things to say about them. http://shop.41hz.com/shop/item.asp?itemid=653
Someday I'll try one in a guitar amp build. The low power ones are quite inexpensive.


Most of the high-power TriPath chips are basically just PWM modulators + gate drivers, requiring at least a pair of MOSFETs to do the actual power handling thing. Similarly one can also find driver chips that are fully analog, basically including about everything of a full power amp but the power transistors.