A DIY audio amplifier design

[HUG-1]

Want to learn about electronic components? Want to have the pride of owning something entirely unique, customised exactly as you want it? This thread concerns the making of a DIY (power) amplifier, how it should perform and what to realistically expect for small money.

[HUG-1]

One question asked repeatedly is 'will amplifier X work with Harbeth speakers Y?' You know the answer is 'yes, if the amplifier is competently designed and working within original specification'.
How good does the amp have to be? How little or how much should it cost? How big or how small? How many component parts? Are all amplifiers basically the same under the lid (yes)?

Old habits die hard. The habit of believing that the amp is supremely important (it isn't) will not die. We hope that this thread will give you a grip on the reality of audio amplifiers. They are about as simple an electronic device as you can imagine ....

[A.S.]

Interesting idea! OK, where to start. Let's start with a bold statement, which if you can disprove, you can win FOC a pair of brand new M40.1 (see here).

1) "We absolutely agree (and hear for ourselves) that if you randomly select any competently designed stereo audio amplifiers working within their original performance specification* (and obviously of enough power output**), and hook them to your Harbeth speakers you will hear a difference between them. That's absolutely assured and expected..."

2) "When precautions are taken to turn down the more 'gainy' (that is, sounds louder) amplifier to match the gain of the quieter amplifier, the big differences that you truly could hear under 1) above diminish or disappear on an instantaneous A-B switchover."

3) As there are no standards for the gain (loudness) of amplifiers at the loudspeakers relating to the marking of the volume control knob on the front panel, you cannot be sure how many volts any amp is producing for any volume control position on the front panel. For example, amp A at volume setting 8 produces 18w output; amplifier B produces 41w at volume setting 8 and amp C, which doesn't have a volume control, produces 110w from a fully driven input stage. It is impossible to compare the sound of these without reducing (in this example) amp B to volume setting 3 (to match A's 18w output) and measuring the output of amp C. Who would think to do that?

Can amplifiers be level matched by ear? Absolutely not! That's why this debate will not die. The failure of the ear as a precision instrument is the source of the problem! You can't use your ear to validate your ear!

Maybe the conclusion of this thread will be that whilst it is entirely possible to make a perfectly functional, perfectly reliable power amp that will drive any Harbeth speaker for the cost of a meal out - and of adequate sonic performance - all of us would prefer the pride of ownership of a professionally made amplifier sitting with our audio equipment, not a DIY lash up. And that is all the excuse you need to invest in a really well made amp, not any considerations of 'audio quality'. Who would want a kit-car when they could own a Bentley?

* Obviously the design has to be thorough and professional. If the design was poor and the amp is unstable, has whistles or distortion issues, and has a frequency response which is not basically flat then it is not a competent audio design or it has slipped out of original specification and needs servicing.

** It's clear that a flea cannot pull a horse. We cannot meaningfully compare a 7W tube amp with a 100W solid state amp when driving a real loudspeaker.

[A.S.]

...Maybe the conclusion of this thread will be that whilst it is entirely possible to make a perfectly functional, perfectly reliable power amp that will drive any Harbeth speaker for the cost of a meal out - and of adequate sonic performance - all of us would prefer the pride of ownership of a professionally made amplifier sitting with our audio equipment, not a DIY lash up....

We should consider the beginner audiophile. Someone like me when I started out. I had the choice of DIY (that was my only choice due to fund issues), or daydreaming of Leak, Quad, Ferrograph etc.. My first three amps were kits - I wish I'd kept them - but one I found pinned up in a museum display (picture to follow).

I stress again .... in my opinion there are much more relevant and important factors in the selection of a hifi amplifier than any claims you will read about 'sound quality'. You are far better to select on the basis of styling, power, brand image and reputation, after-care, user facilities and value for money. These factors are important.

Nobody here can ever factually answer the question 'does amp A sound better than amp B' so why ask that question? The answer will be as vague and personal as asking 'do blondes make better lovers?'. No one can answer that objectively: they can only answer subjectively and that is not an answer, it's merely an opinion. They can answer the question 'is amp A reliable? What after care did you receive from amp B's maker? Can you comment on the styling of amp C? What's the remote control on amp D like? How hot does amp E run? Does F have enough inputs? How heavy is it? Does it hum?'

So, what does today's DIYer have available as an amp kit? Well, for the price of a few beers, an extremely well presented little amp kit ....

More follows

[paolo]

So, what does today's DIYer have available as an amp kit? Well, for the price of a few beers, an extremely well presented little amp kit ....

More follows

My recommendations are:

1. the Zen:circuit by Nelson Pass, comprising single gain stage and easy to construct but a beast of a heat sink. Pictures will follow (not so much a DIY kit but a set of plans)

2. The 46 SET Tube Amp using Tungsol 46s. Comes as a kit from Thomas Mayer, but you need to be aware you are dealing with high voltage!! Ouch. For the fainthearted it comes fully assembled (I am definitely in latter category) In finished form it weighs around 25kg and delivers no more than 2W per channel.

P3ESRs and the 46 SET are magical.

[A.S.]

Here is an excellent little kit amp. Everything you need to make a complete and very high quality monoblock power amp (two needed for stereo). 100W output (with suitable cooling). Flat frequency response. Inaudible hiss. Very low distortion.

It works. Who would take any pride in owning such a DIY box when they could buy a beautifully made amp in a hifi store with proper after care? And that's all the justification you need to buy a proper amp. But in sonic performance terms, can this amp actually be distinguished?

>

[jr_41]

AMB Laboratories is a popular site with DIY headphone enthusiasts. Many of Ti's designs can be adapted to drive speakers instead (though the output is on the low side), and he does provide a power amplifier design.

http://www.amb.org/audio/

I built the AMB M³ amplifier and σ11 PSU to drive headphones, but it can be configured as a 6Wrms per channel (into 8Ω) power amplifier instead. The designs are all free.

Building this project taught me a lot about analogue electronics!

[Spindrift]

Here is an excellent little kit amp. Everything you need to make a complete and very high quality monoblock power amp (two needed for stereo). 100W output (with suitable cooling). Flat frequency response. Inaudible hiss. Very low distortion.

It works. Who would take any pride in owning such a DIY box when they could buy a beautifully made amp in a hifi store with proper after care? And that's all the justification you need to buy a proper amp. But in sonic performance terms, can this amp actually be distinguished?

>

Interesting.
I see no issue in using a DIY amp if it performs well, is reliable and built with care (esp. the solderings and internal fixtures). In fact, a tiny mono amp (as it appears to be) can easily be hidden behind the curtains or amongst furniture.

Is it for sale as such or is it a collection of selected parts/units from Farnell, RS or a similar parts supplier?

[jr_41]

Here is an excellent little kit amp. Everything you need to make a complete and very high quality monoblock power amp (two needed for stereo). 100W output (with suitable cooling). Flat frequency response. Inaudible hiss. Very low distortion.
>

On second examination of that photo there doesn't seem quite enough to build an amplifier, at least safely. Also, why is there a lemon in the photo? Am I missing something - is this a joke?

[A.S.]

To reply to the previous two posts:

1) The complete amp kit is available on the internet as a stereo pair not including the metal box.
2) I count 22 electrical components plus one pcb plus three connectors. The pcb is highly recommended but not essential. The three connectors (blue, green) are sensible for testing. Not shown is a mains input socket. The speaker can be directly wired to the blue socket or a pair of proper speaker sockets connected to the blue socket (or wired directly to the pcb).

The only electrical component that is optional is the LED. Deleting the LED means that there would be 21 essential electrical components. If you reduce the count to 20, the amp will not function at all. If you add one, two, ten, one hundred more components the amp will not function any better. There are no circuit places that would benefit from the addition of even one component. It might look fancier and more sophisticated, but for this all-in-one IC chip, this is really all you need. The mains toroidal transformer (right) and the two black capacitors and the 4-wire rectifier (total 4 parts) could be common to a stereo amp.

Note: If you were to make a stereo kit with single LED, you would need 21 + (21-4) + 1 (LED) = 39 components and not one more. None of these electrical parts have a 'personality' and this circuit could (as a generalisation) be found in a UPS or printer.

Photographed on the kitchen table, the lemon is for scale.

[jr_41]

It was the lack of a mains connector, fuse (though one is usually built into the mains cable, but probably not of a rating suitable to protect the amplifier from overload) and means of earthing the metal case which I was alarmed to find missing from a kit, but in the context of your argument this is being picky.

One thing though, although toroidal transformers are better at reducing EMF, would having the AC section (i.e. the PSU) so close to the signal input in that single, small case not introduce an audible 50hz hum/noise floor? Although again, this could potentially be argued as not relevant to this argument, is it not these kind of considerations which is increase the complexity and cost of an amplifier, at least to a degree?

[A.S.]

Quick answer:

Picture attached of a very successful active speaker module we made some years ago using this IC power amp. The bass section was driven as push-pull; >100W power output.

Marked IEC connector with external user replaceable fuse - star electrical and safety earth - and very close proximity of toroidal transformer to the 3 ICs. Even much closer than shown here, there was no audible hum induction on the output to the woofer (it was extremely quiet to my surprise).

First rate power amps do not need to be complex, heavy or expensive. Nor can they be 'voiced' except by deliberately tweaking the frequency response and/or poor layout design and/or degrading the feedback or temperature stability - all bad practice. Power amps are extremely simple devices: when they work they work.

>

[jr_41]

Impressively simple! If this is all that's required to take on any power amplifier, disregarding power output (though perhaps additional complexity is required to achieve satisfactory signal reproduction at > 100W?), and render the listener unable to distinguish A or B - assuming the levels are matched - then this is genuinely a revelation (at least to people like me - who may have at least a degree of understanding of the theory, but don't have the experience or a true grasp), not to mention you are doing our wallets a great service!

Are we talking purely about power amplifiers here? In point 3 of your post above you mention volume control. If we're talking about volume control etc. then are we not into the realm of integrated or pre/power amplifiers? To continue in the vein of keeping things simple, shall we say that an integrated or pre/power amp adds signal gain control and perhaps input selection? A power amplifier alone is of limited value in the real world, and in my experience, when audio amplification is mentioned it usually infers both. In what way do you consider the addition of these features impacts your argument?

I also note you mention an instantaneous A-B switchover, is there a particular reason for this?

[A.S.]

Impressively simple! If this is all that's required to take on any power amplifier, disregarding power output (though perhaps additional complexity is required to achieve satisfactory signal reproduction at > 100W?), and render the listener unable to distinguish A or B - assuming the levels are matched - then this is genuinely a revelation (at least to people like me ...

Thanks for the feedback. To answer you ...

1.) Yes, it really is that simple. But it can be even simpler. If the sales volume of the amplifier is expected to be sufficiently high, you as the manufacturer could approach the chip maker and ask them if they would 'sweep-up' as many of those twenty-odd electrical components into the chip itself. There would be a trade-off. Those blue and orange two-legged capacitors are, as you can see, rather large proportionate to the IC itself. They would be difficult to embed into the IC package unless it was made much bigger, which would add more cost than the deletion of the external capacitors themselves; that's a no-no. So, most likely, the IC people would offer a compromise where the power output was down rated from, say, 100W to 50W, some restrictions on how hot it could get (etc. etc.) and the super-integrated result would be available.

If you look inside any modern miniature audio equipment (MP3 player, iPod etc.) you will see that there is a very high degree of 'integration', which is a fancy way of saying that there are ICs but not much 'glue' (external components) like capacitors, resistors and so on. This saves cost, saves assembly time and above all, increases reliability. IC are fantastically reliable and will work for decades as we know. Capacitors are dirt cheap and have a very short (and temperature related) lifespan counted in thousands of hours. Getting rid of capacitors in any (high powered) electronic circuit is the first step to improving reliability. The primary durability-limiting component in a power amplifier is the power supply capacitors (the big black cans) as they are working hard, in a hot environment.

2.) The limiting parameter for the power IC is - perhaps surprisingly - simply that of heat dissipation. 100w means 100w - would you want to hold a 100w light bulb just turned off? It's a lot of heat. It melts plastic. So if the heat sink (attached) is big enough, or force cooled (noisy fan? popular with PA stage amplifiers) can carry that heat from the chip it will run at 100W day in, day out.

3.) Yes, that little kit is all you need for a power amplifier. If you only intend to drive it from a 'high level' source such as a typical CD player (2v output) then you could directly connect the CD to the input of the kit. If the CD player has a (remote) volume control, then that's all you need to adjust loudness.

If the CD player does not have a volume control then you have three choices:

A) By a Passive Volume Control and connect the CD to its input and the output to the amp. This is what I use. It is nothing more than a volume control (see picture, it's green) in a nice box: no other components at all.

B) Obtain a Pre-Amplifier with multiple inputs, volume control, ideally tone controls and connect its output to the input of the kit. This may also have a phono input. It may have adjustable in/out gains, as the QUAD 44 does.

c) Add your own volume control and/or input selector to the kit amp! The picture shows what's needed: a volume control and knob, and a switch (rather ugly) to change between two input sources (which in this really fancy arrangement operates the two yellow relays). Finish the kit off with a pair of speaker connectors and some nice quality phono inputs.

Personally, at this stage of my audio life, I would far rather go and buy a well made amplifier with a first-class after care "buy once and forget" than make my own. Under the lid it may be the very same components but what I'm willing to pay for is someone else to take responsibility for servicing it - and something that looks well made - rather than having to roll-up my own sleeves. I'm sure you feel the same. The essential issue is that none of these basic parts have or can have a 'personality'.

The A-B question has been covered at length here I think.

>

[jr_41]

Interesting stuff, particularly the 'integration' concept. Thanks. I hope someone takes up your challenge and that you're able to prove your point. It is tempting to build one of these amplifiers and see how it compares to one of the 'Hi-Fi' offerings, and whether it's possible to distinguish them. You have demonstrated how simple an amplifier is at it's fundamental level with the use of ICs, which in itself is an eye opener.

[A.S.]

Interesting stuff, particularly the 'integration' concept...

I was rooting around in the garage this morning and I found what is surely the last word in 'integration'.

ILP, a British company, took an audio amplifier idea not dissimilar to the kit (previous posts) and potted it into a cavity in an enlarged die-cast heat sink. That means, after they tested the assembled circuit board, they poured a thermo-setting epoxy over it to completely seal it from the atmosphere. So what we have here is an even more highly 'integrated' power amplifier. I'm not sure if it's still available, but I believe that they sold tens of thousands of these over the years. We considered using some in active speakers but the heat sink protruded too far from the back panel of the speaker and would have been damaged in transit or use.

All we need is to provide a safe mains input socket which we solder directly to the transformer's brown and blue wires (the mains switch built into the IEC socket which also has a non-essential mains filter) and solder the red, black, yellow and orange transformer wires directly to pins on the 'pod'. The smoothing capacitors are now 'potted' inside the pod.

We just need a volume control, a phono input and a speaker connector (and ideally a case) and we have a complete mono amp. Not one single additional part is required.

I have no idea if this amp (or indeed the kit) would pass the A-B instantaneous switch over test. It may or it may not; it is most likely to depend on the acuity of the listener. But whether or not, what will come out of the kit or this super-integrated amp will be recognisably good music. The objective of this exercise is to re-emphasise that whilst this sort of DIY solution will 'do the job' there is surely no pride of ownership, and when the audiophile has the funds available, he'd be better to invest in something with a decent cosmetic and above all, minimum risk that it will expectedly fail and destroy his expensive speakers.

That durability, whilst not an attractive marketing buzz word, is IMHO the most important parameter of any power amplifier capable of destroying expensive speakers - as they all are.

>

[jr_41]

Now that's what I call a black box solution!

A fear of damaging expensive equipment (or yourself!) through tinkering with DIY if you're not completely confident is an issue. I think a pair of cheap speakers - to at least check the amplifier is basically working - is a good idea, not to mention a basic understanding of mains electricity and safety best practice.

Returning to my previous argument for a moment, I note that in your cross over designs you use a very high quality PCB (with gold/gold plated tracing?), high quality film capacitors, etc., and in the P3ESR SE, oxygen free copper wiring. Am I correct in understanding these high quality variants were chosen because they're in the signal path and provide an audible improvement? In the kits listed above only the cheapest variants are used (that potentiometer is a far cry from the Alps Blue Velvet you typically find in a hifi amplifier). Given an amplifier is also in the signal path, is the use of higher quality components and wiring not justified there as well?

I'm not sure which if any of the caps in the kit are actually in the signal path (as opposed to the large capacity electrolytic caps which are used in the PSU section and entirely appropriate), but I've read on a number of occasions that electrolytic and ceramic caps are a noticeable degradation vs film caps.

[Pluto]

Not nearly as tightly "integrated" as the ILP module shown above, but have you seen these modules? Top notch British design and manufacture, brilliant value for money.

And they seem to sound rather good too...(OK, I'll get my coat...)

[A.S.]

...I note that in your cross over designs you use a very high quality PCB (with gold/gold plated tracing?), high quality film capacitors, etc., and in the P3ESR SE, oxygen free copper wiring. Am I correct in understanding these high quality variants were chosen because they're in the signal path and provide an audible improvement? In the kits listed above only the cheapest variants are used (that potentiometer is a far cry from the Alps Blue Velvet you typically find in a hifi amplifier). Given an amplifier is also in the signal path, is the use of higher quality components and wiring not justified there as well?

Er ..... never crossed my mind to use, or even bother to investigate the use of 'exotic components'. I'm far too experienced in the art of the marketeer to be sucked into that bog! Needless cost for no benefit = bad engineering.

Our philosophy here is simple: We do not see the provision of spare parts/servicing as profit centre in our company (as most companies do) - we see it as a cost centre. This is a subtle difference.

Profit centre: After care/upgrades are run as a for profit activity and they can be the most profitable department in a company, far more than first-time sales as the amp upgraders know.

Cost centre: The company does not have a dedicated service/upgrade department with expensive staff waiting like vulture. It supplies service parts and takes the hit against general overheads.

We absolutely do not want to be diverting time from hard-pressed production staff to resolving avoidable* customer issues. Every second that is taken off production can never be compensated by the meagre profit on spares (when they are processed at normal prices, not as a super-profitable activity), so we over-engineer to virtually guarantee that what we sell never comes back. We do not plan to get two bites at the cherry. That's the primary thought behind designs here: we never set cost minimisation as a goal: the speaker will cost what it costs and that not negotiable internally. You would be horrified if you could see what we discard (I am) to keep up the standard.

Electrolytics: you may have seen my analysis of these elsewhere here (where???). I'm on a mission to eliminate them.

'Exotic' parts in amplifiers: do the electrons flowing around know that they are in a piece of fancy cable or a swish capacitor? I don't believe that they do. As I've said here before, the use of fancy cable is entirely for marketing demands. Despite our well known position, many markets insist on exotic cable in their speakers and they are willing to pay for it.

*Some customer issues (amplifier failure, child crushes supertweeter) are not avoidable, they are bad luck: we must get those customers up and running as soon as possible. But what we make and leaves our door we have responsibility over and can minimise the risk of failure due to parts/manufacturing by using best quality materials and careful selection and testing.

[jplaurel]

I'm very pleased to see this thread on the Harbeth User Group. Building your own audio gear is a lot of fun and very rewarding. Believe it or not, with little more than a good multimeter, a soldering station and some basic tools, you can easily build a great-sounding amplifier. If you're looking for a way to get off the audio nervosa merry-go-round, DIY is a good way to do it. And as a side benefit, you get to learn something new! How cool is that?

This is my DIY chip amp based on the National LM3876 using only 12 components per channel, including the power supply. It's built onto an old cutting board for now, using a piece of aluminum stock for the chassis and heat sink. I've ordered some proper heat sinks for it and am building a nice aluminum chassis for it using scraps from a local metal supplier. Sounds very nice with the Harbeth P3ESR and took just a few hours to assemble.

http://gallery.me.com/jplaurel/10015...13339172430001

Here's the chip amp hooked up to the Harbeth P3ESR speakers, using the Bel Canto integrated above as a preamp. I'm sure Alan will be pleased to note the zip cord speaker cable!
http://gallery.me.com/jplaurel/10015...13339172550001

One thing I should mention is that when you build these things, be sure to test them on some old speakers before you connect them to you valuable Harbeths. I keep an old pair of Radio Shack Minimus 5 speakers in my shop for just this purpose.