2.4KW amp, anyone?

[fas42]

Let’s get back on topic Frank. Which amp are you designing first? When will you have the schematic finished? When will you buy the parts and start assembly? When can we expect to see the first finished product?

As I mentioned earlier, I aim to have a design that suits the maximum output unit first; it's always easier to scale down rather than the other way round. The overall concept is one that's been floating around for a while, and in part triggered because so many commercial units don't appear to have the real grunt that's required. I don't mean here to dispute that that they can go loud, rather that there is always power to spare, and that they should always appear to be able to do the job effortlessly, with no apparent strain. As I mentioned once, many years ago I heard the top of the line mbl units, and initially they appeared to do well when playing but then the subtle signs started to intrude that indicate that they weren't totally on top of the game ...

The key area of the design initally is have active devices, transistors or equivalent, that can handle the stresses involved of voltage, amps and power dissipation, which means heat. A straight schematic is not the answer, the actual construction in 3D space will be absolutely crucial in making this fly, especially if I don't want it to be hideous expensive. The McIntosh MC2KW is total overkill in terms of materials assembled to achieve a technical result, which is not what I'm after.

So a schematic design has to go hand in hand with how you envisage the parts will be organised on a chassis. The theoretically ideal parts may be a bad choice because actually connecting them together involves awkwardness in the physical positioning of the bits, and makes it all too expensive.

There are valid reasons for people saying that making a good, big amp is hard, and a key one is that you misstep in how it's all assembled. Get this wrong, and even with a great schematic, the end product will be a dud. Remember, I have always said that getting the small details right is crucial, and doing a big amp makes this philosophy even more pertinent ...

So, the final answer is, it will take as long as it takes: I'm interested in a product that pleases and satisfies me, not one that just technically does some of the things right.

Frank

 

[tony ky ma]

No worries, Frank, for all you know I am a mail-order engineer without a real eddication in any country. Ah, such is the power of the net...

As a Diyer has to say something here in a diy forum, I am done with mechanical engineer in Japan but had been teach the most of general engineering subjects (EE Chem etc. no computer at that year) in the first two years before get in to the main, when you want more knowledge for your hobby that you can learn by yourself after school , you don't need certificate for hobby.
The most difficulty in diy's project is collecting all the parts, when you got a new idea, you have try to get what you can get within your budget before to design,
There are two kind of diyers , one is for save money cheaper than market, the other is looking for something non in the market. I am the later one, as it is non in the world you don't know what will happen till it being done, like the tube amp I did with separated the out-put transformer from the out-put tube into another case link with longer cable instead the use of speaker's cable, it will come out problems of safety , cable's insulation and connector is the issue (800v, 100ma), what is the advantage in this way ? you will not know till the amp is done and ABX for it. having all the results in hand then putting them to forum for discussion, lot of people in forum against this idea in safety, it doesn't matter, I know it will not fit all people, but so far so good for 3 years in using, how better than use of speaker cable ? I only know in my system, in other systems I don't. because this is only one part of the whole system needs the similar level quality of the whole system to show out the difference in ABX
tony ma

 

[fas42]

There are two kind of diyers , one is for save money cheaper than market, the other is looking for something non in the market. I am the later one, as it is non in the world you don't know what will happen till it being done, like the tube amp I did with separated the out-put transformer from the out-put tube into another case link with longer cable instead the use of speaker's cable, it will come out problems of safety , cable's insulation and connector is the issue (800v, 100ma), what is the advantage in this way ? you will not know till the amp is done and ABX for it. having all the results in hand then putting them to forum for discussion, lot of people in forum against this idea in safety, it doesn't matter, I know it will not fit all people, but so far so good for 3 years in using, how better than use of speaker cable ? I only know in my system, in other systems I don't. because this is only one part of the whole system needs the similar level quality of the whole system to show out the difference in ABX
tony ma

I like your philosophies and attitude a lot, Tony, especially the concept of outboarding the output transformer -- safety's for wimps!! (I'm only joking, people ...)

I'm obviously looking to do something not on the market, at least not with the right combination of attributes ...

As a side thought, if people want to understand a bit more where I'm coming from, I'm a great fan of the programming language Perl and the philosophies underlying it. Larry Wall is my kind of guy, from his Wikipedia piece:

In the second edition of Programming Perl, Wall (along with coauthors Randal L. Schwartz and Tom Christiansen) outlined the Three Virtues of a Programmer:

1. Laziness - The quality that makes you go to great effort to reduce overall energy expenditure. It makes you write labor-saving programs that other people will find useful, and document what you wrote so you don't have to answer so many questions about it. Hence, the first great virtue of a programmer. Also hence, this book. See also impatience and hubris.
2. Impatience - The anger you feel when the computer is being lazy. This makes you write programs that don't just react to your needs, but actually anticipate them. Or at least pretend to. Hence, the second great virtue of a programmer. See also laziness and hubris.
3. Hubris - Excessive pride, the sort of thing Zeus zaps you for. Also the quality that makes you write (and maintain) programs that other people won't want to say bad things about. Hence, the third great virtue of a programmer. See also laziness and impatience.

Just translate the above to audio meaningful terms, as a gentle guide to my approach ...

Frank

 

[fas42]

Only fair to comment here now, in spite of what I said earlier, and this is only because Gary's comments in the Class D thread stirred up my interest in such matters again, I might be tempted to be lazy, as mentioned above, and go the UcD route. I just discovered Bruno now has a Ucd2k module available, and its capabilities look very promising. It is also very eco nice and all that sort of thing ...

Gary was happy with his efforts in this direction, so the Hypex solution should be able to do the job ...

Again, thoughts?

Frank

 

[mep]

My thoughts are you should have conviction of your direction based on your knowledge and not which way the forum winds are blowing.

 

[fas42]

My thoughts are you should have conviction of your direction based on your knowledge and not which way the forum winds are blowing.

Fair enough, but I'm after an elegant solution to the problem of a high powered amp, that is still "musical", and doesn't cost silly money and effort to build. The actual amplification side of things, for me, is not so important: I know this is back to front as compared to most people's thinking, but here my knowledge and experience serve me well. The experience with the HT has demonstrated that almost anything that amplifies is OK, provided you worry about all the other bits and pieces, like power supplies, interference suppression and contruction techniques.

The one thing with class D is possible high frequency interference from the nature of its operation disturbing sound quality, but by all accounts by others Bruno has this largely under control. I'm prepared to throw any level of shielding, etc, at it to get that under control, I feel this will be less work than what's needed to assemble using linear technology.

If nothing else, it will be a lot lighter to lug around!!

Frank

 

[garylkoh]

Fair enough, but I'm after an elegant solution to the problem of a high powered amp, that is still "musical", and doesn't cost silly money and effort to build. The actual amplification side of things, for me, is not so important: I know this is back to front as compared to most people's thinking, but here my knowledge and experience serve me well. The experience with the HT has demonstrated that almost anything that amplifies is OK, provided you worry about all the other bits and pieces, like power supplies, interference suppression and contruction techniques.

The one thing with class D is possible high frequency interference from the nature of its operation disturbing sound quality, but by all accounts by others Bruno has this largely under control. I'm prepared to throw any level of shielding, etc, at it to get that under control, I feel this will be less work than what's needed to assemble using linear technology.

If nothing else, it will be a lot lighter to lug around!!

Frank

Frank,

Class D isn't easy - even with the Hypex modules.... and it's not the usual suspects. Like you said, Bruno has the EMI/RFI emission largely under control. To make it actually requires effort. I think that Class D has very low PSRR - translated to mean that the design decisions made in the power supply end up being very audible - more audible than Class A or Class AB designs, and much more audible than tube amps.

I don't know if it will be a lot lighter to lug around. I couldn't get a SMPS to work musically, and in my amp using the UcD400 module, I ended up with a transformer weighing 18kg.

 

[fas42]

Frank,

Class D isn't easy - even with the Hypex modules.... and it's not the usual suspects. Like you said, Bruno has the EMI/RFI emission largely under control. To make it actually requires effort. I think that Class D has very low PSRR - translated to mean that the design decisions made in the power supply end up being very audible - more audible than Class A or Class AB designs, and much more audible than tube amps.

I don't know if it will be a lot lighter to lug around. I couldn't get a SMPS to work musically, and in my amp using the UcD400 module, I ended up with a transformer weighing 18kg.

Gary, thanks for your thoughts. Yes, PSRR is one thing that I've noted, from specs at least, is suspect in class D so a high level of attention is needed here; when I did my DIY gainclone I used an extemely tightly controlled, regulated power supply, for substantial audible benefit, so at least I have good experience here ...

Also agree about SMPS, 2 lots of digital may allow for a cute chassis, but isolating the digital workings from the musical signals only becomes harder.

Ultimately the design, as always in such cases, will be a balance of compromises, including that of weight. If it needs weight to get the job done, so be it. What can't be compromised is the firstly the SQ, and then the power: it must be able to achieve full wattage output with no loss of fidelity ...

Again, your input is much appreciated ...

Oh, last thought, there weren't any nasty surprises with the UcD modules by any chance? In other words, something you assumed or were led to believe would not be a problem or would perform up to a standard, but it turned out otherwise?

Frank

 

[fas42]

Just to let people know that things are still happening, the brain juices are cooking at the moment ...

It's obviously not a trivial exercise doing an amp capable of high output: one way is to just throw a lot of money at the concept and use brute force engineering to get the job done. This is how the McIntosh MC2KW was built, and I'm not going down that road. At the moment I'm working on getting some of the key elements of the design decided on, and this splits naturally into 2 areas.

First is the power supply which are far as I'm concerned is the key to the whole exercise: if it's not done right then the end machine will be a dud, and not worth the effort. And so far it's either going to be a conventional affair in the style of Krell, without going to the extremes of custom transformers, just off the shelf parts intelligently hooked together; or, a custom switching power supply setup. Off the shelf SMPS's are too expensive, and not suitable in terms of output. Even though Gary doesn't like switchers it can be done properly, the Chord amps from the UK proved this, and if you go into the design aspects there are all sorts of advantages in being able to handle fairly precisely and in a straightforward manner how the interference artifacts are minimised. Also it will be lighter amd more eco friendly! The big downside is that a significant more effort needs to go into getting the initial design right, I'm not familiar with SMPS. But, in contrast, getting a conventional supply to always behave itself could end up being just as much a time consumer: I haven't done enough yet to decide ...

The second bit is the amplification stuff, obviously enough, and here my thinking will be back to front for most people. Unlike most, I'm a fan of feedback, properly done, and both the DIY gainclone and HT box use this technique at the heart of their functioning, and have proved themselves to me. So the real concept of the 2400W amp is a high quality opamp, with tons of grunt, a big mutha of a gainclone so to speak.

I can hear a lot of you out there crying, "Disaster!! Opamps and FB are worse than climate change for a healthy world, and a decent listening session!". Well, I beg to differ: every time you listen to a recording now you're listening to the sound of such technology, it's all about doing it properly. Some very highly regarded amps have used this technique over the years, but it's not politically correct, so they stay in the background of people's thinking. Meridian, Musical Fidelity, Eidetic in Australia have all used it, and I'm sure there are others -- implementation is everything ...

So one way of looking at things is to say, select an opamp that sounds right to you and then hang a voltage and current multiplier to the the back of it, and let FB take care of the audible misbehaviour of the output stage. If this is done correctly then the output stage used becomes irrelevant, it could be an array of gainclones, MOSFET, BJT, class D and it all sounds the same: the signature, or sound of the amp is not that of the power devices, but that of the opamp. Which, of course, should be "transparent", ie, no personality. The opamp, or equivalent, is the master, the output stage is the slave ...

In that sense I'm comparing all of these options: there is where the next quandry arises, dealing with the voltage swing. I'm after 2.4kW into 8 ohms, that translates to a 400V swing peak to peak. Or, 200V power supply rails. A lot of the "correct" audio output devices won't handle this, the ON BJT's only go up to 350V. Almost, but not quite there! So if I want to use a conventional output stage arrangement then I will need to use non "audio" devices, more effort will be needed to make them behave in a reasonably "linear" way. Or I could go to a bridge solution ...

Alternatively, I could go for a effectively completely modular output solution, using assembled class D circuits or power IC units. The problem of voltage swing is even more exacerbated here, and to solve that one has to do, at least as one solution, what has been called a bridged bridge topology. That is, taking 2 amp circuits, bridging them, doing a second such and bridging that combination. Then things get a little tricky, you need floating power supplies and the inputs need to be floated, things like transformer coupling likely come into the picture.

All right, that's enough for now! It gives anyone interested an idea where my thoughts are heading, it's still all about making major design decisions ...

Frank

 

[JackD201]

Drat! I thought there'd be a bread board pic by now! Just kidding Frankie

- Rude Boy

 

[DonH50]

A bootstrap cascode design would help with the voltage swing, though of course adds other issues...

 

[fas42]

A bootstrap cascode design would help with the voltage swing, though of course adds other issues...

Not familiar with that one, Don. Just Googled, and very little there, and it seems to be all about the input side of things. Have you seen or used this in an output stage?

Frank

 

[DonH50]

Yes but it was painful. The concept is the same but for the swings you are contemplating it is much harder. Look up active cascode and there might be a little more, though that term is now defined as using an opamp around the cascode devices. Some of my old reference texts include bootstrapping for higher voltage swing but it can be a perilous undertaking and I am not sure I would willingly do it again. I was forced into a few times, just once (I think) for a power amp, but more often to handle higher voltage than the latest IC process would allow...

One issue with Class D will be getting HV devices that switch quickly enough; those are opposites from a device physics standpoint. Maybe an interleaved design? Actually, a bridged design would be the easiest, but of course has its own issues... Always a compromise!

 

[fas42]

If I go class D I'll be taking the easy way out, and using high power Hypex modules. Again, it's all about getting power supplies and the fine details right. Unfortunately, this way is also going to require a bridged bridge solution so may not be worth it ...

I've just come across stacked output devices, this works for bipolar, does it fly for MOSFET? Any experience here? I'll have a look at the active cascode references, thanks.

Might just add a bit here about distortion. I want the beast to both make beautiful music and appease Tim, so technically should perform. To be happy about it, and to make sure that the amp is not contributing to any obvious audible distortion, it needs to be at least 0.001% at 1W over the full spectrum, 20-20,000Hz. Reason for that is that if it is hooked up to a high sensitivity speaker, I'll use 100dB/1m as a reasonable benchmark, and you feed it 1W of something the theoretical distortion is 100dB down, in every way inaudible. Now, I'll be happy with that same, 0.001% distortion figure as you up the power, 10W, 100W, 1000W, because you will only need to use such levels on inefficient speakers, which means the distortion component is automatically much lower in volume.

Frank

 

[fas42]

A minor update: had a long chat to an audio friend yesterday, who's an electrostatic's nut, wants to build his own "one day". He's flat out rebuilding his home at the moment, all on his lonesome, and hasn't touched audio gear in years. But he's got a a big stash of interesting stuff in storage, deep storage, to keep the dust away, and this includes Martin Logan CLS IIa's. I thought he said Ia's, but I might be mistaken or he might be mistaken: looking up the net, the real models are I, and IIa.

The point being, he's extremely interested in how the 2400W amp would go with these, we've agreed that he'll drag them out first thing when the prototype amp gets going, to see what they can do. His audio epiphany was hearing Gryphon amps driving the same brand (I think), and he is also a perfectionist by nature, so his thoughts should be an excellent guide as to what standard my beast comes up to ...

I notice from the web that they sound a pretty challenging speaker to drive, the specs say 1 ohm at 20kHz, others say the bass needs lots of current. Any other thoughts on such from people here, and also, seeing that they haven't been driven in years, what should be checked or even fixed before sending power through?

Frank

 

[mep]

If you ever really build this contraption, you better hook it up to a pair of real cheap speakers first before you fry an expensive pair.

 

[fas42]

If you ever really build this contraption, you better hook it up to a pair of real cheap speakers first before you fry an expensive pair.

SOP ...

Frank

 

[fas42]

And a little bit more: I'm totally in tune with Bob Cordell's thinking; his approach and style are extremely agreeable to me. Would you believe he got 0.001% distortion at 20kHz at full power with a MOSFET amp back in '83? Gee, we've come a long way since then ...

Frank

 

[Whatmore]

2.4kw, is that the best you can do Frank?

That's not an amp, THIS is an amp

 

[fas42]

Well, for a start the Earthquake brand of gear is designed for SPL competitions, and maximum gee whiz factor in young fellas' cars: I don't think they'll do too brilliantly hooked up to high end speakers ...

Plus, they're class J, a variant of class D, light as a feather for what they're doing. But the most telling factor is that they're talking about a 1 ohm load, and when bridged; once you drop back to it being a normal amp driving a normal 4 ohm speaker load you've suddenly got a miserable 900W -- gee, what happened to having a decent voltage swing??

Frank