Got huge heatsinks - which amp to build ?

[firestorm]

I'm looking for a decent amp to build. I use a Marantz PM17, source is a Squeezebox 3, speakers Tannoy 611. Room is 3.5 x 5m (10 x 15 feet app). My metalwork skills are basic. My electronic skills are moderate. Little HV experience. But my expectations are high ;-)

The amplifier that has impressed me most in the last few years are a WAD KT88 pp, 300B pp, Mr Liang 845 SE, EL84 pp, Simaudio Moon i5. They all had something to offer, over and above my previous amps. Nad 3030, Mission Cyrus 2, Naim NAP110, Avondale 260z, Marantz PM68 and currently a PM17. I bought and sold amps for because of my budgetary status and because, well because I could. The 260z was built as a partial kit, most of the hard work was done, like the holes in the chasis and heatsinks.

What don't I like about the PM17 ?
Sounds poor at low volume. Piano doesn't sound enough like a piano, I can't hear the notes decay, there no enough air or ambience. The soundstage is not well defined,seems to move. Not as good as the I5 or any of the valve amps I've listed above.

What do I like about the PM17 ?
Its well built. No hum. Looks good. Has a remote. Sounds better than the Naim kit I had. Sounds better than all the midfi stuff I've listened to before. Wife likes it.

So what do you think ? Nelson Pass F3, F5 ? JLH 1996 edition ? NP version of the JLH, the PLH ? The DX amp ? Graham Maynard's GEM ? Moskido ? SKA amp?

Want soul, body, want something that'll impart the music, not just the soundstage, not just the bass slam or the spittle flying past the microphone.

I'd like to try my hand at something that is relatively simple to build, that can stay in the living room and sounds at least as good as the I5. If I had the money and time, I wouldn't mind trying all of them, at the same time. I know each persons opinions are different, but I'd like to draw from the collective experience - the hive mind if you will.

Sorry for the title, I needed something that'll standout, my previous post was hindered by a dry title ... So I thought I try something more dynamic.

http://www.diyaudio.com/forums/solid-state/171614-estimating-heatsink-thermal-resistance.html

 

[sekhar]

u can try dr boras amp they r real good easy 2 build and they sound just lovely . Check out the thread on dr bora by me u can find a lot of information
regards
sekhar

 

[Jay]

With your "high" expectation I only can suggest the Moskido. But it has to be built carefully. No feedback there. But as far as I know, almost all experienced builders gave up HOT amplifiers sooner or later. The Moskido 60W was my last hot amplifier. I grouped suggested amps based on wattage. For 10W there are too many class-A options, tube and SS, I don't keep any. For 30W, Aleph efficient class-A if speaker is sensitive enough, was my favorite but not keeping one, may be later. For less than 90dB speaker I'll go for minimum of 60W class-AB. For 60W I keep the Mooly amplifier (and will build many variants I guess). For 120W I keep the Stochino (for possible modification if any). Just for fun I'm sorting the best AB amp with opamp input (Quad 405 variants, LYNX 3.0, etc). If can live with high wattage class-A, I'll keep the Krell KSA 50W.

Also for fun, I'm building various amps with 5-Ampere IRL510 (not IRF) output mosfets, class-A (JLH topology) and class-AB quasies.

Before amp building become your hobby, please select a very good one even if it has to be expensive, and stay away from building too many amps. It doesn't worth it (in achieving the best sound system) except for a satisfying hobby.

 

[Wavebourn]

You can build a nice class A hybrid amp.

Here is an example:

http://www.diyaudio.com/forums/tubes-valves/147916-wavebourn-tower-iii-high-end-hybrid-amp.html

 

[5th element]

What don't I like about the PM17 ?
Sounds poor at low volume. Piano doesn't sound enough like a piano, I can't hear the notes decay, there no enough air or ambience. The soundstage is not well defined,seems to move. Not as good as the I5 or any of the valve amps I've listed above.

Funny thing is that the amplifier shouldn't be to blame for poor sound at low volume, it doesn't change. From a theoretical stand point the lower the volume the better it should sound - this of course ignoring your ears and the S/N ratio of your environment.

Sound stage definition is almost always related to the loudspeakers and their design, not the amplifier. It will be possible to finely tailor the sound, especially when considering valve amplifier matching to the loudspeakers. But a loudspeaker that's good at imaging will image well regardless of what it is driven by.

What do I like about the PM17 ?
Its well built. No hum. Looks good. Has a remote. Sounds better than the Naim kit I had. Sounds better than all the midfi stuff I've listened to before. Wife likes it.

No hum is a product of good design.

I've built 6 channels of

http://www.diyaudio.com/forums/solid-state/160061-slones-11-4-blameless.html

They are silent in operation, run very cool and sound great. With my loudspeakers they have an absolutely rock solid central image and providing the music contains it, great depth and width. This of course I attest to the design of the loudspeakers and not the amplifier, but that's my personal take on it.

My only concern would be that you appear to have a preference for valves and the ethos the blameless were created under is somewhat opposed to why people like valves - for the way they distort. Stereophile has also reviewed the Sim audio amplifier and it has a somewhat interesting distortion spectrum and I can't help but think that could have been partially responsible for the reason you liked it.

One thing with DIY though is that you often get a lot for your money. In other words, the expensive parts that go into an amplifier are the case, heat-sinks, smoothing caps, transformer and to a lesser degree the output transistors. All of these, provided the secondary voltage on the transformer is chosen with care, can be reused in a good number of different designs. So if you built something and didn't like it, you can always try something else.

 

[ostripper]

I'm looking for a decent amp to build. I use a Marantz PM17, source is a Squeezebox 3, speakers Tannoy 611. Room is 3.5 x 5m (10 x 15 feet app). My metalwork skills are basic. My electronic skills are moderate. Little HV experience. But my expectations are high ;-)

The amplifier that has impressed me most in the last few years are a WAD KT88 pp, 300B pp, Mr Liang 845 SE, EL84 pp, Simaudio Moon i5. They all had something to offer, over and above my previous amps. Nad 3030, Mission Cyrus 2, Naim NAP110, Avondale 260z, Marantz PM68 and currently a PM17. I bought and sold amps for because of my budgetary status and because, well because I could. The 260z was built as a partial kit, most of the hard work was done, like the holes in the chasis and heatsinks.

What don't I like about the PM17 ?
Sounds poor at low volume. Piano doesn't sound enough like a piano, I can't hear the notes decay, there no enough air or ambience. The soundstage is not well defined,seems to move. Not as good as the I5 or any of the valve amps I've listed above.


Want soul, body, want something that'll impart the music, not just the soundstage, not just the bass slam or the spittle flying past the microphone.

I'd like to try my hand at something that is relatively simple to build, that can stay in the living room and sounds at least as good as the I5. If I had the money and time, I wouldn't mind trying all of them, at the same time. I know each persons opinions are different, but I'd like to draw from the collective experience - the hive mind if you will.



http://www.diyaudio.com/forums/solid-state/171614-estimating-heatsink-thermal-resistance.html

I see you want to try them all . I have , but after giving away or throwing out the last one. This becomes expensive and tiresome. My NEW idea is to go MODULAR. Separate power board and a whole slew of classic AB voltage stages to mate with. I wish I thought of this before as I would have a lot of un-wasted resources to choose from now.

Completed :
AX1.1VB - Doug self "blameless" - pure , lab like amp .. simple.

BX1.1VB - Classic bootstrapped amp- (I own 2 of these - 18 months) , like DX , aksa55 ..etc.

CX1.1VB - balanced VAS- good bass , crystal clear vocals (I have 2 of these- 13 months old)

GX1.1VB - APT1 clone - built it but gave it away. Sounds like CX , but better bass (cascoded - needs 70V supplies.)

Power boards -
PB60 - 1 pair any modern output device or driver , C-R-C filtering for voltage board , built in Vbe. (35-50V unit)

PB250 - 4 pair OP devices , SUPER fat rail busses ,ESP project 15 Cap multipliers to power voltage boards , Vbe , and 6800uF X 2 onboard filtering.
(55 - 80V supplies)

ALL voltage boards can run either power board, never get bored. Don't like the amp swap out 7$ to 20$ voltage boards.
ALL are available below on my link (mongrel folder).
5th element is right ... the Blameless is the "cats meow"

example below:
OS

 

[Bigun]

Are you sure there is one amp that will satisfy you ? If you find differences between all these amps you will likely hear them between any other amps you build - each one having it's own personality probably means that you'll not find one design that will satisfy you on all music types for all time.

OS has the idea - make it easy to own more than one amp with something modular. I remember at least one person had built the F5 and F3 into the same chasis with a switch to choose between them depending on the 'mood'

 

[ervinl]

Are you sure there is one amp that will satisfy you ? If you find differences between all these amps you will likely hear them between any other amps you build - each one having it's own personality probably means that you'll not find one design that will satisfy you on all music types for all time.

OS has the idea - make it easy to own more than one amp with something modular. I remember at least one person had built the F5 and F3 into the same chasis with a switch to choose between them depending on the 'mood'

Push pull can also be swicthed to SE, just by changing other MOSFET, other than get input signal from VAS, connect it directly to VR (V+ and V-) to get fixed voltage bias

Ervin L

 

[firestorm]

Thanks for the feedback guys, much appreciated. Theres plenty to digest and mull over.

I consider diy audio to be analogous to cooking and eating. I enjoy both very much. The quality of the ingredients, the preparation, presentation and the eating are equally important. Sometimes I want just the fundamental taste to be centre stage, with little embellishment.


If I had the time and money, I'd like to try them all. A class A, a well designed class AB, a hybrid a la Wavebourn or Moskido at the minimum. A Blameless and unashamedly blameful-rose-tinted variety will be on the cards too, I think D.Self published the design of one in E&WW a few years ago. Something similar to that is being sold commercially by Cambridge Audio, the 840A. Mr. Nelson Pass's F3 will be a good candidate for the tinted version according to what I have read.

OS how would you accomodate the diffferent voltage requirements ? Multi tapped transformers ?

5th Element
I have struggled to find the right speakers for a long time, I fell in to the trap of buying kit that sounded great on demos, but rubbish at home. My last pair of speakers were Dynaudio Audience 52, totally unsuited to the room I have now, they just boomed and chuffed. At the moment I have a pair of Tannoy 611, Rogers JR149 and bits for an open baffle speakers that are on hold at the moment.

I do like valve amplifiers, but the running costs and practicalities put me off. I could consider a valve pre and ss power or perhaps a hybrid, valves doing the VA bit and mosfet/bjt/jfet doing the IA bit.

I think you guys are right, there probably isn't the one amplifier (or speaker) that'll fit the bill 100% of the time. The modular or switchable amplifier deserves more consideration. The major costs of an (diy) amplifier seems to be the power supply, the chassis and heatsinking. The driver and output stage's percentage cost doesn't seem too high.

Now, do I start off with toro with a little soy and wasabi, steamed sea bass or a nice piece of haddock in a light batter ?

 

[Brian Kingsbury]

F5. There's a reason the F5 thread has nearly 1.1 million views ...

Edit:

All of the Pass First Watt designs share a common power supply so you can change from one design to another very easily.

 

[tinitus]

looking at your Tannoy 611

if they sound best with tube amps it could indicate that you need a softie


If its possible, maybe small scale BA-1 with gain trafo instead of driver board, and a simple tube pre with good gain and drive

 

[lanchile]

NXV200 From Aussie Amplifiers

Well, let's not leave out the Aussie Amplifiers, especially the NXV200. Excellent amp kits!!!. I would strongly recommend them for sure.

 

[5th element]

With regards to the transformer.

Providing the output stage is suitably built, the rail voltages define the maximum output power.

For a single pair of output transistors the max you'd really want to use is around 40 volts. This gives you a bit under 100 watts into 8 ohms and the output pair should be able to supply around 150 watts into 4 ohms. These are resistive loads mind you, so in reality the +- 40 volt supply on a single pair is only really 8 ohm stable.

With Class A the output power is significantly reduced per output pair, now this is if you're accepting only class A usage. The amount of class A power is directly proportional to the level that the amplifier is biased to.

If you are accepting only class A usage, and no jumps into class AB, then you'd normally use around 20-25 volt rails, giving around 25-35 watts of class A into an 8 ohm load.

The other option is to power the amplifier with 40 volt rails and apply enough biasing to provide say 25 watts of class A operation. This gives the added benefit of being able to provide much larger output swings on dynamic material, the extra output is obviously class AB.

Note that class A gives the best distortion performance. Class B comes in second with class AB coming in third.

In my opinion there is an attraction towards a class A amplifier, however I see the disadvantages as unacceptable and much prefer the added efficiency of a class B output stage. If and when we can build fusion reactors I may change my mind on this.

 

[firestorm]

Hmm, so if have a 300VA, 4 x 25v transformer I could configure it to be suitable for 200W (8R) AB amp or a Class A amp pushing 50W and the rest as heat.

For AB, I assume the transition needs to be high enough not to be noticed in normal listening ? So that its only really an issue when you've got test equipment pushing out sine waves?

The extra distortion from an AB biased output stage comes from the transition between A to AB ?

A straight Class B amp would be distorting all the time?

 

[Bigun]

Now, do I start off with toro with a little soy and wasabi, steamed sea bass or a nice piece of haddock in a light batter ?

Go with the steamed bass. This method of cooking fish ultimately benefits the freshest fish, perhaps consider some Black Cod as an alternative.

Distortion in ClassAB has more than one source. There is the issue of cross-over distortion and gm doubling. Best thing is to read up some of the stuff published by D. Self on this topic.

p.s. why not simply try building an F3. Better than pottering around on the internet

 

[AndrewT]

D.Self calls his Blameless a ClassB amplifier. He shows us exactly how to set this using Vre.

When you look at what he has done you will see that everyone else in audio calls this ClassAB with bias set to minimise distortion. I and some other refer to this as optimal bias ClassAB.

Optimal bias ClassAB has the lowest distortion of all the classes except ClassA.

 

[5th element]

Hmm, so if have a 300VA, 4 x 25v transformer I could configure it to be suitable for 200W (8R) AB amp or a Class A amp pushing 50W and the rest as heat.

25 Volt secondaries will rectify to give +- about 35 volts. There's always some variation in the mains voltage, but most mains transformers are sold for 220/230 volt areas. We are supposed to be on 230 I think in the UK, but that doesn't happen, I just measured the mains in here as 242V. Usually you have to reduce the rectified voltage calculation by a couple of volts for diode losses, but as our mains runs a little hot it helps to balance it back out again.

+- 35 volt rails will give you ~70 watts into an 8 ohm load. You'd need three pairs of output transistors to manage this as Class A I think.

If you wire those secondaries now in series you'd get +- 70 volts, which is good for around 280 watts into 8 ohm. I'd want to use 5 pairs of high power output transistors for this, to give it real current driving capabilities, however the 300 VA transformer wouldn't be adequate for that!

If I were you I'd spend some time looking into various designs, see which ones tickle your fancy.

On the whole these would probably fall into two groups. One group would be class B and would function well off of +-40 volt rails (30 volt secondaries) and use either 1 or two pairs of output transistors. This would include DXs blame ST, 'my' blameless, the symasym etc.

The second group would be the class A amps that would probably function well off of around +-25-30 volts (18 volt secondaries).

For AB, I assume the transition needs to be high enough not to be noticed in normal listening ? So that its only really an issue when you've got test equipment pushing out sine waves?

This is about half right. See the job of an amplifier into a perfect resistive load is about as easy as it gets. Loudspeakers throw phase angles into the impedance equation that alter the draw of current. This can place significantly higher demands on the output stage and could theoretically be pushing your 30 watt class A amplifier constantly into class AB because your speakers demand a lot of current at say 300hz, even under normal listening levels.

What you would want is for the amplifier to provide 100% class A operation under normal listening levels and only swap into AB for large crescendos. As a lot of music nowadays is highly compressed and this can make the crescendos somewhat non existent.

Also, when testing one might use sine waves, which are not a close approximation at all to music, but they normally drive purely resistive loads, which is peanuts for the output stage to drive. Music may not be like a pure sine wave, but the transients can reach the same peak levels and it's these that you're interested in for class AB operation - do you include these within the class A side of things, or let the amp slip into class AB?

The extra distortion from an AB biased output stage comes from the transition between A to AB ?

A straight Class B amp would be distorting all the time?

Yes to the first question, not quite to the second.

in a power amplifier you've got the positive and negative sections, these two at some point must meet up and this point is called zero crossing, IE zero volts, the point where the xaxis is intersected. This creates a region where conduction has to pass from the npn to the pnp transistor in the output stage. With an underbiased amplifier this creates a bit of a mess throwing up high order harmonic content that is displeasing to our ears.

If optimally biased this region is perfectly 'smoothed' out you could say, this is a class B amplifier and it rates as number 2 on the low distortion podium.

Class A improves upon the situation, as both of the output transistors are conducting all of the time there is no 'crossover' from the npn to the pnp side of things and it takes the top spot on the distortion podium.

Class AB falls in between the two. In class B you want as perfect a hand over from the npn to the pnp side as you can get. Underbias and you have a sudden blip in the hand over, overbias and you create a similar blip, but this time it's because both transistors are now conducting, rather then just one over a small region.

In class A both transistors are conducting all the time.

In class B only one side is conducting at any one time

In class AB, both transistors are conducting some of the time and the other time only half are and it's this transition that creates the increase in distortion.

I should here point out, that there are other highly detrimental sources of distortion in constructing a high performance amplifier. Due to the fact that the current draw in a class A is constant, it handily removes the worst of these. This isn't time for you to think, hurrah I can ignore them though, because if you do, when the amplifier transitions from A to AB the performance could end up being 50-100x worse.

Realistically though this is getting ahead of the game and what you need to do is draw up a short list of amplifiers you would like to try out. Once done you can investigate how to go about getting PCBs, and the rest of the construction details, including how to best maximise the money you will need to spend.

Edit - What AndrewT said, there are a couple of different ways to define the classes and we don't want to get confused here. Simply put there's

1) Class A.
2) A class where the transition from the npn to the pnp is as good as you can make it.
3) A class that results from the over driving of a class A amp.

 

[AndrewT]

In class A both transistors are conducting all the time.

yes, but I argue that both transistors must actively control the current passing to the load, rather than simply conducting.

In class B only one side is conducting at any one time

that's where you and D.Self are misusing the definition of ClassB.
ClassB is where each of the push and pull halves operate over 180degrees of the 360degrees of the sinusoidal waveform. This is equivalent to you saying ONLY one transistor at a time sends current to the load.

In class AB, both transistors are conducting some of the time and the other time only half are and it's this transition that creates the increase in distortion............................

creates the increase in distortion compared to ClassA. ClassAB gives less distortion than ClassB. An overbias of ClassAB also gives less distortion than ClassB

............................... Due to the fact that the current draw in a class A is constant

No.
Most ClassA topologies do not pass constant current through the supply rails. Whether it's single ended or push pull, the current in the supply rails is modulated with the audio signal. For push pull the extent of that modulation is shown by considering an extra 1A going to the load, the demand is met by 0.5A from one supply rail while the other sinks 0.5A less current. When the 1A of current is being sourced from the load it's the opposite rails which source and sink the 0.5A. The output current is the difference in supply rail current. when zero current passes to the load then the supply rail currents are equal and constant (quiescent state), when the output is not zero, the currents in the supply rails change from the quiescent state value to a new value that is alternately higher then lower than the quiescent value.

If you look at balanced amplifiers and ensure that the increase in one rail exactly matches the opposite phase decrease in the other rail and do the same for the other polarity supply then the matching but out of phase currents in the two supply rails cancel and the common PSU sees a constant load, but only if there is a common PSU feeding all four supply rails of the balanced amplifier.

There is one topology of ClassAB where there is constant current but it is very rarely implemented. A single ended ClassA stage with the load returned to the supply rail rather than the ground rail.

 

[5th element]

yes, but I argue that both transistors must actively control the current passing to the load, rather than simply conducting.

Indeed it was a poor choice of words and a lack of thought on my part originally. I know this, but it wasn't until you mentioned it that it became apparent. So thank you

that's where you and D.Self are misusing the definition of ClassB.
ClassB is where each of the push and pull halves operate over 180degrees of the 360degrees of the sinusoidal waveform. This is equivalent to you saying ONLY one transistor at a time sends current to the load.

Lets take the definition of a class AB amplifier to represent the blameless I created and let's say its an optimally biased class AB amplifier.

If it is driving a purely resistive load, I assume we could call it a class B amplifier?

However when driving any real life load, current would need to be sourced from the opposite side too due to the reactive loading and hence can't be called a class B amp.

creates the increase in distortion compared to ClassA. ClassAB gives less distortion than ClassB. An overbias of ClassAB also gives less distortion than ClassB

Isn't this simply debating the usage of the terms Class A, class B and class AB?

I believe I loosely re-evaluated the definitions at the end of my last post. Maybe you missed it as I edited it in.

No.
Most ClassA topologies do not pass constant current through the supply rails. Whether it's single ended or push pull, the current in the supply rails is modulated with the audio signal. For push pull the extent of that modulation is shown by considering an extra 1A going to the load, the demand is met by 0.5A from one supply rail while the other sinks 0.5A less current. When the 1A of current is being sourced from the load it's the opposite rails which source and sink the 0.5A. The output current is the difference in supply rail current. when zero current passes to the load then the supply rail currents are equal and constant (quiescent state), when the output is not zero, the currents in the supply rails change from the quiescent state value to a new value that is alternately higher then lower than the quiescent value.

This is interesting. Lets say we're talking about a sine wave for example.

Also lets say we've got a class A amplifier biased so that under quiescent conditions 1 amp is flowing.

Finally lets say that the amplifier is driving a load such that the amplifier is working to it's absolute limits with regards to class A operation, but it never goes beyond it. The load is purely resistive.

For example at zero crossing no current is flowing to the load, so 0.5 amps is sourced and sunk by each power rail respectively. Or rather should that be 1 amp?

However as we approach the peak of the sine wave, more and more current is sourced by the positive power rail and the increase in demanded current returns through the return ground. Whilst at the same time less and less current is sunk by the negative supply rail and this reduction is proportional to the increase in the return current flowing through the ground?

Does this explain why a 100% class A amplifier is it's the most efficient when being driven to it's maximum level?

 

[5th element]

I just noticed that there's a poll at the top of this thread and instead of answering that I'll post here.

You should build an amplifier that under no normal conditions will be forced to clip. This will by definition set the maximum output power that will be required and from there you can make an informed decision about what amplifiers may be suitable.

As an extra question.

Lets say you've built an amplifier that is going to be driven on +- 40 volt rails. You've biased it so that it will provide 20 watts of class A power into an 8 ohm load.

Naturally it will step into being an over-biased class AB amplifier when those 20 watts are exceeded. Or if never driven that hard, it will always remain under class A operation.

Why would anyone want to build a class A amplifier that works on say +- 18 volt rails that will provide those same 20 watts of class A, but will then clip instead of transitioning into an overbiased class AB?