ExtremA class A amp

[Piotr]

Hi!

Have anyone read about (or even better, built) the new class A amp designed by Bruno Putzeys?

Bruno is the guy behind the UcD amps from Hypex.

The ExtremA is a DIY amp and you can read more at;

http://www.hardwareanalysis.com/content/article/1842.2/

I'm about to build two channels.

There's also a thread at diyaudio about the amp;

http://www.diyaudio.com/forums/showthread.php?threadid=96853

/Peter

[Victor]

ExtremeA is one great amp.

I read the article when it first appeared and I was impressed. It is not a 'me too' amp at all, - it stands out from the crowd in all respects. It kind off reminds me of Ampzilla design, - which is my other favorite amp.

Anyway, the fully symmetric, complementary topology and the Class A to boot with no ground reference is perhaps the best way to go as far as power amps are concerned. Nothing that the likes of Bryston, Levinson, ADI and many others even come close. The specs alone are in the category all by themselves here,- virtually no distortion!

I am also considering building the thing. Tell me, - how do you intend to house the amp? What box and the heat sinks are you using?

[AptosJeff]

No comment?

I'm surprised there hasn't been more response here. I'd like to know what some of you other amp guys think. JonMarsh? Karma?

[Piotr]

I'm also surprised about the "lack" of response about the amp. Maybe things will change when people start to listen to and measure the little beast. The distortion seems to be record low.

I did plan an amp similar to the ExtremA but I don't have enough knowledge to pull a project like that off from scratch.. Obviously I was very glad to see the thread over at diyaudio.com.

I'm not 100% sure how I will build the amp. The PCB's arrived the other day and I'm writing a list of components to order. Probably it will be monoblocks and I'll likely scale it for 100W (even though it's tempting to make it an 50W amp to save some on the electric bill and heat in the room). I have a couple of heat sinks lying around but I'm not sure how to chose the proper one.. I have two sizes and my gut feeling (comparing to my other class A amps, the Zapsolutes from LC audio) tells me to go with the biggest I got.

The chassis will likely be built as a tower to save space on the floor. No standard "black-box-from-Japan" here. I want to make it as compact as possible without risking hum and noise from the toroids.


/Peter

[TacoD]

It was published last year in the magazine Elektuur (Dutch Elektor). The design is from the same guy (Bruno) which designed the Hypex UCD class D amplifiers. The pcb's and output devices are expensive that's the reason I did not build this amp and stick to my Pass clones.

[Victor]

Piotr,

I am going to order the PCBs soon. Right now I am thinking about the chassis and the heat sinks. If you go with 100 W class-A approach then the heat sinks will have to be truly huge and the weight of the amp will be about 30-40pounds, assuming it is a mono-block with a 500 VA transformer.

Picture a finned heat sink at least 6 inches tall and 12-14 inches long with 1/3 inch thick fins extending to about 2.5-3 inches. I would like to buy a pre-made chassis suitable for a mono-block and simply slap the heat sink on it. Question is, - where to get the box? Any ideas?

Victor

[Hdale85]

There is a place www.diyenclosures.com they have a 1013-HS but they have been out of stock for a while. He's supposed to be getting more eventually but he's a private dealer kind of guy with limited funds as he does this in his spare time sort of thing. They are very nice though.

[ThomasW]

http://www.par-metal.com/plist.htm

[Victor]

Thanks Thomas and Dougie for the link to parmetal and diyenclosures. I do like the HS-1013 deal. Let’s see if this 1013 model will become available.

I re-read the article on ExtremeA and noticed that there is no mention of current in-rash protection. Those large filter capacitors of 33,000 uF each will definitely require some form of slow charging on the start up. Unless I missed the mention of the start-up circuit somewhere, we will need to provide this capability for sure.

[Piotr]

Yes I figure some kind of softstart is necessary.

About filter caps, any idea in using four pole caps with a constant current design like ExtramA?

Any point in using external powersupply? Guess not since the noise levels seems to be extremly low as it is. Guess it will be low enough as long as the toroids are mounted for minimum "radiation" aimed towards the amp board.

Rectifier, any 35A will do or are there significant differences in these?

I've been thinking on the scaling of the amp. Do you think it would be possible to use a BIG heat sink and scale the amp for something like 200W but at the same time lower the quiscent current to 50W or so? Iow. 50W class A and 200W A/B.


I will likely build the amps without much concern about the finish and looks of it. Will probably hide it behind a curtain of some sort which I will also use for hiding absorbers on the back wall of the listening room.


Oh.. about heatsinks.. I was thinking of building the chassis like any standard stereo amp with heatsinks on each side, left and right (still separate mono's though). The idea is to us a relatively narrow dimension so the amp circuit board can bridge from the left to the right heat sink with one pair of transistors on each heat sink. Then I realized that that could casue some problem with unequall temperature on the two heat sinks due to outer circumstances such as open window/door or "closeness" to other objects or a near wall. This would possibly make the DC shift on the output I guess?


Many thoughts now, will start building next week!

/Peter

[Hdale85]

Just mount one channel to each side of the chassis. No reason in trying to put one set of transistors from one on one side and the other set on the other side.

[Piotr]

Just mount one channel to each side of the chassis. No reason in trying to put one set of transistors from one on one side and the other set on the other side.

Thanks but... no! I will build them as monoblocks. The reason I was thinking on that solution is that the biggest heatsinks I have may be a tad to small hence my wish to use two heatsinks for one channel. It would also be nice constructionwise to build the chassi as a "normal" amp with one heat sink on each side, just bolting some sheets of aluminum to them. Two good reasons iow. My concern as mentioned is different temperatures on the two heatsinks that possibly could cause drift of the outputs.

/Peter

[JonMarsh]

Very interesting looking design, even if I'm more a fan of non feedback designs. Would be really interesting to see testing at the level that say, John Atkins of Stereophile does. A real world issue is stability with reactive loads (particularly capacitive), and amplitude sensitivity to load impedance. If it uses as much feedback as I suspect, that's probably pretty low. Scaling this to a bit more power would be an interesting project, using the extended beta outputs from On Semi or Toshiba.

For a true class A amp you do need to take some care about things like choice of rectifiers and the hetainking arrangements- unlike conventional amps, they're always running at pretty high load levels. If you do your own power supply, I'd look closely at capacitor ripple current and temperature ratings, as well as ESL. If I didn't already have a very nice sounding "small" amp in my Ayre AX-7, (and if I had more spare time), I'd be tempted to build some of these. For me, a more interesting approach then Pass's low power DIY designs.

The proof is in the listening, though, not the sine wave tests.

~Jon

[Piotr]

Interesting indeed Jon!

I also use to be a "fan" of non feedback designs (or rather non loop feedback/non global loop feedback) but I am a bit less dogmatic these days and believe that done right and used at the right place feedback can work fine.. really fine. Acutally my current setup is totally "no feedback". The SACD player uses the Zapfilter from LC audio and the amps are the class A Zapsolutes.

I believe that if a design is relatively linear befor feedback is applied, fast and running in class A (preferably bridged), feedback does no harm, rather the other way around.

I will try to do some extensiv testing of the ExtremA's and will also put it thru a "bypass listening test" as used by LTS, the Swedish "sound technology society". Loading the amp with a complex impedance load and listen to a signal before and after it has passed the amp. Few amps or other components can pass music signals without altering it and coloring it enough to hear it. Of hundreds of amps that has been put thru these tests I think only a Bryston amp did pass without detection. That means no one in the panel could here if the signal was taken before or after the amp.

My gut feeling is that the ExtremA is close to perfect and will be non detectable in a "before/after" type listening test.

/Peter

[JonMarsh]

That's a good rigorous approach to take. I'm a little surprised that a Bryston passed that test, but I've heard since they redesigned their grounding system in the newer ones a few years back, and a few other tweaks, they are much better.

~Jon

[Kevin Haskins]

Bruno certainly knows his stuff. I'm working on a PCB for something more at my skill level (design wise) with the new National LME49810. Might be interesting to build one of the Class A amps just to have as a reference.

[capslock]

The design is interesting. However, I wouldn't want to have a class A nagging my ecological conscience. Also, at 0.21 €/kWh, one had better design in an auto-power-off.

Modifying this to semi class A (i.e. about 0.5 A bias) will mean a redesign of the bias circuit unless one wants to live with about 1R of emitter resistance. I am surprised the bias circuit will not ring with as little roll-off as the schematic indicates.

I'm pretty sure replacing the BD139/140 drivers with something more linear will do good, even in a class A design. The Sankens are excellent.

Scaling this to a bit more power would be an interesting project, using the extended beta outputs from On Semi or Toshiba.

Are you implying those are superior? They may be slightly cheaper, but to me the Sankes are preferable because of their higher guaranteed f_t and lower thermal resistance. h_fe linearity and capacitances are in the same ballpark for all devices.

[Piotr]

Good points there Capslock!

I have a powerdown function on my current class A amps for non critical listening and I agree that it would be a nice option on the ExtremA. The price for power is lower here up north and I also heat my small house with electric power. In other words I will not be as much an ecological bandit as I could have been. :B Only during the summer I would waste energy.. guess I'll have to use a class D amp during summer months if I want to be able to sleep at night then.

Actually I move in the right direction anyway.. becasue my current amps idles at aprox. 750W so I will actually reduce my "audioheating" by a third since the 100W ExtremA idles at 500W. Scaling the amp for 50W would be even better and could be enough with "only" 250W of heat produced. Eventually I'll use the ExtremA for the highest registers only.

/Peter

[JonMarsh]

The design is interesting. However, I wouldn't want to have a class A nagging my ecological conscience. Also, at 0.21 €/kWh, one had better design in an auto-power-off.

Modifying this to semi class A (i.e. about 0.5 A bias) will mean a redesign of the bias circuit unless one wants to live with about 1R of emitter resistance. I am surprised the bias circuit will not ring with as little roll-off as the schematic indicates.

I'm pretty sure replacing the BD139/140 drivers with something more linear will do good, even in a class A design. The Sankens are excellent.



Are you implying those are superior? They may be slightly cheaper, but to me the Sankes are preferable because of their higher guaranteed f_t and lower thermal resistance. h_fe linearity and capacitances are in the same ballpark for all devices.

I like the front end design of the Extreme amp, but not the bias system or output stage. I've been simulating some circuits with the On Semi Thermal Trak parts, NOT using any of their recommended configurations, and that, plus the power dissipation capability and VCE are my issues.

By extending the power a bit, I'm talking 250-300 watts @ 8 ohms. Something like the MX-R, which also uses the Thermal Traks.

Having a two level bias capability would be interesting, perhaps. For the ECO folks and for the Class A- the thing is, partial heavy AB with conventional circuits and devices usually results in higher output stage nonlinearity then a very well designed low level AB bias stage.

[Johnloudb]

I've been playing around with a variable bias class A amp circuit of my own design. The circuit monitors the output current and bumps up the bias current when needed. The bias comes down slowly, so it's not audible. I haven't yet done much Spice modeling yet, been too busy with other projects.

I favor simple circuits with linear open loop, and small amounts of feedback.

Having a two level bias capability would be interesting, perhaps. For the ECO folks and for the Class A- the thing is, partial heavy AB with conventional circuits and devices usually results in higher output stage nonlinearity then a very well designed low level AB bias stage.

So you're saying that a class AB output stage, biased heavily into class A, is less linear than one with less bias current? Didn't know that. That shoots down my plan B. :B

Is that just with BJTs or MOSFETs also?

[capslock]

I like the front end design of the Extreme amp, but not the bias system or output stage. I've been simulating some circuits with the On Semi Thermal Trak parts, NOT using any of their recommended configurations, and that, plus the power dissipation capability and VCE are my issues.

By extending the power a bit, I'm talking 250-300 watts @ 8 ohms. Something like the MX-R, which also uses the Thermal Traks.

Having a two level bias capability would be interesting, perhaps. For the ECO folks and for the Class A- the thing is, partial heavy AB with conventional circuits and devices usually results in higher output stage nonlinearity then a very well designed low level AB bias stage.

Yeah, the OnSemi app note appears to have been authored by monkeys, doesn't it?

I have a bunch of ThermalTraks sitting here, just haven't had the time to play with them yet. As for the Toshibas or regular OnSemis, I see no real advantage for them over the Sankes which happen to be just a tad faster.


What is the MX-R? I presume its an amp, not a sports car? Do you have a link?


Last I checked, Leach had concluded the optimal bias (Self-speak) or gm-doubling (non-Self lingo) thing was a myth. Anything wrong with his paper?


Greetings

Eric

[JonMarsh]

I've been playing around with a variable bias class A amp circuit of my own design. The circuit monitors the output current and bumps up the bias current when needed. The bias comes down slowly, so it's not audible. I haven't yet done much Spice modeling yet, been too busy with other projects.

I favor simple circuits with linear open loop, and small amounts of feedback.



So you're saying that a class AB output stage, biased heavily into class A, is less linear than one with less bias current? Didn't know that. That shoots down my plan B. :B

Is that just with BJTs or MOSFETs also?

As Caplocks metions, for a standard class A-A/B, as you raise the bias from near zero, yous see an optimal bias point initially with low crossover notch, then as you increase it somewhat, the gm shifts upwards and things are not quite as perfectly linearly, unless you're biased heavily enough to be in true class A. This is most easily measured in an open loop circuit, BUT, it is ALSO dependent on the type of output transistors, even different bipolars having varying sensitivity.

A lot of folks who don't agree with Self's overall philosophy may throw the baby out with the bathwater- he has made a lot of valid observations and measurements- just, in some cases, he's dogmatic about things that are somewhat easily circumvented.

[JonMarsh]

Yeah, the OnSemi app note appears to have been authored by monkeys, doesn't it?

I have a bunch of ThermalTraks sitting here, just haven't had the time to play with them yet. As for the Toshibas or regular OnSemis, I see no real advantage for them over the Sankes which happen to be just a tad faster.


What is the MX-R? I presume its an amp, not a sports car? Do you have a link?


Last I checked, Leach had concluded the optimal bias (Self-speak) or gm-doubling (non-Self lingo) thing was a myth. Anything wrong with his paper?


Greetings

Eric

Hello Eric,

By monkeys genetically engineered for subnormal IQ and creativity.

However, I know my friend Charles Hansen did spend some time wrestling wtih this aspect; though the Ayre MX-R sounded very good when I first heard them in January of 2006, he was a few months late shipping because of further development work.

I'll post some links later, and more comments- must get back to a session at the IBM Power and Cooling Symposium.

~Jon

[ThomasW]

Ayre MR-X mono-block machined from a single piece of billet aluminum

[JonMarsh]

there should be a good online review at Stereophile now, too. The measurements are solid, but the only real test is listening. I'm not objective, though, so don't pay too much attention to my comments. :W

------------------------------------
from my phone at the IBM Power and Cooling Symposium

[Piotr]

Wow that Ayre chassi is some serious stuff. I can promise mine will not look anything like that.. :B


Trying to find all the components now for ExtremA which is a litle harder than I thought it would be but piece by piece I'm making slow progress on collecting the parts.


/Peter

[capslock]

Hello Eric,
IBM
~Jon

Invented By Monkeys? I was not happy at all with my T20 laptop years ago, but the T63 I have right now is fine.

[JoshK]

Just give into the dark side and accept the truth of tubes....

[JonMarsh]

Btdt

[Piotr]

Just give into the dark side and accept the truth of tubes....

Nah.. I don't want a good sounding amp.. I want a non-sounding amp!


/Peter

[Jonasz]

Nah.. I don't want a good sounding amp.. I want a non-sounding amp!

http://www.ska-audio.com/diy/

Ordered four channels for the tweeters and mids... :B

[JoshK]

Sorry, not trying to derail the thread, was just making a joke. I like tube amps, I also like SS amps...I have no axe to grind.

[Piotr]

Sorry, not trying to derail the thread, was just making a joke. I like tube amps, I also like SS amps...I have no axe to grind.

Don' be sorry, I took it as a joke and tried to be a little funny as well.


But to be serious for a little while.. my goal is simply to build an amp that is neutral and transparent to the incoming signal. I do not want an amp that sound subjectively good, but simply deliver the signal as "a wire with gain" so to speak. Perhaps there are cheaper solutions for that goal but the beaty of this design and the extreme performance was to tempting for me to resist.. just had to get a pair of those..

In the end I hope I will have a true reference amp that other can be tested against. I have for example listened to a LM3886 chipamp that was impossible to detect in a "before/after" listening test. The value of such a test depends a lot on the material you listen at, at what levels and of course the surrounding gear you use.


/Peter

[Piotr]

http://www.ska-audio.com/diy/

Ordered four channels for the tweeters and mids... :B

Seems like a nice solution Jonasz!

Maybe a get-together in the future if you'r back in the land of the wikings? Would be fun to compare and perhaps listen in a "LTS aproved before/after test".


/Peter

[Victor]

Well, - it seems I found a good box for my upcoming stereo version of ExtremeA. I have 2 old Bryston 4B NRBs that I bought used in the mid 90’s. I do not need the power of 4B and I do not like the output cascode approach that Bryston undertook with the 4B design.

Those who are familiar with Bryston products would know that all Bryston amps are built on the essentially same circuit topology; - they are just scaled for different power. So, - for a while now I was thinking of turning those old 4Bs NRBs into a modern 4-channel 3BSST with a twist.

My plan is to take the 4 circuit boards from the pair of 4Bs and put them into one box and use the other box for the stereo ExtremeA.

In the process I intend to remove the cascodes, effectively turning the 4B into 3B and considerably upgrade the overall design to be actually superior to what Bryston is building now. Bryston’s box is great; -it is truly modular and is very suitable for Class A design. All I need is a pair of more substantial heat sinks which are easy to get.

These days an old, but not too old, Bryston 4B NRB can be had for about $300-400. Much of its internal guts can be used for the ExtremeA. You get among many other things:
1. an excellent on/off switch,
2. the thermal breaker
3. the back plate with all necessary connectors including the power connector with a fuse box and the XLR connectors
4. The pair of power transformers. Bryston uses excellent toroidal 500VA transformers for the 4B. The voltage rating on those is about 60 V RMS per winding. It is a fairly simple matter to expose the transformer wire and unwind as many turns as it takes to match the requirement of the ExtremeA.

All in all,- it is a $300-400 well spent particularly considering what you get and, more importantly, absolutely no need for laborious metal work. In fact, methinks, that one will spend more on those parts if purchased alone. If anyone is interested in the 4B NRB-3B SST a with a twist metamorphosis, I will work out a procedure and will make it available.

Comments on this plan?

Victor

[JonMarsh]

Well, - it seems I found a good box for my upcoming stereo version of ExtremeA. I have 2 old Bryston 4B NRBs that I bought used in the mid 90’s. I do not need the power of 4B and I do not like the output cascode approach that Bryston undertook with the 4B design.

Those who are familiar with Bryston products would know that all Bryston amps are built on the essentially same circuit topology; - they are just scaled for different power. So, - for a while now I was thinking of turning those old 4Bs NRBs into a modern 4-channel 3BSST with a twist.

My plan is to take the 4 circuit boards from the pair of 4Bs and put them into one box and use the other box for the stereo ExtremeA.

In the process I intend to remove the cascodes, effectively turning the 4B into 3B and considerably upgrade the overall design to be actually superior to what Bryston is building now. Bryston’s box is great; -it is truly modular and is very suitable for Class A design. All I need is a pair of more substantial heat sinks which are easy to get.

These days an old, but not too old, Bryston 4B NRB can be had for about $300-400. Much of its internal guts can be used for the ExtremeA. You get among many other things:
1. an excellent on/off switch,
2. the thermal breaker
3. the back plate with all necessary connectors including the power connector with a fuse box and the XLR connectors
4. The pair of power transformers. Bryston uses excellent toroidal 500VA transformers for the 4B. The voltage rating on those is about 60 V RMS per winding. It is a fairly simple matter to expose the transformer wire and unwind as many turns as it takes to match the requirement of the ExtremeA.

All in all,- it is a $300-400 well spent particularly considering what you get and, more importantly, absolutely no need for laborious metal work. In fact, methinks, that one will spend more on those parts if purchased alone. If anyone is interested in the 4B NRB-3B SST a with a twist metamorphosis, I will work out a procedure and will make it available.

Comments on this plan?

Victor

Sounds good to me, Victor- it's the same reason I buy older Aragons- they're solid amps to begin with, and a good platform for building something else with new electronics- hence, the AragonX project.


I just haven't had time to work in the last two years to finish my new channel electronics. But that cloud has had a silver lining, as the Bias Track transistors have come out from On Semi, and with that and my lastest simulations, I think I'm getting near a final design.

[Drew]

In the process I intend to remove the cascodes...

Victor

I know next to nothing about analog circuit design, so I have a question: why remove the cascode? My understanding of a cascode is that it is an elegant and effective way of reducing distortion without having to use feedback. What disadvantages do they have?

Thanks.

[Victor]

Sounds good to me, Victor- it's the same reason I buy older Aragons- they're solid amps to begin with, and a good platform for building something else with new electronics- hence, the AragonX project.

I just haven't had time to work in the last two years to finish my new channel electronics. But that cloud has had a silver lining, as the Bias Track transistors have come out from On Semi, and with that and my lastest simulations, I think I'm getting near a final design.

And I thought I was the first one to think of this!

Indeed Jon, this is the easiest way to go, particularly for someone like me who is hands-on challenged. If I can stay away from the metal work, - I am happy. I confess I have never seen an Aragon amp, - perhaps I should look at it.

I know next to nothing about analog circuit design, so I have a question: why remove the cascode? My understanding of a cascode is that it is an elegant and effective way of reducing distortion without having to use feedback. What disadvantages do they have?

Drew,

To answer your question I need to look at the Bryston circuit in some detail. I hope that the following is not too involved. Here it goes…

The cascade is generally a good thing. It does improve the linearity of the overall circuit, by fixing the collector voltage of the gain-generating transistor at the constant DC potential and thereby taking the non-linear base-collector capacitance out of the equation. This is not quite what Bryston is doing and I am guilty of calling Bryston’s output stage cascoded.

Bryston 4B amp has a very clever output stage and at a first glance you almost want to say that there is no way it can be stable, but it is. Bryston’s output stage has voltage gain, which a text book no-no and it comes due to the active loading of one out of the two output transistors used.

The load transistor in an active load pair is not a cascode transistor in a classic sense, although it looks like one, - it is really an active load. Considering the positive voltage rail for the moment, Bryston uses both PNP and NPN power transistors at the output. The PNP transistor will generate gain, while the NPN will not, but the NPN will see a higher voltage at its base, due to the gain of the pre-driver that drives both of those power transistors. So the voltages appearing at the collector of the PNP and the emitter of the NPN are roughly the same. The overall feedback will make them to be nearly identical.

Each of those power NPN and PNP transistors is loaded with a PNP power transistor. This configuration creates a nice current buffer and will have voltage gain somewhat free of bandwidth-reducing Miller effect.

This is all very nice in theory but as far as I know Bryston is the only manufacturer that managed this thing to work in the power output stage. There are potential problems which Bryston design does not escape. One thing for sure, for the overall scheme to work the NPN-PNP pair must be fairly evenly matched in terms of their Vbe-Ic characteristics. This is quite impossible to achieve in practice, resulting in uneven operation. The 4b output stage has two current generating sections. However due to the transistor mismatch one section will almost definitely work harder then the other.

The good thing is this; - the feedback around the output stage tries to keep the output transistors in the linear range. This alone is responsible for the exemplary THD numbers that Bryston publishes.

Now, what are the reasons for going into this complexity? The way I see it, - it has to do with manufacturing. This scheme allows Bryston to use the same transistors and nearly identical circuit topology in all of their amplifiers regardless of the power output. The 4B specifically simply divides up the voltage of the output stage so each transistor thermally works about the same with about 40 collector-emitter Volts per transistor. Notice that the power rails are set at plus-minus 85Volts! Alternatively the higher output power amplifier might need heftier transistor. But NOT in Bryston’s case! Here all the amps from the 2B all the way to 4B use the same devices.

I do not like the complexity where none is necessary. I also do not like the system that cannot be made linear without heavy application of feedback. Call me old-fashion, but that is what I think. Bryston amps linearity completely depends on the feedback. The ST and subsequent SST versions take this even further but increasing the feedback and hence producing even better THD numbers. I am not criticizing Bryston, but I would make different engineering choices. Personally I like the approach taken by Dr. Leach, - see here http://users.ece.gatech.edu/~mleach/lowtim/

So in the end, the active loading, or cascodes in my original post, do not contribute to the overall linearity, - they simply the vehicle to create a higher power amplifier. If you were to simulate the circuit as I have done, you will see that the 3B will actually produce better THD then the 4B, but the difference is slight due to the action of the feedback around the output stage. The Open-Loop gain should be a little higher in the 4B case, so the feedback should in theory work better, but the added complexity and the resistor choices negate this greatly. So the 3B topology wins here on linearity alone. In fact I expect the 2B to be the king of all Brystons amps, because it is the simplest design they manufacture with only one pair of output transistors.

I intend to turn my two 4Bs into a quad 3B in one box. In the process I will provide much better input stage voltage regulation then the SST line, I will use the same output transistors as the SST line uses but in a different package and I will replace the input transistors with a low-noise matched pairs. However more importantly, and this is critical in opinion, I will completely change the overall feedback network. I will do away with an electrolytic capacitor completely.

This capacitor although it sees little voltage still is a source of significant non-linearity. There is a way to use a plastic (polypropylene) capacitor in its place but not in the same way. The overall transfer function will not change. My simulations show that a decrease in THD by a factor of more then 5 certainly possible as a result. I further will increase the amount of raw capacitance available to the amplifier from 10 mF to perhaps 20 mF per rail. I will also have to get new transformers.

The end result should be an amp that is better then 3B SST. I have yet to do all this, so I reserve judgment until the deed is done.

regards,
Victor

[Hdale85]

Be nice if I could find a dead amp on ebay or something to use for my chassis for my amp. I can't find much though. I'm just doubting my chassis building skills.

[Drew]

Wow, Victor, thanks for the detailed reply to my question. It obviously took a lot of time and I appreciate it.

On first read, I understood maybe 70%, so I'll have to read it a few more times. Analog circuit design has always fascinated me. A pity that I don't have the time to really dive into it.

It's funny - the engineer in me wants to believe that making a linear amplifier at audio frequency ranges should be an easy thing. Or, at the very least, we should have converged on the ideal solution by now. All these different topologies and competition quite frankly surprise me. There's always more to learn...

[Piotr]

Most of the components will arrive soon and I'm about to buy some new solder for the ExtremA board.

What do you think about this solder from Kester:

http://www.kester.com/en-us/products/prodcat_detail.aspx?pid=44

It's called '285' and contains 2% rosin flux. Sn/Pb/Ag. Anyone used this particular solder?

edit: should add that the ExtremA board and the supply board is of goldplated "thru holes" type.

/Peter

[JonMarsh]

My personal opinion is that this would be a good solder for things like speaker crossovers, point to point wiring in a vacuum tube amp, chasis wiring in general.

My personal favorite in this class of Rosin based solders is Cardas Quad Eutectic.

For a PCB work, a different approach may be more effective. I have had very good results with organic water soluble cored flux wire, but the versions using lead are hard to find or not available, and I have no experience with the lead free versions. A Lead free example is the 331 solder. Hot water is used to rinse or clean the board; lead and board wetting is very good, and after cleaning there is no residue that can exhibit hydroscopic action (rosin can).

These days the "no clean" types are more popular, due to fewer production steps, but that means there can be some residual flux on the board. A good low residue lead free solder is the 275.

flux Rosin residue can be removed by stronger solvents, such as alchohol, but there is more risk of component damage. I still have a number of older spools (NOS) of the water clean stuff, and that's still all I use for hand assembly of PCBs.

[Piotr]

So John,

are you suggesting a more mildly activated rosin?

Maybe even less rosin.. say 1% instead of 2%?

The info on the '285' says the flux residues are non-corrosive non-conducting so what could be the problem in your opinion?

In an earlier project I used a silver solder that was easy to use and gave nice shiny "solders". Don't remember the kind of rosin but I did not clean the board or the "solders" and everything looks fine close to ten years later.

I have an old spol of multicore silver solder with1% roin flux in two canals. This is non easy touse, it doesn't wet like the one I used earlier and therefore I thought that Kester '285' like a good easy to use solder. Aaah choices choices and more choices!

/Peter

[Johnloudb]

So John,

are you suggesting a more mildly activated rosin?

Maybe even less rosin.. say 1% instead of 2%?

The info on the '285' says the flux residues are non-corrosive non-conducting so what could be the problem in your opinion?

In an earlier project I used a silver solder that was easy to use and gave nice shiny "solders". Don't remember the kind of rosin but I did not clean the board or the "solders" and everything looks fine close to ten years later.

I have an old spol of multicore silver solder with1% roin flux in two canals. This is non easy touse, it doesn't wet like the one I used earlier and therefore I thought that Kester '285' like a good easy to use solder. Aaah choices choices and more choices!

/Peter

I used Wonder Solder, rosin core solder 12 years ago, never washed it off and joints still look good. It is supposed to be corrosive, but just haven't seen that yet.

John

[JonMarsh]

So John,

are you suggesting a more mildly activated rosin?

Maybe even less rosin.. say 1% instead of 2%?

The info on the '285' says the flux residues are non-corrosive non-conducting so what could be the problem in your opinion?

In an earlier project I used a silver solder that was easy to use and gave nice shiny "solders". Don't remember the kind of rosin but I did not clean the board or the "solders" and everything looks fine close to ten years later.

I have an old spol of multicore silver solder with1% roin flux in two canals. This is non easy touse, it doesn't wet like the one I used earlier and therefore I thought that Kester '285' like a good easy to use solder. Aaah choices choices and more choices!

/Peter

Organic mild acid fluxes were developed in the 70's to provide a water washable flux that had good cleaning and wetting properties even with somewhat problematic surfaces (oxidized metal, including tin, copper) that could be cleaned easily without solvents. The main caveat is that the board MUST be cleaned within 24 hours of soldering; immediately within the production flow is preferred. Do that, and you get great solder joints and a very clean professional looking board that won't absorb any moisture in the flux residue- because it's not present. Now, moisture absorption is not a big factor in most circuits, except high impedance FET and vacuum tube gear. Still, I'm not a fan of flux residue on boards. OK, I'm a bit OCD, I suppose! :W

In production, mild organic acid flux for electronics been long superceded by low residue fluxes for wave soldering. But these fluxes are not as effective in dealing with any oxidation on leads or trace surfaces, and are designed primarily for wave soldering, not so much for hand soldering. This is a factor in why component plating solderability is a much bigger deal in the electronics world than 20 years ago. Assembly processes these days are very tightly controlled, and prevention of any metal oxidation in PCB's and components is an important quality factor.

Rosin fluxes all require something stronger than water to clean, and even just rubbing alcohol can be a problematic solvent for some plastics used in electronic components.

In relatively low impedance circuitry like the Extreme A, may not really make much difference- for me, it's just a workmanship thing.

If you go with a "conventional" rosin core solder, I can't recommend Cardas Quad Eutectic highly enough- I've used a number of other silver bearing solders in the past, and the Quad Eutectic is formulated similar to what we use for chip die attach- the combination of lead, tin, copper and silver melts and hardens at just one temperature- it's practically impossible to get a cold solder joint, it flows great, and the flux cleans well but has relatively low residue for a rosin flux. I've been using it exclusively for speaker crossovers and point to point wiring the last five years.

~Jon