AKSA too Expensive
AKSA may be great sounding... and I guess that is the ultimate worth of a component but for what you get parts-wise it just seems too expensive. Also Mr. Dean monitors the entire web like a hawk looking for any references to the AKSA name and makes sure that the circuit is not published anywhere and that people are guilted into not even thinking about publishing details.
Greg
AKSA may be great sounding... and I guess that is the ultimate worth of a component but for what you get parts-wise it just seems too expensive. Also Mr. Dean monitors the entire web like a hawk looking for any references to the AKSA name and makes sure that the circuit is not published anywhere and that people are guilted into not even thinking about publishing details.
Greg
Hi!
Although oviously my native tongue is not english (and sadly so), I bought the PCBs for Death of Zen because I thought that this amp would be the ultimate destruction 
of all other zen-style-class-A-amps (at least you could get the impression while reading the introductory article on ESP 
)...
Maybe someone can help me with my probs, but I will post it in another thread...
Bye,
/Arndt - who's still sometimes using /Halo style endings, and who right now is very much enjoying the arrival of his TI samples... although one part is still missing, will be delivered when available...
Peter Daniel said:
Isn't the name itself suggest that this is a dead amp?
Although oviously my native tongue is not english (and sadly
so), I bought the PCBs for Death of Zen because I thought that this amp would be the ultimate destruction of all other zen-style-class-A-amps (at least you could get the impression while reading the introductory article on ESP )...Maybe someone can help me with my probs, but I will post it in another thread...
Bye,
/Arndt - who's still sometimes using /Halo style endings, and who right now is very much enjoying the arrival of his TI samples... although one part is still missing, will be delivered when available...
Re: AKSA too Expensive
Still he provides a lot of free knowhow to us all,
in his postings here at Forum.
We are all in different situations and have different concerns,
that will/have to guide our thinkings/doings.
-----------------------------------------------------------------
We can find all the greater joy in that people
like Jonathan Carr , jcarr at diyAudio
and Nelson Pass , Nelson Pass at diyAudio
are in situations, where they can/want to,
so willingly share their "secrets" in amplifier designs.
More people, that are generous givers in one or another way
at this forum, I am sure you can add yourself.
-----------------------------------------------------------------
/halo - knows what our actions will bring us, in the end:
- if you give - others will give to you
- if you withhold - others will withhold from you
Hugh Dean, AKSA, may have his reason.GringoAudio said:AKSA may be great sounding... and I guess that is the ultimate worth of a component but for what you get parts-wise it just seems too expensive. Also Mr. Dean monitors the entire web like a hawk looking for any references to the AKSA name and makes sure that the circuit is not published anywhere and that people are guilted into not even thinking about publishing details.
Greg
Still he provides a lot of free knowhow to us all,
in his postings here at Forum.
We are all in different situations and have different concerns,
that will/have to guide our thinkings/doings.
-----------------------------------------------------------------
We can find all the greater joy in that people
like Jonathan Carr
, jcarr at diyAudioand Nelson Pass
, Nelson Pass at diyAudioare in situations, where they can/want to,
so willingly share their "secrets" in amplifier designs.
More people, that are generous givers in one or another way
at this forum, I am sure you can add yourself.
-----------------------------------------------------------------
/halo - knows
what our actions will bring us, in the end:- if you give - others will give to you
- if you withhold - others will withhold from you
THE AKSA AMP & PHILOSOPHY
Folks,
I guess I have some explaining to do. I'm very happy to do this, as DIY is my passion, amplifiers are my life, and 'you're a swell bunch of guys' as the North American contributors might echo......
TOPOLOGY
The AKSA is a Bailey topology, using a single input diff pair, a resistively driven single ended Class A voltage amplifier, a bootstrap current supply to the VAS, and a conventional darlington output stage. It has been tweaked to within a Hertz of its life, and is available in two versions, 55W and 100W, with an upgrade called the 'Nirvana' if desired. Both variants are very compact, with small parts count, and both ratings are conservative.
PERFORMANCE
Modesty forbids, but it will routinely compete with off-the-shelf amplifiers costing five times as much. THD on the 100W AKSA at 1KHz 10W into 8R is 0.0045%, and at 20KHz it is 0.03%. There are plenty of comments on the AKSAs in Audioasylum and the new dedicated forum at audiocircle.com; have a look for yourself. It's not as good sonically as an 845 SET, but it's good for a SS amplifier, and for the money the AKSA is better than reasonable.
COMMERCIAL PHILOSOPHY
I run a shoestring organization, trimmed to the basics, like Bill Baker of Response, or Scott Nixon of DAC Attack. This is my sole income, although I do enjoy a military superannuation from a past life. At this point, my wife draws a modest salary from the business; I do not. The business is self-financing, and growing steadily. Aspen now has two amplifiers, two preamplifiers, and a two way floor standing loudspeaker. Presently I am working on a larger amplifier, a new loudspeaker, and a DAC.
The electronics industry is extremely fast-paced, and protection of Intellectual Property becoming increasingly difficult, strangled by the high price of attorneys and the ineffectiveness of patent law. In an entire amplifier, a change of just 10% to one component no longer infringes the patent, and with so much of the technololgy fabricated in the Far East where laws are different it is almost impossible to guarantee your circuit won't be reverse engineered and used elsewhere, often by mass producers who quickly destroy a customer base. These are the facts of consumer electronics; not if it happens, but when. In this situation you need to make realistic assessments of your product; how much control do you have over it, and for how long?
Possible solution: Protect the circuit as much as possible (by patrolling the net like a 'hawk' as one disgruntled contributor opines!), operate on an 'honor' system, and offer unparalleled service for those constructors who commit with their pockets and build the kit. Service remains king in any technology. If the customer has immediate email recourse to the designer, there's a good chance the problem will be resolved quicker, and this has repercussions for the business model. I must admit that this approach has been quite successful, and I enjoy wonderful relations with many customers around the world who have successfully built the AKSA, love the sound, and seem very happy with the service backup.
The AKSA is not expensive, and sells for around two thirds the price of many expensive, European and US kits. A few observations here and there that buyers get fewer parts with their kits compared to others in the market prompt this response from me. You must pay a little more for a design which uses less parts, because this is clearly superior technology. Complexity does not a better amplifier make, and if you want a complicated amplifier at low cost, build the many offerings in Elektor and tens of DIY magazines like it. I'll say it again; the better designs are simpler, (study Jean Hiraga's designs and refute this!!). There should be more effort to make things simpler in life, because this means they are easier to build, easier to get going, simpler to fix, and cheaper to make. The secret of any good design - no matter what technology - is to identify the components and circuit blocks which determine the performance and pare them back whilst optimizing them; this leads to a better mousetrap. This is the reason General Motors still persist in 2003 with building a pushrod, two valve Small Block, and long may it live!! A design so good, and so effective, and so flexible, and so cheap to make, the Small Block V8 is universally understood, and inexpensive. There is a world of difference between cheap and inexpensive, and similar arguments apply to most of the legendary products of the industrial and consumer world.
AKSA HERITAGE
The AKSA is my third full blown amplifier design. I began at age 46 with a single ended mosfet stage and tube voltage amp called the Glass Harmony. The tube stage took two hours to design. The output stage about six months. I owe much to Nelson for the design of the output stage, although I've turned his Zen amplifier on its head. The interface between the two circuit blocks required 18 months of intensive work, with a lot of input from other, cleverer people than me, such as Kevin O'Connor, from London Power in Canada, another brilliant man, like Nelson. The GH (28W per channel) remains my best amplifier, but it is not commercial, as I am deeply concerned about how to protect the IP, which is utterly unique. However, I plan to release the Class A Glass Harmony in the years ahead as a fully built and tested, off-the-shelf amplifier.
My second amplifier was the Lifeforce, which incorporated a triode in a conventional PP SS Class AB amplifier. It sounded pretty good. The third was the AKSA, which during development slowly took form as 'tube-sounding SS'. This was intended, and involved a LOT of work and testing.
The influences on the AKSA design come from AR Bailey, John Linsley Hood, Doug Self, Jean Hiraga, T. Giesberts, Graham Dicker and Leo Simpson (two very clever Australians), and numerous application notes from Motorola, Hitachi and RCA. It was not developed from Rod Elliott's P3A, although I have looked his design over and agree there is passing resemblance. We must remember that it is not generally possible to look at a schematic and divine how an amp will sound; the sonics exist on a different astral plane to the schematic! In truth, the 'sound' of an amplifier is equal parts schematic, layout and parts; and this needs to be factored into every design.
MY BACKGROUND
I try to keep backstage at Aspen Amplifiers because I despise the cult of personality and want to center the business around the customer and the product. However, since some of the judgements here are a little harsh, and certainly ill-informed, a few words are in order. My previous twenty year career was as an academic in the Australian Army; I am an English and Science/Maths teacher. In later years I switched to computing, and hold a Masters degree in BIT. However, this field is a little dry for me these days, and my real love is for electronics and mechanics. I am not a natural businessman, and have to work very hard on the admin side, but I love people, and enjoy wonderful relations with the customer base. I love what I do with a passion, and will do it until I drop at my PC. Music is an important part of my life; I was classically trained as a pipe organist though I don't play these days.
To those who would ascribe venal or self-serving motives to me, I suggest you email me, ask the questions, read the answers, and perhaps even, buy the product!
In closing, I thank the many contributors who have championed my AKSA, particularly MB from Singapore, who is a committed music lover, passionate DIYer, and good friend.
Thank you for the bandwidth,
Cheers,
Hugh
Folks,
I guess I have some explaining to do. I'm very happy to do this, as DIY is my passion, amplifiers are my life, and 'you're a swell bunch of guys' as the North American contributors might echo......
TOPOLOGY
The AKSA is a Bailey topology, using a single input diff pair, a resistively driven single ended Class A voltage amplifier, a bootstrap current supply to the VAS, and a conventional darlington output stage. It has been tweaked to within a Hertz of its life, and is available in two versions, 55W and 100W, with an upgrade called the 'Nirvana' if desired. Both variants are very compact, with small parts count, and both ratings are conservative.
PERFORMANCE
Modesty forbids, but it will routinely compete with off-the-shelf amplifiers costing five times as much. THD on the 100W AKSA at 1KHz 10W into 8R is 0.0045%, and at 20KHz it is 0.03%. There are plenty of comments on the AKSAs in Audioasylum and the new dedicated forum at audiocircle.com; have a look for yourself. It's not as good sonically as an 845 SET, but it's good for a SS amplifier, and for the money the AKSA is better than reasonable.
COMMERCIAL PHILOSOPHY
I run a shoestring organization, trimmed to the basics, like Bill Baker of Response, or Scott Nixon of DAC Attack. This is my sole income, although I do enjoy a military superannuation from a past life. At this point, my wife draws a modest salary from the business; I do not. The business is self-financing, and growing steadily. Aspen now has two amplifiers, two preamplifiers, and a two way floor standing loudspeaker. Presently I am working on a larger amplifier, a new loudspeaker, and a DAC.
The electronics industry is extremely fast-paced, and protection of Intellectual Property becoming increasingly difficult, strangled by the high price of attorneys and the ineffectiveness of patent law. In an entire amplifier, a change of just 10% to one component no longer infringes the patent, and with so much of the technololgy fabricated in the Far East where laws are different it is almost impossible to guarantee your circuit won't be reverse engineered and used elsewhere, often by mass producers who quickly destroy a customer base. These are the facts of consumer electronics; not if it happens, but when. In this situation you need to make realistic assessments of your product; how much control do you have over it, and for how long?
Possible solution: Protect the circuit as much as possible (by patrolling the net like a 'hawk' as one disgruntled contributor opines!), operate on an 'honor' system, and offer unparalleled service for those constructors who commit with their pockets and build the kit. Service remains king in any technology. If the customer has immediate email recourse to the designer, there's a good chance the problem will be resolved quicker, and this has repercussions for the business model. I must admit that this approach has been quite successful, and I enjoy wonderful relations with many customers around the world who have successfully built the AKSA, love the sound, and seem very happy with the service backup.
The AKSA is not expensive, and sells for around two thirds the price of many expensive, European and US kits. A few observations here and there that buyers get fewer parts with their kits compared to others in the market prompt this response from me. You must pay a little more for a design which uses less parts, because this is clearly superior technology. Complexity does not a better amplifier make, and if you want a complicated amplifier at low cost, build the many offerings in Elektor and tens of DIY magazines like it. I'll say it again; the better designs are simpler, (study Jean Hiraga's designs and refute this!!). There should be more effort to make things simpler in life, because this means they are easier to build, easier to get going, simpler to fix, and cheaper to make. The secret of any good design - no matter what technology - is to identify the components and circuit blocks which determine the performance and pare them back whilst optimizing them; this leads to a better mousetrap. This is the reason General Motors still persist in 2003 with building a pushrod, two valve Small Block, and long may it live!! A design so good, and so effective, and so flexible, and so cheap to make, the Small Block V8 is universally understood, and inexpensive. There is a world of difference between cheap and inexpensive, and similar arguments apply to most of the legendary products of the industrial and consumer world.
AKSA HERITAGE
The AKSA is my third full blown amplifier design. I began at age 46 with a single ended mosfet stage and tube voltage amp called the Glass Harmony. The tube stage took two hours to design. The output stage about six months. I owe much to Nelson for the design of the output stage, although I've turned his Zen amplifier on its head. The interface between the two circuit blocks required 18 months of intensive work, with a lot of input from other, cleverer people than me, such as Kevin O'Connor, from London Power in Canada, another brilliant man, like Nelson. The GH (28W per channel) remains my best amplifier, but it is not commercial, as I am deeply concerned about how to protect the IP, which is utterly unique. However, I plan to release the Class A Glass Harmony in the years ahead as a fully built and tested, off-the-shelf amplifier.
My second amplifier was the Lifeforce, which incorporated a triode in a conventional PP SS Class AB amplifier. It sounded pretty good. The third was the AKSA, which during development slowly took form as 'tube-sounding SS'. This was intended, and involved a LOT of work and testing.
The influences on the AKSA design come from AR Bailey, John Linsley Hood, Doug Self, Jean Hiraga, T. Giesberts, Graham Dicker and Leo Simpson (two very clever Australians), and numerous application notes from Motorola, Hitachi and RCA. It was not developed from Rod Elliott's P3A, although I have looked his design over and agree there is passing resemblance. We must remember that it is not generally possible to look at a schematic and divine how an amp will sound; the sonics exist on a different astral plane to the schematic! In truth, the 'sound' of an amplifier is equal parts schematic, layout and parts; and this needs to be factored into every design.
MY BACKGROUND
I try to keep backstage at Aspen Amplifiers because I despise the cult of personality and want to center the business around the customer and the product. However, since some of the judgements here are a little harsh, and certainly ill-informed, a few words are in order. My previous twenty year career was as an academic in the Australian Army; I am an English and Science/Maths teacher. In later years I switched to computing, and hold a Masters degree in BIT. However, this field is a little dry for me these days, and my real love is for electronics and mechanics. I am not a natural businessman, and have to work very hard on the admin side, but I love people, and enjoy wonderful relations with the customer base. I love what I do with a passion, and will do it until I drop at my PC. Music is an important part of my life; I was classically trained as a pipe organist though I don't play these days.
To those who would ascribe venal or self-serving motives to me, I suggest you email me, ask the questions, read the answers, and perhaps even, buy the product!
In closing, I thank the many contributors who have championed my AKSA, particularly MB from Singapore, who is a committed music lover, passionate DIYer, and good friend.
Thank you for the bandwidth,
Cheers,
Hugh
I understand that your I.P. costs money
The heart of every great product is Intellectual Property. Your I.P. is obviously well founded and heavily expensive at the R & D end. I appreciate that and look forward to being able to afford one of your AKSA units someday. Thank you for the extended and earnest response. I have no desire to steal your ideas I just bemoan the fact that I can't afford them yet. And they include heatsinks which are expensive and often hard to find, for hobbiest, unto themselves.
The heart of every great product is Intellectual Property. Your I.P. is obviously well founded and heavily expensive at the R & D end. I appreciate that and look forward to being able to afford one of your AKSA units someday. Thank you for the extended and earnest response. I have no desire to steal your ideas I just bemoan the fact that I can't afford them yet. And they include heatsinks which are expensive and often hard to find, for hobbiest, unto themselves.
Hugh,
Thanks for telling us a bit about the topology of your amps. It
seems your design philosphy is a minimalistic approach to
the "traditional" design and that much of the "secret" is in
the choice of component values (in contrast to focussing on
complex topologies, that is). I know from previous discussions
that you take a lot of care about the details and listening.
Thanks for telling us a bit about the topology of your amps. It
seems your design philosphy is a minimalistic approach to
the "traditional" design and that much of the "secret" is in
the choice of component values (in contrast to focussing on
complex topologies, that is). I know from previous discussions
that you take a lot of care about the details and listening.
The topology of the AKSA seems to be superficially similar to the Elektor BDX66/67 darlington amplifier by Giesberts, first published in 1982. That was a very nice sounding amplifier, and widely-copied. Quite naturally, since it sounds much better than the MOSFET-based crescendo published in the December 1982 issue, which received a lot of attention then.
Linuxguru,
Superficially, yes, you are right. However, there is one fundamental difference with this design over almost all others which explains the almost tube like sonics. I won't go into it, and I can count the number of people on one hand who have picked this from the schematic, but unquestionably it is the reason this amp has been so very popular and continues to rank high on the list.....
I would say a Linux man in Bangalore would be VERY busy!! It is good to see you can still smell the roses on your days off!
Cheers,
Hugh
Superficially, yes, you are right. However, there is one fundamental difference with this design over almost all others which explains the almost tube like sonics. I won't go into it, and I can count the number of people on one hand who have picked this from the schematic, but unquestionably it is the reason this amp has been so very popular and continues to rank high on the list.....
I would say a Linux man in Bangalore would be VERY busy!! It is good to see you can still smell the roses on your days off!
Cheers,
Hugh
Hugh, thanks for the reply. I haven't seen the AKSA schematic, and I respect your decision to restrict its distribution, while broadly describing the topology here.
I've been playing around with a commercially-available derivative of the Giesberts darlington design, made in India. It sounded terrible out of the box, but I was able to fix most of its problems and it sounds fairly clean now for its price. It used 2 x bc558 for the input stage, bc547 for the single-ended voltage amplifier and TIP122/127 for the Class-B output. I swapped out the TIP122/127 with TIP142/147, the bc547 with a 2sd667, and changed the quiescent crossover biasing on the output darlingtons from about 2.5 Vbe to about 3.1 Vbe.
There are a couple of things which I'm not sure about:
1) It has a 100R resistor between the VAS and the base of each darlington of the output stage. Apart from possibly squelching spurious oscillations, is there any purpose for this? The darlingtons have a very high hFE, so the 100R resistor is useless as a protection device against a short on the output. It's also connected between two AC high-impedance nodes (collector of VAS and base of darlingtons), so it can't contribute much, if anything, to the AC picture. If at all anything, it may contribute a very slight HF roll-off in conjunction with Cbe of the output darlingtons. Is it advisable to short out the two 100R resistors?
2) What is the effect of substituting a small-signal darlington in place of the standard single-ended transistor in the VAS? Gain-bandwidth is generally lower for small-signal darlingtons, but this transistor has an external miller capacitance (dominant-pole compensation) at any rate. In conjunction with the bootstrapped current source, this limits the slew rate at the collector of the VAS. What are the pros and cons of substituting a darlington here? I substituted a 2sd1978 small-signal darlington here, and I could not discern much of a difference in sound quality - perhaps a little reduced crossover distortion due to higher loop gain.
I've been playing around with a commercially-available derivative of the Giesberts darlington design, made in India. It sounded terrible out of the box, but I was able to fix most of its problems and it sounds fairly clean now for its price. It used 2 x bc558 for the input stage, bc547 for the single-ended voltage amplifier and TIP122/127 for the Class-B output. I swapped out the TIP122/127 with TIP142/147, the bc547 with a 2sd667, and changed the quiescent crossover biasing on the output darlingtons from about 2.5 Vbe to about 3.1 Vbe.
There are a couple of things which I'm not sure about:
1) It has a 100R resistor between the VAS and the base of each darlington of the output stage. Apart from possibly squelching spurious oscillations, is there any purpose for this? The darlingtons have a very high hFE, so the 100R resistor is useless as a protection device against a short on the output. It's also connected between two AC high-impedance nodes (collector of VAS and base of darlingtons), so it can't contribute much, if anything, to the AC picture. If at all anything, it may contribute a very slight HF roll-off in conjunction with Cbe of the output darlingtons. Is it advisable to short out the two 100R resistors?
2) What is the effect of substituting a small-signal darlington in place of the standard single-ended transistor in the VAS? Gain-bandwidth is generally lower for small-signal darlingtons, but this transistor has an external miller capacitance (dominant-pole compensation) at any rate. In conjunction with the bootstrapped current source, this limits the slew rate at the collector of the VAS. What are the pros and cons of substituting a darlington here? I substituted a 2sd1978 small-signal darlington here, and I could not discern much of a difference in sound quality - perhaps a little reduced crossover distortion due to higher loop gain.
Hi LXG,
You are asking difficult questions which require long explanation! Here goes........!
The base of a transistor can exhibit negative resistance in certain circumstances, particularly where they are being turned on and off constantly. This means reductions in potential could result in higher base current, a surefire recipe for oscillation. A base stopper resistor of 100R confers positive resistance to the base at all times and prevent any spurious oscillation of the device. This is a common problem with the emitter follower and base stoppers are, in my view, mandatory on Class AB amplifiers. They do reduce transconductance of the output stage, no question, but there is a high feedback factor here and the loss of transconductance is essentially linear and easily corrected by feedback because the base resistance is principally ohmic.
In my experience this merely reduces loading on the tail of the LTP, improving feedback factor by a few db and resulting in slightly lower distortion as you say. It does not improve sonics.
A better, often used technique is to drive the VAS with an emitter follower, a la Self, as this also reduces loading on the LTP. However, even that approach has been criticised because the emitter follower passes a similar current to the VAS and thus its base bias is much the same. However, it does have a higher (and more stable) input impedance than a common emitter amplifier, particularly if no degeneration has been used.
The Self amp is capable of very high performance, at least in measured terms, and can be tweaked for good sonics. Persevere, you will produce something pretty good, but more linear transistors than the TIP are advised, such as the Toshiba C5200/A1943 series, which are very good and inexpensive. And use Self's Type II EF output stage - with the right capacitor (100nF, foil polyprop) with low dielectric absorption, it sounds very good indeed.
Hope this helps,
Cheers,
Hugh
You are asking difficult questions which require long explanation! Here goes........!
The base of a transistor can exhibit negative resistance in certain circumstances, particularly where they are being turned on and off constantly. This means reductions in potential could result in higher base current, a surefire recipe for oscillation. A base stopper resistor of 100R confers positive resistance to the base at all times and prevent any spurious oscillation of the device. This is a common problem with the emitter follower and base stoppers are, in my view, mandatory on Class AB amplifiers. They do reduce transconductance of the output stage, no question, but there is a high feedback factor here and the loss of transconductance is essentially linear and easily corrected by feedback because the base resistance is principally ohmic.
In my experience this merely reduces loading on the tail of the LTP, improving feedback factor by a few db and resulting in slightly lower distortion as you say. It does not improve sonics.
A better, often used technique is to drive the VAS with an emitter follower, a la Self, as this also reduces loading on the LTP. However, even that approach has been criticised because the emitter follower passes a similar current to the VAS and thus its base bias is much the same. However, it does have a higher (and more stable) input impedance than a common emitter amplifier, particularly if no degeneration has been used.
The Self amp is capable of very high performance, at least in measured terms, and can be tweaked for good sonics. Persevere, you will produce something pretty good, but more linear transistors than the TIP are advised, such as the Toshiba C5200/A1943 series, which are very good and inexpensive. And use Self's Type II EF output stage - with the right capacitor (100nF, foil polyprop) with low dielectric absorption, it sounds very good indeed.
Hope this helps,
Cheers,
Hugh
An idea from this post by jcx in this forum:
http://www.diyaudio.com/forums/showthread.php?postid=1008105#post1008105
Basically, he suggests using bootstrapped feedback from the output, using a resistive divider with the same ratio as the voltage feedback to the inverting input of the LTP. This bootstrapped feedback is however added to the quiescent current fed to the emitters of the LTP. If the ratios of the resistive dividers match, the AC input impedance at the bases of the LTP increases dramatically.
This is trivial to implement in the Self-Giesberts-AKSA topology and derivatives:
Just add a resistor from the upper end (at the junction of the two resistors) of the VAS current-source bootstrap capacitor, to the emitters of the PNP input LTP.
As an example, consider a PNP LTP with a resistor of 10k from an 8.2v voltage source (say a zener) being the current source of (8.2 - 0.7)/10k = 0.75 mA.
If the main voltage-series feedback resistive divider is 1k:39k for a gain of 40, choose a current-source bootstrap resistor of 390k from the upper end of the bootstrap cap to the emitters of the LTP - this adds a small quiescent current component to the LTP. If the bootstrap junction is at +20v, this additional current is about (20-0.7)/390k ~= 50 uA, which doesn't alter the LTP DC biasing significantly. However, the AC impedance at the input increases dramatically.
This trick can be used independently of the collector load of the LTP - either a current-mirror, or a resistive load feeding the VAS.
I tried it with the Giesberts darlington amplifier derivative, and it is sonically viable - it sounds no worse than without the feedback, and probably much better at high input amplitudes, because both the transistors in the LTP are now operating at almost constant Vbe. If this hasn't already been incorporated in AKSA, it is a viable upgrade.
http://www.diyaudio.com/forums/showthread.php?postid=1008105#post1008105
Basically, he suggests using bootstrapped feedback from the output, using a resistive divider with the same ratio as the voltage feedback to the inverting input of the LTP. This bootstrapped feedback is however added to the quiescent current fed to the emitters of the LTP. If the ratios of the resistive dividers match, the AC input impedance at the bases of the LTP increases dramatically.
This is trivial to implement in the Self-Giesberts-AKSA topology and derivatives:
Just add a resistor from the upper end (at the junction of the two resistors) of the VAS current-source bootstrap capacitor, to the emitters of the PNP input LTP.
As an example, consider a PNP LTP with a resistor of 10k from an 8.2v voltage source (say a zener) being the current source of (8.2 - 0.7)/10k = 0.75 mA.
If the main voltage-series feedback resistive divider is 1k:39k for a gain of 40, choose a current-source bootstrap resistor of 390k from the upper end of the bootstrap cap to the emitters of the LTP - this adds a small quiescent current component to the LTP. If the bootstrap junction is at +20v, this additional current is about (20-0.7)/390k ~= 50 uA, which doesn't alter the LTP DC biasing significantly. However, the AC impedance at the input increases dramatically.
This trick can be used independently of the collector load of the LTP - either a current-mirror, or a resistive load feeding the VAS.
I tried it with the Giesberts darlington amplifier derivative, and it is sonically viable - it sounds no worse than without the feedback, and probably much better at high input amplitudes, because both the transistors in the LTP are now operating at almost constant Vbe. If this hasn't already been incorporated in AKSA, it is a viable upgrade.
Hmmm LXG,
the collector (cf. output) impedance of a CCS, even an imperfect one, is in the megohms range. If you bootstrap the supply rail for this CCS, you increase this impedance somewhat because now there is no Vak variation across the CCS active device.
Considering we are dealing with inter-emitter impedances limited to 2 x degeneration, say not much more than 200R, I'm not surprised that there is no sonic effect. And the current variation due to smallish differences in voltage across the CCS will not be very large at all, perhaps less than 3%.
OTOH, if you replaced the bootstrapped CCS with a simple, single resistor, it would now have constant voltage across it with signal throughput, and it would thus have a dynamic impedance in the megohms, without the usual Cob limitations of an active device.
THIS might sound a bit better - though I doubt you'd hear it. However, consider further, if you changed the ratio of the bootstrap divider network so that it were dissimilar to the CLG, what then would be the effect on the sound?
I know the answer to this question, in fact, so I'd suggest you have a shot, as the implementation is trivial, as you rightly point out.......
Cheers,
Hugh
the collector (cf. output) impedance of a CCS, even an imperfect one, is in the megohms range. If you bootstrap the supply rail for this CCS, you increase this impedance somewhat because now there is no Vak variation across the CCS active device.
Considering we are dealing with inter-emitter impedances limited to 2 x degeneration, say not much more than 200R, I'm not surprised that there is no sonic effect. And the current variation due to smallish differences in voltage across the CCS will not be very large at all, perhaps less than 3%.
OTOH, if you replaced the bootstrapped CCS with a simple, single resistor, it would now have constant voltage across it with signal throughput, and it would thus have a dynamic impedance in the megohms, without the usual Cob limitations of an active device.
THIS might sound a bit better - though I doubt you'd hear it. However, consider further, if you changed the ratio of the bootstrap divider network so that it were dissimilar to the CLG, what then would be the effect on the sound?
I know the answer to this question, in fact, so I'd suggest you have a shot, as the implementation is trivial, as you rightly point out.......
Cheers,
Hugh
> OTOH, if you replaced the bootstrapped CCS with a simple, single resistor, it would now have constant voltage across it with signal throughput, and it would thus have a dynamic impedance in the megohms, without the usual Cob limitations of an active device.
Actually, this is the configuration I'm using. A simple resistor from a zener to the emitters of the LTP (I noticed that the authorized equivalent schematic of AKSA dispenses with the zener and uses a 1000uF cap instead).
>if you changed the ratio of the bootstrap divider network so that it were dissimilar to the CLG, what then would be the effect on the sound?
Hmm - if the ratio of the bootstrapped FB is less than that of the main voltage-series FB divider, the dynamic impedance of the CCS doesn't increase as much, but it still works. If the ratio is greater, I guess there could be dynamic negative-impedance structures and possible instability/oscillations. I'll try it out and see what happens.
Actually, this is the configuration I'm using. A simple resistor from a zener to the emitters of the LTP (I noticed that the authorized equivalent schematic of AKSA dispenses with the zener and uses a 1000uF cap instead).
>if you changed the ratio of the bootstrap divider network so that it were dissimilar to the CLG, what then would be the effect on the sound?
Hmm - if the ratio of the bootstrapped FB is less than that of the main voltage-series FB divider, the dynamic impedance of the CCS doesn't increase as much, but it still works. If the ratio is greater, I guess there could be dynamic negative-impedance structures and possible instability/oscillations. I'll try it out and see what happens.
Have any of you guys thought about this fact:
By using a straight resistor instead of CCS or bootstrap to load the IP
dif pair, the IP devices work in almost constant power mode.
IOW as the IP voltage rises, and the voltage across IP (N ch) device
gets less, the current increases proportionately.
Do some simple calcs you will see that it comes very close to zero
power modulation.
Maybe this is why a lot of people prefer R loaded IP pair?
cheers
Terry
By using a straight resistor instead of CCS or bootstrap to load the IP
dif pair, the IP devices work in almost constant power mode.
IOW as the IP voltage rises, and the voltage across IP (N ch) device
gets less, the current increases proportionately.
Do some simple calcs you will see that it comes very close to zero
power modulation.
Maybe this is why a lot of people prefer R loaded IP pair?
cheers
Terry
> By using a straight resistor instead of CCS or bootstrap to load the IP
dif pair, the IP devices work in almost constant power mode.
Maybe I'm missing something, but that's true of any diff amp, regardless of the current source, as long as the DC component in the input signal has been filtered out.
If there is a DC component, it's not generally true even in the resistive current source case. Consider a resistive current source R biased from (say) 0.5 Vs. The current through it is (Vs/2 - Vbe)/R. Consider an input DC voltage equal to the -ve supply rail. The current through the resistor (and hence, the LTP) has increased to (1.5Vs - Vbe)/R, ~3x the quiescent current. But the Vce drop across the LTP has dropped from (Vbe+Vs) all the way down to Vbe, and that ratio could be much more than 3x, depending on Vs. The power dissipated in the LTP drops as a result.
> Maybe this is why a lot of people prefer R loaded IP pair?
Plain resistive sources have some advantages, including the fact that it's easier to make a real resistor that looks like an ideal resistor, without much by way of parasitic capacitances, etc. The simplicity of a resistor is hard to beat from a minimalist point of view.
Ideally, you want to use active devices only to obtain gain - either current gain or voltage gain. Everything else should be managed with passives. The bootstrapped resistive current source contributes a bit towards this goal.
dif pair, the IP devices work in almost constant power mode.
Maybe I'm missing something, but that's true of any diff amp, regardless of the current source, as long as the DC component in the input signal has been filtered out.
If there is a DC component, it's not generally true even in the resistive current source case. Consider a resistive current source R biased from (say) 0.5 Vs. The current through it is (Vs/2 - Vbe)/R. Consider an input DC voltage equal to the -ve supply rail. The current through the resistor (and hence, the LTP) has increased to (1.5Vs - Vbe)/R, ~3x the quiescent current. But the Vce drop across the LTP has dropped from (Vbe+Vs) all the way down to Vbe, and that ratio could be much more than 3x, depending on Vs. The power dissipated in the LTP drops as a result.
> Maybe this is why a lot of people prefer R loaded IP pair?
Plain resistive sources have some advantages, including the fact that it's easier to make a real resistor that looks like an ideal resistor, without much by way of parasitic capacitances, etc. The simplicity of a resistor is hard to beat from a minimalist point of view.
Ideally, you want to use active devices only to obtain gain - either current gain or voltage gain. Everything else should be managed with passives. The bootstrapped resistive current source contributes a bit towards this goal.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.