You should read more carefully. Bandwidth and Slew rate are different thing. Inadequate slew rate??? Increase slew rate from 200V/us to 1500V/us and you can hear the difference.
dado
This argument seems to have been brought about by an english language problem.
Post1104 seems to show Esp knows little and yet post1118 states he agrees with the norm.
But read 1104 again and be kind enough to insert the correct english words into the text and I believe you will see what he meant but did not write.
Post1104 seems to show Esp knows little and yet post1118 states he agrees with the norm.
But read 1104 again and be kind enough to insert the correct english words into the text and I believe you will see what he meant but did not write.
But did i make a mistake, feeling some kind of irony in your question ?Are-you this kind of audiophile thinking there is no salvation without vinyl records, Quad speakers, tube amps and silver plated wires Teflon isolated with cryogenic treatment?
Yes , you did miss it and i'm the type of guy replacing my 20w SET with an SSA V-12
...Now If you see then have a drink and a smile ...it's all good............
Easy Fred,
a v-8 cant sound like an V-12 , go for the 4 extra cylinders..[/QOUTE] Easy Fred, a v-8 cant sound like an V-12 , go for the 4 extra cylinders..[/QUOTE]
Hi Wayne,
If I have the money, why not. I tried matching hFE the other day and out of 20 pairs, I'm lucky enough to come up with 4 pairs. That means, I need hundred pairs or more for a V12 amp. Right now, economy isn't good so I'm just going for what I can afford.
a v-8 cant sound like an V-12 , go for the 4 extra cylinders..[/QUOTE]Hi Wayne,
If I have the money, why not. I tried matching hFE the other day and out of 20 pairs, I'm lucky enough to come up with 4 pairs. That means, I need hundred pairs or more for a V12 amp.
Right now, economy isn't good so I'm just going for what I can afford. 
I do not want to burst anyone's bubble, but matching components is a little rediculous to say the least.
I had a production run with some 100 amps. I decided I wanted one for myself and to match every pair of transistors perfectly at several values of Ic.
After adjusting off-set and bias it was no different from any of the other 99. Listening to it and measuring it made absolutely no difference at all.
I have since matched nothing apart from making sure the same part number is written on them and they come from the same manufacturer.
Maybe there is an insignificant difference if using zero negative feedback, but I doubt it.
I had a production run with some 100 amps. I decided I wanted one for myself and to match every pair of transistors perfectly at several values of Ic.
After adjusting off-set and bias it was no different from any of the other 99. Listening to it and measuring it made absolutely no difference at all.
I have since matched nothing apart from making sure the same part number is written on them and they come from the same manufacturer.
Maybe there is an insignificant difference if using zero negative feedback, but I doubt it.
Last edited:
About this speed of amps question, some precisions.
In open loop, bandwitch and slew rate are correlated. CR introduce a kind of artificial increase of the bandwidth, but cannot do nothing regarding slew rate, witch is the ability of your amp to follow transients (fast changes of amplitude).
To give an image, at an average speed of 60miles/h an old VW will be able to follow a Ferrari between two cities. Never to follow his accelerations keeping the same distance.
When we where in analog world for our audio sources (Vinyls) it was not too much a problem: the bandwidth and the slew rate of the signals where limited in the same time and in a linear and correlated way by the inertia of the mechanical grooving heads and their bandwidth filters. Now, we live in a digital world, and signals from CDs can present slew rate more than ten time higher than a linear source with a 20 000hz sinusoidal limit.
Deigning amps, the good practice is to set each stage speed faster than the one preceding. And to design them for they limit they speed/bandwidth in a linear way. That where Output Fets presents an advantage against bipolar: they are faster, and more linear at high frequencies.
The Current feed back is a way to get better slew rate, with negative feedback, because the correction signal (CR) will not across a slow and non linear input stage, introducing an additional delay before to be subtract to the the original input signal. CR can present high levels at transients, and generate distortion if they are applied in a non linear way. Reason why CR had a bad reputation during some times.
Cascodes are too good practices to get rid of distortions at Hfs. That where Lazy Cat is accurate all the way in his designs, and why everyone find his designs are sounding "easy".
The way to limit the TIM is trivial, in a sense: never apply a signal with edges the amp cannot follow. So, build the fastest (slew rate =open loop bandwidth) amp you can, and limit the max slew rate of the incoming signals in the input, in a linear way -means by a low pass filter- above the amp open loop capabilities.
I believe that, for preamps, 50V/µs is a good value, according to the 1V level ( it needs much more in DACs output stages before filtering) and a perfect amp have to follow this speed by its gain: My SSA have 30 gain factor, and a 1500Vµs of slew rate: i'm at least in the rails.
I reduce the bandwidth in the input stage by a low pass filter at 1Mhz, to get a near flat phase response curve up to 10 000hz.
About Class D, by design, the slew rate is very hight, as the power stages are using a high clock square waves frequency: They present in theory a good match to Digital sources.
Some friends of mine (in witch i believe too ) just changed his high end previous amps for such D amps in his active system, and report sound is even better, and said mr Hiraga was astonished by his set. That why i want to try, i do not believe in fashion. I believe direct comparison with my actual SSA/Crescendo amps will be a difficult challenge for my future D amps. Will see witch will be the winner. I do not believe in "thin" sounding or any bad reputation of class D (coming from cheap PA amps) witch are those kind of audiophiles rumors i do not care about. And i believe that, some times, reducing distortion give the feeling of "simplifying" the signal. A better amp can sound "thiner" if some agreeable distortion is removed. I believe in "transparency", not agreeable sounding amps, i'm not an "audiophile".
SSA amp is very transparent and easy, not "agreeable".
In open loop, bandwitch and slew rate are correlated. CR introduce a kind of artificial increase of the bandwidth, but cannot do nothing regarding slew rate, witch is the ability of your amp to follow transients (fast changes of amplitude).
To give an image, at an average speed of 60miles/h an old VW will be able to follow a Ferrari between two cities. Never to follow his accelerations keeping the same distance.
When we where in analog world for our audio sources (Vinyls) it was not too much a problem: the bandwidth and the slew rate of the signals where limited in the same time and in a linear and correlated way by the inertia of the mechanical grooving heads and their bandwidth filters. Now, we live in a digital world, and signals from CDs can present slew rate more than ten time higher than a linear source with a 20 000hz sinusoidal limit.
Deigning amps, the good practice is to set each stage speed faster than the one preceding. And to design them for they limit they speed/bandwidth in a linear way. That where Output Fets presents an advantage against bipolar: they are faster, and more linear at high frequencies.
The Current feed back is a way to get better slew rate, with negative feedback, because the correction signal (CR) will not across a slow and non linear input stage, introducing an additional delay before to be subtract to the the original input signal. CR can present high levels at transients, and generate distortion if they are applied in a non linear way. Reason why CR had a bad reputation during some times.
Cascodes are too good practices to get rid of distortions at Hfs. That where Lazy Cat is accurate all the way in his designs, and why everyone find his designs are sounding "easy".
The way to limit the TIM is trivial, in a sense: never apply a signal with edges the amp cannot follow. So, build the fastest (slew rate =open loop bandwidth) amp you can, and limit the max slew rate of the incoming signals in the input, in a linear way -means by a low pass filter- above the amp open loop capabilities.
I believe that, for preamps, 50V/µs is a good value, according to the 1V level ( it needs much more in DACs output stages before filtering) and a perfect amp have to follow this speed by its gain: My SSA have 30 gain factor, and a 1500Vµs of slew rate: i'm at least in the rails.
I reduce the bandwidth in the input stage by a low pass filter at 1Mhz, to get a near flat phase response curve up to 10 000hz.
About Class D, by design, the slew rate is very hight, as the power stages are using a high clock square waves frequency: They present in theory a good match to Digital sources.
Some friends of mine (in witch i believe too ) just changed his high end previous amps for such D amps in his active system, and report sound is even better, and said mr Hiraga was astonished by his set. That why i want to try, i do not believe in fashion. I believe direct comparison with my actual SSA/Crescendo amps will be a difficult challenge for my future D amps. Will see witch will be the winner. I do not believe in "thin" sounding or any bad reputation of class D (coming from cheap PA amps) witch are those kind of audiophiles rumors i do not care about. And i believe that, some times, reducing distortion give the feeling of "simplifying" the signal. A better amp can sound "thiner" if some agreeable distortion is removed. I believe in "transparency", not agreeable sounding amps, i'm not an "audiophile".
SSA amp is very transparent and easy, not "agreeable".
Last edited:
[/QOUTE] Easy Fred,
Hi Wayne,
If I have the money, why not. I tried matching hFE the other day and out of 20 pairs, I'm lucky enough to come up with 4 pairs. That means, I need hundred pairs or more for a V12 amp.
Right now, economy isn't good so I'm just going for what I can afford. [/QUOTE]Yikes ......
OK... this is in contrast to conventional wisdom, now will apply to the SSA also, or does this design call for matching outputs.....
You wield an interesting double edge sword ESP..
So your friend and Mr Hiraga like the agreeable sound of added class-d distortion, but you will not accept the opinions of audiophiles, because of the removal of said agreeable distortion ...
...
Thanks Nico, you just saved Fred and i a bundle .....
Fred have enuff outputs for two V12's.....and "Audiophiles prefer to listen to music at a quality level that is as close to the original performance as possible using high fidelity components"...Wiki
See music .. wahlah ... your an audiophile .....
...
Last edited:
I am sorry but I differ with the statement. Audiophiles listen to snippets of impressive sounding stuff and they use irrelevant adjectives like speed, slam wham and after the two second snippet sound follows it with another snippet of another impressive sounding passage. I have yet to meet an audiophile that will sit and enjoy an entire performance. Two seconds into a performance he is already thinking what he wants to hear next,
By The way Christophe, my SSA earlier tonight I could measure 368V/uS slew and by carefully optimizing all components I ended up with 1378V/uS SR - but I must say very honestly it did not move me anymore than before, the music still sounded exactly like before.
By The way Christophe, my SSA earlier tonight I could measure 368V/uS slew and by carefully optimizing all components I ended up with 1378V/uS SR - but I must say very honestly it did not move me anymore than before, the music still sounded exactly like before.
They report a "no" Class-D distortions.
With reported 0.005% measured distortion+noise, and near no measurable IM, where do you can find an add ?
Do you generalize any bad experience, or just report audiophile rumors ?
Class D is, i'm sure of that, the future of power audio technologies. May-be time to change your mind against your prejudices ?
As measures are not the whole thing, time to experiments, listens, improvements.
I will not be alive in 30 years, but i remember similar reactions, 30 years ago, when i tried to convince idiophiles of the advantages of Current feed back topology, and solid state.
Seems they love to live in the past, in a museum, with their tubes, old turntables, old 33rpms, refurbished bad old esoteric speakers etc...
The don't listen to music, they listen to their hifi equipment. In fact they don't even listen, they look-at, demonstrate, and talk about!
http://www.dcx2496.fr/img/Img00160.jpg
I believe that. Those two numbers are quite a hight level of performance. And differences can be more thin than our memory is able to distinguish, or those differences introduced by humidity or air temperature ;-)
I am curious to know where do you attempt this improvement ? Changing physical implantation ? Changing components ? Playing with impedances and currents ?
Agree what you said about audiophiles. You describe exactly those i know. they live in a show room, trying to convince themselves, (with no success), they own the ultimate system. And each time they listen to something better, they are deseperated. That do not let so much time to listen to music.
A far i'm concerned, i'm more enjoyed to find a new album i love, that i can listen even on my phones with my mobile, than a new amplifier. And my system was the same during 20 years. ;-)
A good movie gives more pleasure than a good theater screen where you make projections of various spectacular movies scenes.
In fact, this a different pleasure to build systems, learn more, and reach goals in musical reproduction, that you can use not so often in your life's free time. Sitting there and closing you eyes in the dark, listening a good album. Specially if you have wife, kids and dogs ;-)
There is a good way to know if we are an audiophile or not. When a friend comes in your home, did you talk about music, did you make them discover the last album you love, or did they bring the last they love, or did you talk with them about your equipment ?
Last edited:
Guys,
Not to be rude about this, but could we get back to the SSA topology?
Question of Lazy Cat:
Have you tried the SSA with asymmetrical source resistors? This, I understand, increases H2 which has a masking effect on the higher harmonics, and is reported to improve sound quality.
The fact that it can be verified easily is a bonus.....
Hugh
Not to be rude about this, but could we get back to the SSA topology?
Question of Lazy Cat:
Have you tried the SSA with asymmetrical source resistors? This, I understand, increases H2 which has a masking effect on the higher harmonics, and is reported to improve sound quality.
The fact that it can be verified easily is a bonus.....
Hugh
Hello Nico
I am about to put together one SSA and evaluate also. I will follow the 40W Lazycat example as enhanced by you. Early info from a friend who attempted it, is it was a donkey to bias up. Looks like every Vbe difference to local bias resistors in the drivers and small value changes for the current feeding resistors to input stage can easily throw the symmetric current bridge out of balance. After a few TO-92s had bitten the dust due to random over-bias as drivers, Philips original BD140/BD139 survived the tweaks and stayed. No Rf step network capacitors were needed for stability. Now after a change I will explain, they run cold anyway. Rest are A970/C2240 GR hfe grade, and K1058/J162 Renesas. Without any gate resistors it showed a jittery bias initially when having the TO-92 drivers, but not sure why, I wasn't there. It behaved itself with 100R stoppers. The main solution we ended up with over telephone discussions, was to replace the 330R common resistor in the drivers with an LED. That made it bias right away without much offset and without half power stage being on or off in random, or erratically shooting to Amperes bias together as it did before. Turned that 330R brown at a point too. It can still wander higher by 50-100mA for bias current value as it heats up during first hour, but its safe. It settles to ~200mA total current draw per rail. He reports a very good subjective result already, with no noises of any kind, and I will arrange to inspect the prototype for real soon, will report additionally. I attach the tweaked schematic that resides in that working veroboard build for anyone who may find it useful. This one is a a bit more sensitive for 25dB practical gain also. R11 is a trimmer. D6 is a 1.85 Vf green LED on DMM diode test. I will try for close Vbe parts when I will attempt mine. To tie with quoting you above, it's likely that this one parameter is important here, and/or resorting to resistor values fine tweaks than any hfe worries. Will look testing for 28dB gain and 970/2240 as drivers along with higher R5,R6,R13,R14 for lower THD, and unequal gate stoppers initially as well. Don't know if 5pF-10pF comp caps will be needed across R13,14 as in your prototype if everything will go well with small faster drivers, have to see in practice. Did you use those for stability issues, or for better square waves?
Best
Attachments
Well done Salas and nice feed-back regarding your findings. Of course all information that you supplied is useful and relevant information, thank you.
Unless you play music to the extent that you approach rail voltages and clipping is unsymmetrical, I believe that some minor offset is not bothersome to anyone. The offset may be measured by some distortion analyzers and reported as higher THD although it is not really the case.
I am unfortunately not a purist and my objective is to listen to music at volume levels that I feel comfortable with therefor I would seldom tweak a system to the nth degree because once started you never stop. It is similar to the guy that tunes his Alfa Romeo every week-end, there is no real reason to do this, it is just customary and part of the ritual.
Once this amplifier settles at operating temperature and playing music you will never know whether an offset of one microvolt or one hundred millivolt actually exists.
The thing with this amplifier is that even changing the gain affects both off-set and bias simply because it is a balanced bridge, but whether output is actually in perfect balanced or slightly unbalanced it still is a bridge and will remain stable.
What I might add is that if you tie transistors together that is not matched then you do have a variance problem, but if they are some distance apart it is not a problem as each will stabilize at ambient plus. If it is important to glue them together then I would agree that they should be matched as ones temperature would affect the other.
Unless you play music to the extent that you approach rail voltages and clipping is unsymmetrical, I believe that some minor offset is not bothersome to anyone. The offset may be measured by some distortion analyzers and reported as higher THD although it is not really the case.
I am unfortunately not a purist and my objective is to listen to music at volume levels that I feel comfortable with therefor I would seldom tweak a system to the nth degree because once started you never stop. It is similar to the guy that tunes his Alfa Romeo every week-end, there is no real reason to do this, it is just customary and part of the ritual.
Once this amplifier settles at operating temperature and playing music you will never know whether an offset of one microvolt or one hundred millivolt actually exists.
The thing with this amplifier is that even changing the gain affects both off-set and bias simply because it is a balanced bridge, but whether output is actually in perfect balanced or slightly unbalanced it still is a bridge and will remain stable.
What I might add is that if you tie transistors together that is not matched then you do have a variance problem, but if they are some distance apart it is not a problem as each will stabilize at ambient plus. If it is important to glue them together then I would agree that they should be matched as ones temperature would affect the other.
Last edited:
You'r right, Nico. Looking at the offset of any amp with an oscillo in the millivolt range can be pretty depressing, but it is like measuring the changes in earth rotation during the start of a formula one race ;-)
As long as your bass speaker doesn't moves or you can hear an annoying "Clock" connecting-them, i believe, on my side, there is no problem.
I do not care, as you, to glue together the input devices, but i take care to let time for my system stabilize in temperature before listening to music, as i do when i tuned-it.
Power on and a coffee cup.
And, if i have a problem with a rock'nroll offset , i prefer to add a condenser in the CR loop to reduce the gain at low frequencies. I know, it is supposed to be bad if you need to use electronic capacitors, but, when one cannot hear any degradation of his pleasure...
Last edited:
An example could be had from a Wheatstone Bridge for those who remembers the guy. It is used to measure an unknown quantity (resistance, capacitance, etc) by means of balancing the currents in the two opposing legs of the bridge. You could measure miniscule difference in components because of its differential in current and the resulting off-set in potential.
Lazy Cat may or may not agree with my explanation but this amp functions in a similar manner. At switch on, regardless of how well things are matched, the bridge will "amplify" every imbalance in component and the changes will remain apparent until all components have settled and stabilized. Which is what Salas mention here. Even tiny imbalances in current in the two legs would be visibly changing continually. I know what was to be expected so it never bothered me and there is no damage done to your speakers. 100 mV offset would be dissipating 80mW in your 8 ohm speaker for a short while. You can listen to music immediately if you like, I doubt if any but the purest of platinum ear guys ever detect this.
That is why I said before this amp should be adjusted when stable, ten minutes later checked and re-adjusted, until you are satisfied with what you see and that will be the amp's stable state.
If you feel it is the audiophile thing to do and wait until your amp is "warmed up" before listening to it, then this is the perfect amp for the audiophile)
One would assume that the negative feedback should correct this phenomena, but it actually could also worsen it. I tend to think an hour is a little over-stated because mine does stabilise within maybe ten or fifteen minutes. Furthermore, mine drifts both positive and negative, pretty wildly at first, but once at operating temperature it is rock solid, well a few mV up and down. But don't fear fellows this is neither serious nor any form of failure.
I hope my explanation is of some help. LC may want to chip in his five pence worth, but let me assure you, if I thought Lazy Cat was bull dusting us I would have accused him as the ultimate sinner early in the thread.
I have been very pleased with the performance of this simple little amp for several weeks now, it remains one of the most satisfying products I have made and with very little effort. It worked first time and after I set it I never bothered to poke around inside again.
My set of HD800 are plugged in permanently and I had no reason to worry about the off-set damaging them.
Lazy Cat may or may not agree with my explanation but this amp functions in a similar manner. At switch on, regardless of how well things are matched, the bridge will "amplify" every imbalance in component and the changes will remain apparent until all components have settled and stabilized. Which is what Salas mention here. Even tiny imbalances in current in the two legs would be visibly changing continually. I know what was to be expected so it never bothered me and there is no damage done to your speakers. 100 mV offset would be dissipating 80mW in your 8 ohm speaker for a short while. You can listen to music immediately if you like, I doubt if any but the purest of platinum ear guys ever detect this.
That is why I said before this amp should be adjusted when stable, ten minutes later checked and re-adjusted, until you are satisfied with what you see and that will be the amp's stable state.
If you feel it is the audiophile thing to do and wait until your amp is "warmed up" before listening to it, then this is the perfect amp for the audiophile
)One would assume that the negative feedback should correct this phenomena, but it actually could also worsen it. I tend to think an hour is a little over-stated because mine does stabilise within maybe ten or fifteen minutes. Furthermore, mine drifts both positive and negative, pretty wildly at first, but once at operating temperature it is rock solid, well a few mV up and down. But don't fear fellows this is neither serious nor any form of failure.
I hope my explanation is of some help. LC may want to chip in his five pence worth, but let me assure you, if I thought Lazy Cat was bull dusting us I would have accused him as the ultimate sinner early in the thread.
I have been very pleased with the performance of this simple little amp for several weeks now, it remains one of the most satisfying products I have made and with very little effort. It worked first time and after I set it I never bothered to poke around inside again.
My set of HD800 are plugged in permanently and I had no reason to worry about the off-set damaging them.
Salas, I seem never to answer your questions. At first when I have simmed the amp I saw a possible oscillation at around 10MHz or so, but also at the time I was fooling with the gain and the pFs across the FB resistor seemed to solve the problem, but later I noticed that adjusting anything fooled with the bandwidth and the phase margin. When I built it and scoped it there was no oscillation and have since removed the caps altogether, so there is no valid reason for them any longer.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.