I used to totally flummox people back when I was in retail by pointing out that the power supply was an important part of the sound of an amplifier. Just not in the ways they thought.
It's easy to fall into the trap of thinking of it as purely a filtering/storage device for DC, but to look at it as a short to ground for AC (i.e. signal, etc.) headed back upstream struck some people as pretty odd. Both aspects are important.
More important still is how it sounds.
Jocko,
According to many my sanity is a lost cause. You and I are in basic agreement on most things, I think.
Seems ironic to me to see science assume the role of religion. That is, accept without question a given point of view. Anything or anyone who dissents is "obviously" in league with the devil and must be tarred and feathered. Funny...I thought that the fundamental concept of science was to keep an open mind and be ready to accept new possibilities as they opened up. Tis a pity to see folks cloak themselves in all the finery of science and strike a pose, saying that such-and-such is the work of the devil without even trying it. Yes, it takes work. You've got to tear into a functioning circuit, make changes, then sit down and listen. Time consuming, yes. Pain in the butt sometimes, too. Still, it's a remarkably efficient way to separate the men from the boys, assuming that you are willing to educate and then trust your ears.
Grey
It's easy to fall into the trap of thinking of it as purely a filtering/storage device for DC, but to look at it as a short to ground for AC (i.e. signal, etc.) headed back upstream struck some people as pretty odd. Both aspects are important.
More important still is how it sounds.
Jocko,
According to many my sanity is a lost cause. You and I are in basic agreement on most things, I think.
Seems ironic to me to see science assume the role of religion. That is, accept without question a given point of view. Anything or anyone who dissents is "obviously" in league with the devil and must be tarred and feathered. Funny...I thought that the fundamental concept of science was to keep an open mind and be ready to accept new possibilities as they opened up. Tis a pity to see folks cloak themselves in all the finery of science and strike a pose, saying that such-and-such is the work of the devil without even trying it. Yes, it takes work. You've got to tear into a functioning circuit, make changes, then sit down and listen. Time consuming, yes. Pain in the butt sometimes, too. Still, it's a remarkably efficient way to separate the men from the boys, assuming that you are willing to educate and then trust your ears.
Grey
Elso,
what I think he is trying to say is that for all AC analysis for a circuit you treat the power supply just as you treat ground. That is, in simple words that the power supply should have a very low impedance not only for DC but also for AC. Like using ceramic coupling caps.
While this has been known for engineers since the earlier part of last century it is now 'discovered' by the audio folks.
what I think he is trying to say is that for all AC analysis for a circuit you treat the power supply just as you treat ground. That is, in simple words that the power supply should have a very low impedance not only for DC but also for AC. Like using ceramic coupling caps.
While this has been known for engineers since the earlier part of last century it is now 'discovered' by the audio folks.
Well, if you had done any AC analysis of a circuit like a simple CE-stage you would know what I mean...
But anyway, I will try to explain to a chemist like you
, although I miss your muppet avatar picture.
The words impedance and Thèvenin equivalent are pretty much the key words here. Look them up if you don't know what they mean. For low frequencies, any decent power supply will have very low output resistance. But as the frequency rises, the series ESR (=equivalent series resistance) of the filter capacitors rises and thus the impedance of the power supply. Electrolythic capacitors have poor high-frequency ESR and thus you often run them in paralell with ceramic/film capacitors.
If you have a regulated power supply the high-frequency load regulation will be much worse than the low frequency regulation. This is due to a finite gain-bandwidth product of the regulator.
The same thing goes for opamps for example and their power supply rejection ratio (PSSR), that is their sensitivity for noise on the power supply lines. Have a look in any datasheet and see how PSSR drops with frequency.
I can't explain this any simpler than this, maybe someone else can.
Regards
/Magnus
But anyway, I will try to explain to a chemist like you
, although I miss your muppet avatar picture.The words impedance and Thèvenin equivalent are pretty much the key words here. Look them up if you don't know what they mean. For low frequencies, any decent power supply will have very low output resistance. But as the frequency rises, the series ESR (=equivalent series resistance) of the filter capacitors rises and thus the impedance of the power supply. Electrolythic capacitors have poor high-frequency ESR and thus you often run them in paralell with ceramic/film capacitors.
If you have a regulated power supply the high-frequency load regulation will be much worse than the low frequency regulation. This is due to a finite gain-bandwidth product of the regulator.
The same thing goes for opamps for example and their power supply rejection ratio (PSSR), that is their sensitivity for noise on the power supply lines. Have a look in any datasheet and see how PSSR drops with frequency.
I can't explain this any simpler than this, maybe someone else can.
Regards
/Magnus
Well, the size and quality of the transformer mostly has to do with low frequency impedance and level of shielding from mains injected noise and very little with high frequency AC performance.
Actually I don't think you have. Just putting in millions of high-quality capacitors everywhere is NOT going to cure this. Actually it can make things worse! That is because capacitors have a SRF (=self resonant frequency) and particulary high grade plastic film capacitors have a very high resonant Q. Pick the wrong combination and you will have lots of undamped ringing in the power supply. I can not explain this in an easy way because you simply have to know the basic theories behind oscillators and AC circuit theory. But their are excellent application notes on this out there, particulary by LT (Jim Williams of course!), AD and Guido Tent.
You simply have to measure every aspect of the power supply with a good spectrum analyzer and oscilloscope. Listening tests are essentially useless here.
I don't want so sound like a world-class azzhole because I know that you are a very nice person with a passion for DIY audio. But as a trained chemist I think you will understand that there is simply an ocean of difference between the work done by proffessional electrical engineers and that by audio hobbyists.
MRI scanners, SQUID-input stages (yes I am in med. instrumentation!), microwave devices and complex RF links like DS-CDMA are simply light-years ahead of audio.
It's like the difference between me stirring in a cup of tea and the best chemists working on genetically produced substances or the neuro-surgeon doing his stuff - work that really is like magic to me!
So let's celebrate the differencies between people and share our common passion for DIY audio!
/Magnus
Swedish Chef said:
Are you saying that one cannot hear any difference between a good and a flawed power supply? If it is not possible to distinguish them with our ears (at the end, we want to listen music with our gear, isn't it?), why should one take care and spend money and time to design a good power supply?
Well, would be nice to make a time travel to the future: if you are light years ahead, it should be a hands down task for you to give us a schematic for an amp, say, in 2015....
Saluti, Tino
Out of Line
Originally posted by Swedish Chef
http://www.diyaudio.com/forums/showthread.php?postid=496979#post496979
Maybe you can tell me what is going on?
Now what do you propose as a cure?
I am aware of these and studied them all. Actually it was me who posted a link to a Jim Williams article warning that the wrong combination of bypass caps can ring. You probably never heard of the KWAK-CLOCK, which is quite a success in audio circles
Concluding I consider your post out of line, simply trying to impress me with High-Tech and impressive words like Thevenin equivalent, SQUID, MRI, DS-CDMA. microwave devices etc.
Originally posted by Swedish Chef
Actually I do NOT agree. The toroidal transformer has a much higher bandwidth and lets through more hass. There was an Audio amateur article about that. See also:
http://www.diyaudio.com/forums/showthread.php?postid=496979#post496979
Maybe you can tell me what is going on?
Now what do you propose as a cure?
I am aware of these and studied them all. Actually it was me who posted a link to a Jim Williams article warning that the wrong combination of bypass caps can ring. You probably never heard of the KWAK-CLOCK, which is quite a success in audio circles
The listening test is for me the ultimate test as this is the end result that matters. If I have to put in Bluetac or a bundle of straw I would do it. I always check the powersupplylines for oscillation.
Thank you for that.
The problem with professional engineers is they are born without ears. Not their fault. That makes the evaluation by listening pretty hard.
MRI scanners, SQUID-input stages (yes I am in med. instrumentation!), microwave devices and complex RF links like DS-CDMA are simply light-years ahead of audio.
It's like the difference between me stirring in a cup of tea and the best chemists working on genetically produced substances or the neuro-surgeon doing his stuff - work that really is like magic to me!
So let's celebrate the differencies between people and share our common passion for DIY audio!
/Magnus
Concluding I consider your post out of line, simply trying to impress me with High-Tech and impressive words like Thevenin equivalent, SQUID, MRI, DS-CDMA. microwave devices etc.
Elso,
Let's be systematic here...
Yes, that is what I was saying with "level of shielding from mains injected noise". I have not read the article in AA but generally a toroid has much higher primary -> secondary interwinding capacitance than an EI transformer of the same size, thus letting through more garbage.
Any quality transformer will have at least one electrostatic shield between primary and secondary windings to reduce this capacitance, usually to be connected to protective earth.
See "Grounding and Shielding Techniques in Instrumentation" by R. Morrison, ISBN #0-471-83805-5 for a very good discussion on transformers. A stunning interwinding capacitance of 0.05 pF has been obtained with proper shielding techniques. You are not likely to see one of those in audio though.
Well, I am not trying to brag about my golden ears here, simply because they are not golden but very human indeed. I have played in about 8 different orchestras and although I am by no means one myself worked with a few outstanding musicians. And I have used my ears "professionally" for 1.5 years during my duty on a submarine in the Swedish Navy.
The thing with audio is that it is like football - the ones who have all the expert opinions are usually the ones sitting in the TV sofa having never touched a ball. The guys who know how to play keep their mouths shut and concentrate on the game.
It is not my intention to impress you with a lot of technical buzzwords. But for anyone building electronics not knowing what a Thèvenin equivalent is like someone practicing medicine not knowing what the pulmonary artery is. I am not trying to **** you off but it is just as simple as that.
/Magnus
Let's be systematic here...
Yes, that is what I was saying with "level of shielding from mains injected noise". I have not read the article in AA but generally a toroid has much higher primary -> secondary interwinding capacitance than an EI transformer of the same size, thus letting through more garbage.
Any quality transformer will have at least one electrostatic shield between primary and secondary windings to reduce this capacitance, usually to be connected to protective earth.
See "Grounding and Shielding Techniques in Instrumentation" by R. Morrison, ISBN #0-471-83805-5 for a very good discussion on transformers. A stunning interwinding capacitance of 0.05 pF has been obtained with proper shielding techniques. You are not likely to see one of those in audio though.
Well, I am not trying to brag about my golden ears here, simply because they are not golden but very human indeed. I have played in about 8 different orchestras and although I am by no means one myself worked with a few outstanding musicians. And I have used my ears "professionally" for 1.5 years during my duty on a submarine in the Swedish Navy.
The thing with audio is that it is like football - the ones who have all the expert opinions are usually the ones sitting in the TV sofa having never touched a ball. The guys who know how to play keep their mouths shut and concentrate on the game.
It is not my intention to impress you with a lot of technical buzzwords. But for anyone building electronics not knowing what a Thèvenin equivalent is like someone practicing medicine not knowing what the pulmonary artery is. I am not trying to **** you off but it is just as simple as that.
/Magnus
Thanks for the interest... btw, I took the pic of the bear myself.
The power supply does use ~500,000 ufd of capacitance, split of course between a + & - rail. I have been using HEXFRED 400 amp diodes for some time now, and before that Shottkys with snubbers...
As far as current draw from the wall, you get about 300 watts of quiescent current draw, out of which ~90watts is from two channels of "high bias" operation of the 12 MOSFET devices per channel.
The only really large current draw problem is on turn on. For that there is a slow start circuit that limits the draw on turn on when the cap bank has to charge up from 0 volts. Obviously, if you happenend to hit the switch exactly the right point on the AC mains waveform the current would be absolutely astounding, whereas the rest of the time it would just be extremely high. High enough to blow a 20 amp breaker most of the time. So, we don't do that. We use a slow start to charge the cap bank under current limited conditions.
As far as SMPS supplies go, I am skeptical as to their acceptability for this application. Yes, there are amps with SMPS supplies now - I suspect mostly due to the unavailability at any reasonable price of what we formerly knew as "Computer Grade Electrolytics."
Also, please keep in mind that this amp was designed in 1990, which is now almost 14 years ago!
My reservations about SMPS for *this application* goes something like this:
- tough to kill the noise completely
- always tracking the signal, causing lag
- doesn't like to sit around and charge caps
- voltage regulation does not always mean current available at the right time
- not likely to provide current peaks like this supply can...
- peak current has to be drawn on demand from the AC mains all the time (some dependency on the stiffness of the mains)
Ok, having said that, I've been thinking about how to use an SMPS effectively for this application. So, it could happen.
I think that when you spend some time looking at it, you end up thinking Class D amplifier, not power supply. Maybe not.
Oh, almost forgot, the point of the big *** power supply is simple. On complex, forte passages this amplifier "stays together" when listened to on low distortion speakers - even the *same* amplifier without the extra supply *does not*, and most commercial amps (not all) clearly do not. A minor difference perhaps, but this amp is designed to provide performance and not ignore any minor differences. This is audible, and easy enough to discern during listening tests.
The specs are up on the webpages. The circuit is fully complementary, DC coupled as the webpages say. FET input, casoded, current sourced, etc...
Clearly, the Symphony No.1 is an amplifier that is not for everyone, nor necessary for many. You may or may not be able to understand that based upon your personal audiophile experiences or not. Other DIYers may take away some ideas from it that they can use for their own projects.
For the objectivists, this amp measures exceptionally well.
For the subjectivists, this amp makes very little sound of its own, and provides a subjectively open window on the source that comes before it and the speakers that come after it without imparting its own "signature." But that is only the opinion of those who own them or have heard it... :- )
Having said all that, there is no one single perfect amp, and none will be everything for everybody. In the end it comes down to this being a question of *ART* and individual preferences.
_-_-bear
The power supply does use ~500,000 ufd of capacitance, split of course between a + & - rail. I have been using HEXFRED 400 amp diodes for some time now, and before that Shottkys with snubbers...
As far as current draw from the wall, you get about 300 watts of quiescent current draw, out of which ~90watts is from two channels of "high bias" operation of the 12 MOSFET devices per channel.
The only really large current draw problem is on turn on. For that there is a slow start circuit that limits the draw on turn on when the cap bank has to charge up from 0 volts. Obviously, if you happenend to hit the switch exactly the right point on the AC mains waveform the current would be absolutely astounding, whereas the rest of the time it would just be extremely high. High enough to blow a 20 amp breaker most of the time. So, we don't do that. We use a slow start to charge the cap bank under current limited conditions.
As far as SMPS supplies go, I am skeptical as to their acceptability for this application. Yes, there are amps with SMPS supplies now - I suspect mostly due to the unavailability at any reasonable price of what we formerly knew as "Computer Grade Electrolytics."
Also, please keep in mind that this amp was designed in 1990, which is now almost 14 years ago!
My reservations about SMPS for *this application* goes something like this:
- tough to kill the noise completely
- always tracking the signal, causing lag
- doesn't like to sit around and charge caps
- voltage regulation does not always mean current available at the right time
- not likely to provide current peaks like this supply can...
- peak current has to be drawn on demand from the AC mains all the time (some dependency on the stiffness of the mains)
Ok, having said that, I've been thinking about how to use an SMPS effectively for this application. So, it could happen.
I think that when you spend some time looking at it, you end up thinking Class D amplifier, not power supply. Maybe not.
Oh, almost forgot, the point of the big *** power supply is simple. On complex, forte passages this amplifier "stays together" when listened to on low distortion speakers - even the *same* amplifier without the extra supply *does not*, and most commercial amps (not all) clearly do not. A minor difference perhaps, but this amp is designed to provide performance and not ignore any minor differences. This is audible, and easy enough to discern during listening tests.
The specs are up on the webpages. The circuit is fully complementary, DC coupled as the webpages say. FET input, casoded, current sourced, etc...
Clearly, the Symphony No.1 is an amplifier that is not for everyone, nor necessary for many. You may or may not be able to understand that based upon your personal audiophile experiences or not. Other DIYers may take away some ideas from it that they can use for their own projects.
For the objectivists, this amp measures exceptionally well.
For the subjectivists, this amp makes very little sound of its own, and provides a subjectively open window on the source that comes before it and the speakers that come after it without imparting its own "signature." But that is only the opinion of those who own them or have heard it... :- )
Having said all that, there is no one single perfect amp, and none will be everything for everybody. In the end it comes down to this being a question of *ART* and individual preferences.
_-_-bear
A simplistic, and unfancy way of explaining the relationship between the power supply and the amplifier and what comes out:
If you consider an output stage (ignoring the effect of feedback or balanced output stages - for simplicity) there is a device that is doing the controlling of the power. It usually has three main conditions: quiescent, off and on... obviously being modulated "on and off" with the signal.
If we look at the *on* state we find a device that is conducting strongly, and has a close to *zero* impedance (when turned on the to maximum). Which we could approximate with a bit of wire or a very low value resistor (for the instant that it is *on* that much) for that instant.
Now, in that *on* state, it is very close to having the *power supply* merely connected directly to the load! So whatever appears on that rail of power supply in effect appears at the load.
What happens in practice in solid state PP amps, or any amp with feedback is that the power supply ripple and noise is reduced by said feedback. Usually this parameter is called PSRR by the opamp people. Power Supply Rejection Ratio... but in practice the higher the power output and the greater the peak current drawn from *any supply* the *worse* the supply looks as far as ripple.
(For opamp supplies that are regulated, this doesn't exactly apply since the opamp's draw is always way under the supplies ability to provide clean current at a given voltage... this is the reason that some amp designers have tried to build power amps with regulated supplies!)
So, the harder you drive your amp, the *worse* the ripple rejection becomes in effect (assuming it is good to begin with) if for no other reason than the ripple is increased with power drawn!
Which in part is why some amps that sound pretty good at low or moderate levels appear to get "muddled" with complex passages at high levels... there are other mechanisms at play here, but this is one of them.
A simple explanation that I hope illuminates the concept somewhat...
_-_-bear
If you consider an output stage (ignoring the effect of feedback or balanced output stages - for simplicity) there is a device that is doing the controlling of the power. It usually has three main conditions: quiescent, off and on... obviously being modulated "on and off" with the signal.
If we look at the *on* state we find a device that is conducting strongly, and has a close to *zero* impedance (when turned on the to maximum). Which we could approximate with a bit of wire or a very low value resistor (for the instant that it is *on* that much) for that instant.
Now, in that *on* state, it is very close to having the *power supply* merely connected directly to the load!
So whatever appears on that rail of power supply in effect appears at the load.What happens in practice in solid state PP amps, or any amp with feedback is that the power supply ripple and noise is reduced by said feedback. Usually this parameter is called PSRR by the opamp people. Power Supply Rejection Ratio... but in practice the higher the power output and the greater the peak current drawn from *any supply* the *worse* the supply looks as far as ripple.
(For opamp supplies that are regulated, this doesn't exactly apply since the opamp's draw is always way under the supplies ability to provide clean current at a given voltage... this is the reason that some amp designers have tried to build power amps with regulated supplies!)
So, the harder you drive your amp, the *worse* the ripple rejection becomes in effect (assuming it is good to begin with) if for no other reason than the ripple is increased with power drawn!
Which in part is why some amps that sound pretty good at low or moderate levels appear to get "muddled" with complex passages at high levels... there are other mechanisms at play here, but this is one of them.
A simple explanation that I hope illuminates the concept somewhat...
_-_-bear
mikeks said:Bear, i am not sure about your heatsink provision though...
Little point methinks, in providing a vast supply reservoir, and output devices to boot if your thermal management hardware will not permit these to be exploited to the full....
The heatsinks are quite adequate.
Designed to be so.
They run slightly warm, not more. :- )
Efficient and larger than you apparently seem to think from the pictures! The amp weighs in at 128lbs, btw...
_-_-bear
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.