Fast diodes have much sharper transition between conducting and non-conducting state (end of half-sine transition to zero line), i.e. the slope (derivative) changes suddenly when they switch off. This generates high-frequency spectrum with compounds of higher order than that for slow diodes. If this is preferable, then use fast diodes.
What current do you wish to run through the shunt element?
A minimum suggested is 1.5A, same as the device under power.
This means 60W dissipated in the shunt element, and another 60W in the Class A amplifier. 240W total for two channels is pretty high......
What are the advantages of a shunt regulator? Do you believe it will have less ripple?
Best device to use is a p type semiconductor such as an IRF9140, rated at 125W 19A. I suggest the source on the rail, the drain at ground, with the gate driven from an opamp. You will need a large series resistor between supply and rail, and it should drop at least 25% of the supplied voltage, viz 10V. At 3A, this will dissipate a further 30W; your amp and power supply is beginning to look spectacularly inefficient - and for what purpose? A shunt supply won't sound any better in this application than a nicely designed CLC........
Cheers,
Hugh
A minimum suggested is 1.5A, same as the device under power.
This means 60W dissipated in the shunt element, and another 60W in the Class A amplifier. 240W total for two channels is pretty high......
What are the advantages of a shunt regulator? Do you believe it will have less ripple?
Best device to use is a p type semiconductor such as an IRF9140, rated at 125W 19A. I suggest the source on the rail, the drain at ground, with the gate driven from an opamp. You will need a large series resistor between supply and rail, and it should drop at least 25% of the supplied voltage, viz 10V. At 3A, this will dissipate a further 30W; your amp and power supply is beginning to look spectacularly inefficient - and for what purpose? A shunt supply won't sound any better in this application than a nicely designed CLC........
Cheers,
Hugh
hmm...
so in PSUD what value do I use for the ohms of the capacitor?? Is this the ESR of the cap?
Finally... where can I get my hands on 2 20mH chokes that will handle 1.5A each.
which bridge rectifier shall I use?
My caps say:
U32D 105C 0022L 83
50v 68000uF
what power transformer do I need for 40vdc at output??
Thanks!!!
so in PSUD what value do I use for the ohms of the capacitor?? Is this the ESR of the cap?
Finally... where can I get my hands on 2 20mH chokes that will handle 1.5A each.
which bridge rectifier shall I use?
My caps say:
U32D 105C 0022L 83
50v 68000uF
what power transformer do I need for 40vdc at output??
Thanks!!!
No ....... you need to do a little research
No, the recovery characteristics of fast soft recovery diodes gernerate less RFI. That's the whole rational for using them is to minimize the RF generated during the reverse recovery conduction that gets in the amplifier through the supplies. The magnitude of the reverse recovery current and softness of the recovery slope are parameters to look at. Much more emphasis on the soft recovery characteristics has resulted in much less RFI than in the older fast diode types like the Internation Rectifiers Hexfreds. This is pretty old news for high end audio design and was known over fifteen years ago. Do a search on the DIY audio forum, this topic has been discussed in some detail here. Do a little research before snapping off with sounded to me to be a sarcastic reply. BTW I did EMI reduction design and high speed protection circuit design when I worked for Alcatel and have few years of experience with this subject.
I'll give you a head start in fact:
diyAudio Forums > Top > Amplifiers > Solid State > High Speed Diodes
PMA said:
No, the recovery characteristics of fast soft recovery diodes gernerate less RFI. That's the whole rational for using them is to minimize the RF generated during the reverse recovery conduction that gets in the amplifier through the supplies. The magnitude of the reverse recovery current and softness of the recovery slope are parameters to look at. Much more emphasis on the soft recovery characteristics has resulted in much less RFI than in the older fast diode types like the Internation Rectifiers Hexfreds. This is pretty old news for high end audio design and was known over fifteen years ago. Do a search on the DIY audio forum, this topic has been discussed in some detail here. Do a little research before snapping off with sounded to me to be a sarcastic reply. BTW I did EMI reduction design and high speed protection circuit design when I worked for Alcatel and have few years of experience with this subject.
I'll give you a head start in fact:
diyAudio Forums > Top > Amplifiers > Solid State > High Speed Diodes
Hi,
Precisely...
While I'm certainly no expert on shunt regs, the darned things are so hopelessly inefficient no one seems to want them.
A discrete series reg. can be configured in such a way that most of the input ripple is cancelled out just the same.
What you do is inject part of the ripple into the error amp, it should inverse polarity, and the output is superimposed on the output of the regs series device where it's essentially cancelled out due to opposing phase.
At least that's what I often do wit tubed regulators anyway.
If it can be done with tubes, it sure can be done with sand I'd reckon...
Yeah...that was one of the better threads we had on the forum, IMHO.
In a nutshell, it's not the "fast" attribute of the diode that matters but the "soft" part.
After all, how "fast" does a 60/50 Hz rectumfrier have to be ?
Cheers,
Precisely...
While I'm certainly no expert on shunt regs, the darned things are so hopelessly inefficient no one seems to want them.
A discrete series reg. can be configured in such a way that most of the input ripple is cancelled out just the same.
What you do is inject part of the ripple into the error amp, it should inverse polarity, and the output is superimposed on the output of the regs series device where it's essentially cancelled out due to opposing phase.
At least that's what I often do wit tubed regulators anyway.
If it can be done with tubes, it sure can be done with sand I'd reckon...
Yeah...that was one of the better threads we had on the forum, IMHO.
In a nutshell, it's not the "fast" attribute of the diode that matters but the "soft" part.
After all, how "fast" does a 60/50 Hz rectumfrier have to be ?
Cheers,
Re: hmm...
Whoa! Do you want us to build the thing for you?
If you don't know what voltage power transformer you need
I suggest reading some basic electronics books before you
even get started. You will be dealing with mains voltages
so you need to have a good basic knowledge of electrical
design.
WRT chokes, most transformer winders will make these no
problem.
Cheers,
Terry
AudioGeek said:
Whoa! Do you want us to build the thing for you?
If you don't know what voltage power transformer you need
I suggest reading some basic electronics books before you
even get started. You will be dealing with mains voltages
so you need to have a good basic knowledge of electrical
design.
WRT chokes, most transformer winders will make these no
problem.
Cheers,
Terry
AudioGeek,
Terry is probably right; you need to read a few basic books on power supply, if only to avoid getting into serious trouble with mains voltages.
As an aside, you are dealing with the world of physics, and the most important aspect of this world is the word 'compromise'. Nothing is for nothing, more of this invariably means less of that, and the art of design often becomes the management of compromise. This is never more so than in audio, which is further complicated by psycho-acoustic issues - human aural perception - which lurks in the background of all designs, particularly the greats in this noble pursuit, like Hiraga, JLH, NP, to name three.
The ESR of the filter cap is exactly what you should be using in your PSUD program; but it is definitely not as low as the capacitance would have you believe. At 120Hz, a 20mH choke will have an impedance of 15 ohms, and a 68,000uF will be around 0.04 ohms. If we assume around 1Vpp of ripple at the first cap (a most pessimistic assessment, incidentally), then by the second cap, at the PS output, this will have reduced to 3.3mV. This is probably close enough for government work....
I should add that I have nothing against shunt regulation at low power and often use it with receiver tubes. I find they sound better in this application, and the best combination I've found is a SS shunt regulator and a tube as shunt element, as used in the Supratek preamplifiers.
Pavel/Pavel, I was not aware of Malichi's information on these UFSR diodes. On a suggestion, I tried them some years back, and the results were definitive in a subjective sense. This was enough for me, as I go entirely on the psycho-acoustics. The engineering is clearly very important, and it's always part and parcel of the design process, but it is the psycho-acoustics which actually sells the product. My chief concern as an amplifier seller is 'How does it sound?' This simple question causes an enigmatic mix of challenge, wonderment and contempt in any engineer, and I can see why. But I am not selling to engineers, as they build their own amps, and very good they are too. I sell to hobbyists, people who primarily love their music and have some technical skills, but who are not usually engineers. Most have a tertiary education, and are very smart, so it's almost impossible to blow smoke up their !@#$, if you get my drift......
In the space of about ten posts, I think the personality of each participant has emerged, very interesting. My attitude these days (and purely because I don't enjoy the brain damage of conflict), is that I have an opinion which I'm happy to present if asked, but which I prefer not to defend in any committed way. Time is short; this is not religion, there is no need for a jihad, and if my opinion is wrong, no problem. I have built all this stuff and done a lot with power supplies, and enjoyed the results after years of trial and error. I would rather not focus on why, or justify my choice with math, because I have no choice but to get on with the serious business of making a living in this tough field. Only my own conscience and sales are strictly significant, and I try to balance the two transparently.
In closing, I'll give a salutory example. For years I resisted buying Black Gates because I felt they were absurdly expensive. This stuff was snake oil, and totally unjustified. Until I was forced, for commercial reasons, into producing upgrades for my two power amplifiers, that is.......
Now, this week in fact, I will move a ridiculous amount of money to Japan for a very large order of Black Gates. It seems last orders should be made by September, as sadly they will stop production. It seems to me - and I'm only one opinion of course - that there is just nothing comes close in large capacitance ranges. The NX series in particular is so good that it clearly beats most film caps, and in truth is much cheaper bang for buck than the usual staples, like the RelCaps. I think these caps are so good for signal path work I can think of nothing better...... and I have no association with Jelmax or Rubycon, BTW.
I will eventually try the Tantalum resistors from Japan. Perhaps I will see more light on the road to Damascus, perhaps not. But I must try them before I condemn them.
Cheers,
Hugh
Terry is probably right; you need to read a few basic books on power supply, if only to avoid getting into serious trouble with mains voltages.
As an aside, you are dealing with the world of physics, and the most important aspect of this world is the word 'compromise'. Nothing is for nothing, more of this invariably means less of that, and the art of design often becomes the management of compromise. This is never more so than in audio, which is further complicated by psycho-acoustic issues - human aural perception - which lurks in the background of all designs, particularly the greats in this noble pursuit, like Hiraga, JLH, NP, to name three.
The ESR of the filter cap is exactly what you should be using in your PSUD program; but it is definitely not as low as the capacitance would have you believe. At 120Hz, a 20mH choke will have an impedance of 15 ohms, and a 68,000uF will be around 0.04 ohms. If we assume around 1Vpp of ripple at the first cap (a most pessimistic assessment, incidentally), then by the second cap, at the PS output, this will have reduced to 3.3mV. This is probably close enough for government work....
I should add that I have nothing against shunt regulation at low power and often use it with receiver tubes. I find they sound better in this application, and the best combination I've found is a SS shunt regulator and a tube as shunt element, as used in the Supratek preamplifiers.
Pavel/Pavel, I was not aware of Malichi's information on these UFSR diodes. On a suggestion, I tried them some years back, and the results were definitive in a subjective sense. This was enough for me, as I go entirely on the psycho-acoustics. The engineering is clearly very important, and it's always part and parcel of the design process, but it is the psycho-acoustics which actually sells the product. My chief concern as an amplifier seller is 'How does it sound?' This simple question causes an enigmatic mix of challenge, wonderment and contempt in any engineer, and I can see why. But I am not selling to engineers, as they build their own amps, and very good they are too. I sell to hobbyists, people who primarily love their music and have some technical skills, but who are not usually engineers. Most have a tertiary education, and are very smart, so it's almost impossible to blow smoke up their !@#$, if you get my drift......
In the space of about ten posts, I think the personality of each participant has emerged, very interesting. My attitude these days (and purely because I don't enjoy the brain damage of conflict), is that I have an opinion which I'm happy to present if asked, but which I prefer not to defend in any committed way. Time is short; this is not religion, there is no need for a jihad, and if my opinion is wrong, no problem. I have built all this stuff and done a lot with power supplies, and enjoyed the results after years of trial and error. I would rather not focus on why, or justify my choice with math, because I have no choice but to get on with the serious business of making a living in this tough field. Only my own conscience and sales are strictly significant, and I try to balance the two transparently.
In closing, I'll give a salutory example. For years I resisted buying Black Gates because I felt they were absurdly expensive. This stuff was snake oil, and totally unjustified. Until I was forced, for commercial reasons, into producing upgrades for my two power amplifiers, that is.......
Now, this week in fact, I will move a ridiculous amount of money to Japan for a very large order of Black Gates. It seems last orders should be made by September, as sadly they will stop production. It seems to me - and I'm only one opinion of course - that there is just nothing comes close in large capacitance ranges. The NX series in particular is so good that it clearly beats most film caps, and in truth is much cheaper bang for buck than the usual staples, like the RelCaps. I think these caps are so good for signal path work I can think of nothing better...... and I have no association with Jelmax or Rubycon, BTW.
I will eventually try the Tantalum resistors from Japan. Perhaps I will see more light on the road to Damascus, perhaps not. But I must try them before I condemn them.
Cheers,
Hugh
How about Half Fast diodes
I'm confused as to what type of fast diode you are reffering to. Many people are having success with Shottky diodes (both with and without snubber networks) as well. The origional FREDs (Fast Recovery Epitaxial Diodes) do have a soft recovery compared with the diodes in a conventional bridge rectifier. They are a great improvement but some of the new Fast Recovery diodes have much softer recovery characteristics and better sonics than the early Hexfreds. I like the IXYS and the Fairchild Stealth fast recovery diodes and they seem to be pretty easy to find.
I'm confused as to what type of fast diode you are reffering to. Many people are having success with Shottky diodes (both with and without snubber networks) as well. The origional FREDs (Fast Recovery Epitaxial Diodes) do have a soft recovery compared with the diodes in a conventional bridge rectifier. They are a great improvement but some of the new Fast Recovery diodes have much softer recovery characteristics and better sonics than the early Hexfreds. I like the IXYS and the Fairchild Stealth fast recovery diodes and they seem to be pretty easy to find.
my concern
my concern was i didn't want to spend $$$$$$$ on a power transformer and chokes and to end up with a voltage of less than 40vdc at 1.5 current draw.
OK... so I will use my caps...
time to find some 20mH chokes....
any good sources for these???
would a 50v transformer work? what VA rating?
Thanks!!!!
my concern was i didn't want to spend $$$$$$$ on a power transformer and chokes and to end up with a voltage of less than 40vdc at 1.5 current draw.
OK... so I will use my caps...
time to find some 20mH chokes....
any good sources for these???
would a 50v transformer work? what VA rating?
Thanks!!!!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.