One thing that's not being said, is the importance of power supply quality on the sound is variable depending on the circuit used. PSRR figures in highly to the sound quality vs. power supply quality debate.
For example look at the gainclones, people are building 50 watt power amps with 1000 uF filter caps. they can get away witht his because of the high PSRR of those chips.
I would argue that if you have a 90 dB PSRR, then any regulated power supply should sound the same. However many circuits (even op-amps and those power gaincard type chips) don't have that high PSRR at all audio frequencies.
Just my view from the peanut gallery...
Sheldon
For example look at the gainclones, people are building 50 watt power amps with 1000 uF filter caps. they can get away witht his because of the high PSRR of those chips.
I would argue that if you have a 90 dB PSRR, then any regulated power supply should sound the same. However many circuits (even op-amps and those power gaincard type chips) don't have that high PSRR at all audio frequencies.
Just my view from the peanut gallery...
Sheldon
I'd be inclined to agree, in principle.
The problem is I've never seen a circuit with that PSRR across the audio band, taking into account circuit noise gain etc.
Peranders,
You're right, but my point is any competent amp circuit (not obviously flawed designs) can be made to sound good, if you address the details.
Topology is almost irrelevant, but PSU's bring big benefits to everything I've tried.
Andy.
PS quality
Given the above comments on power supply quality and regulators, what regulator specifications and ps "quality" are sufficient? Meaning in the type of small, low voltage digital and +/- analog supplies controlled with 3-pin regulators, and used in DACs and related digital/analog audio designs, as opposed to power amps.
It is one thing to say that a component is not good enough, or that another is "better", but what noise and regulation specs are actually required for a supply which will not add noise and distortion to the output?
Given the above comments on power supply quality and regulators, what regulator specifications and ps "quality" are sufficient? Meaning in the type of small, low voltage digital and +/- analog supplies controlled with 3-pin regulators, and used in DACs and related digital/analog audio designs, as opposed to power amps.
It is one thing to say that a component is not good enough, or that another is "better", but what noise and regulation specs are actually required for a supply which will not add noise and distortion to the output?
have you flipped
"(You can not flipover these IC type references like the LM329 for the negative supply)."
Actually you could run a positive series reference for a negative reference voltage. Tie Vout to gnd and power Vin from the positive supply. You put a current on the ground terminal and V-out (the ground terminal) is the negative output reference. You are running the regulator series and shunt. Vin could be the regulated positive supply for best effective PSRR for the negative reference. Why bother? There seem to be more good series references than shunt. I seem to remember a few that can be used in either mode.
With seperate transformer secondaries you can tie two positive regulators together, for one to act as a negative regulator (V out of one tied to ground of the other) since the input voltages are isolated by the seperate transformer secondaries not be tied together the GND / Vout node is the only galvanic connection.
I have built a lot of line level circuits with ground as the most positive voltage. Scott Nixon used to joke that he had to turn the schematics upside down to figure out how they work. The circuits had ground for the supply and the load in parallel. Ground looks like the positive supply and the gound return. It is a shunt amplifier much the same priciple as a shunt regulator.
You guys in Australia swap all positive and negative descriptions
since you guys are upside down you realize........
"(You can not flipover these IC type references like the LM329 for the negative supply)."
Actually you could run a positive series reference for a negative reference voltage. Tie Vout to gnd and power Vin from the positive supply. You put a current on the ground terminal and V-out (the ground terminal) is the negative output reference. You are running the regulator series and shunt. Vin could be the regulated positive supply for best effective PSRR for the negative reference. Why bother? There seem to be more good series references than shunt. I seem to remember a few that can be used in either mode.
With seperate transformer secondaries you can tie two positive regulators together, for one to act as a negative regulator (V out of one tied to ground of the other) since the input voltages are isolated by the seperate transformer secondaries not be tied together the GND / Vout node is the only galvanic connection.
I have built a lot of line level circuits with ground as the most positive voltage. Scott Nixon used to joke that he had to turn the schematics upside down to figure out how they work. The circuits had ground for the supply and the load in parallel. Ground looks like the positive supply and the gound return. It is a shunt amplifier much the same priciple as a shunt regulator.
You guys in Australia swap all positive and negative descriptions
since you guys are upside down you realize........
The answer is 42
"It is one thing to say that a component is not good enough, or that another is "better", but what noise and regulation specs are actually required for a supply which will not add noise and distortion to the output?"
You have heard the joke about the guy who ask how much something cost and the salesman saying how much you got?
It dependent on how clean your input voltage to the regulator is,
the PSRR and load currents of the circuit whos supply teminals are being regulated, and whether the circuit is anglog, digital, or vegtable. Noise, PSRR, and output impedance of the regulator vary dramatically with frequency.
And I'll bet you were afraid you would get a complicated answer like "It depends."
"It is one thing to say that a component is not good enough, or that another is "better", but what noise and regulation specs are actually required for a supply which will not add noise and distortion to the output?"
You have heard the joke about the guy who ask how much something cost and the salesman saying how much you got?
It dependent on how clean your input voltage to the regulator is,
the PSRR and load currents of the circuit whos supply teminals are being regulated, and whether the circuit is anglog, digital, or vegtable. Noise, PSRR, and output impedance of the regulator vary dramatically with frequency.
And I'll bet you were afraid you would get a complicated answer like "It depends."
The answer maybe 42
...but have you asked the right question
Rather than "what regulator specifications and ps "quality" are sufficient" how about "what regulator specifications and ps "quality" are relevant"
Aren't we a helpful bunch?
Bit like the guy asking the Irishman for directions "Well I wouldn't start from here, if I were you"
Andy (with a degree of fun).
...but have you asked the right question
Rather than "what regulator specifications and ps "quality" are sufficient" how about "what regulator specifications and ps "quality" are relevant"
Aren't we a helpful bunch?
Bit like the guy asking the Irishman for directions "Well I wouldn't start from here, if I were you"
Andy (with a degree of fun).
Re: have you flipped
Yes Fred you are right. But it were you own words that it was more complicated, you remember? But actually I thought the 8-pin references are so expensive I was better out using one for all regulators.
To pick up the old thread I use the low-pass filter circuit as in the OP176 datasheet and the negative reference i.e. -5V is obtained by a inverting opamp as in the Ref02 datasheet. 10k resistors and 5k to ground. This works well but not much better than LM329's.
Everyone:
How about a great supply + a great amplifier circuit?
The Kevin Gilmore circuit contains more "subtilities" but I leave these to discover by the sharpeyed reader themselve. In other words no more hints. This sentence courtesy of Jocko!
Fred Dieckmann said:
Yes Fred you are right. But it were you own words that it was more complicated, you remember? But actually I thought the 8-pin references are so expensive I was better out using one for all regulators.
To pick up the old thread I use the low-pass filter circuit as in the OP176 datasheet and the negative reference i.e. -5V is obtained by a inverting opamp as in the Ref02 datasheet. 10k resistors and 5k to ground. This works well but not much better than LM329's.
Everyone:
How about a great supply + a great amplifier circuit?
The Kevin Gilmore circuit contains more "subtilities" but I leave these to discover by the sharpeyed reader themselve. In other words no more hints. This sentence courtesy of Jocko!
3-Pin Reg replacements
Koinichiwa,
Now I'll probably be appearing "nasty", but I fail generally to see the attraction of the Jung Reg. It is marginally stable, overly complex and has at best a moderate behaviour when faced with transients.
I would like to first point out that the "sound" of a PSU is very rarely in the DC behaviour of the supply OR in the noise. I find the "sound" of a PSU to be almost entierly determined by it's AC behaviour.
If you use 78XX/79XX regs and their fairly high noise and poor ripple rejection is no issue (it would be for a clock oscillator or a DAC chip with very poor PSRR, in most other circuits it is not) you can simply stick a really large value low impedance electrolytic on their output and I find most 78XX/79XX (in fact all I tried) to sound very similar with a few 1,000uF after the Reg. Most PCB's have enough space.
If you want something simple that actually dominates the AC impedance of the supply you could use for currents up to around 100mA the LM6171 as simple follower with a suitable reference on the input. This way the regulator will not only source current, but if required it will also sink it.
The EIN of the LM6171 is around 4uV over 100KHz bandwidth, the PSRR is > 80db below 10KHz. By adding a suitable Transistor as "Gyrator" in the positive or negative supply line the PSU line rejection can be increased far past the 80db possible with the OPA alone, I'd think 100db+ of ripple rejection should be possible in this case and the dropout Voltage would be minimised, this would likely approach the ripple rejection of something like the Jung Reg at much less effort.
The resulting "regulator" will be mostly limied by the Noise from reference for noise and the "speed" of the regulator will be very high.
With a Reference, J-Fet CCS to supply the reference, a suitable RC circuit to kill noise on the reference, two resistors to set the output voltage and one capacitor to keep AC gain at unity plus one transistor, one resistor and one capacitor we have a circuit which, relised mostly in SMD can rival the little regs from Audiocom for size (and at likley better performance).
Total part count:
1 DIL-8 IC
4 Resistor (SMD)
1 J-Fet CCS (SMD)
1 Reference (SMD)
1 200mA BJT (SMD) [darlington would be good]
3 Capacitors - several uF (Os-Con, BG whatever)
I'd suggest a layout that would place the DIL 8 IC such that it can be fitted against normal heatsinks, with the reference circutry above it.
For higher dissipation/current there are TO220 encased Power Op-Amp's which could be placed inside the feedback loop of another Op-Amp to work as "power" output stage with the first Op-Amp dominating the AC behaviour.
Anyway, some thoughts on the "78XX/79XX replacement" project. By cutting out the series pass transistor the whole regualtor becomes much more capable within the dissipation limit of the Op-Amp.
Sayonara
Koinichiwa,
Now I'll probably be appearing "nasty", but I fail generally to see the attraction of the Jung Reg. It is marginally stable, overly complex and has at best a moderate behaviour when faced with transients.
I would like to first point out that the "sound" of a PSU is very rarely in the DC behaviour of the supply OR in the noise. I find the "sound" of a PSU to be almost entierly determined by it's AC behaviour.
If you use 78XX/79XX regs and their fairly high noise and poor ripple rejection is no issue (it would be for a clock oscillator or a DAC chip with very poor PSRR, in most other circuits it is not) you can simply stick a really large value low impedance electrolytic on their output and I find most 78XX/79XX (in fact all I tried) to sound very similar with a few 1,000uF after the Reg. Most PCB's have enough space.
If you want something simple that actually dominates the AC impedance of the supply you could use for currents up to around 100mA the LM6171 as simple follower with a suitable reference on the input. This way the regulator will not only source current, but if required it will also sink it.
The EIN of the LM6171 is around 4uV over 100KHz bandwidth, the PSRR is > 80db below 10KHz. By adding a suitable Transistor as "Gyrator" in the positive or negative supply line the PSU line rejection can be increased far past the 80db possible with the OPA alone, I'd think 100db+ of ripple rejection should be possible in this case and the dropout Voltage would be minimised, this would likely approach the ripple rejection of something like the Jung Reg at much less effort.
The resulting "regulator" will be mostly limied by the Noise from reference for noise and the "speed" of the regulator will be very high.
With a Reference, J-Fet CCS to supply the reference, a suitable RC circuit to kill noise on the reference, two resistors to set the output voltage and one capacitor to keep AC gain at unity plus one transistor, one resistor and one capacitor we have a circuit which, relised mostly in SMD can rival the little regs from Audiocom for size (and at likley better performance).
Total part count:
1 DIL-8 IC
4 Resistor (SMD)
1 J-Fet CCS (SMD)
1 Reference (SMD)
1 200mA BJT (SMD) [darlington would be good]
3 Capacitors - several uF (Os-Con, BG whatever)
I'd suggest a layout that would place the DIL 8 IC such that it can be fitted against normal heatsinks, with the reference circutry above it.
For higher dissipation/current there are TO220 encased Power Op-Amp's which could be placed inside the feedback loop of another Op-Amp to work as "power" output stage with the first Op-Amp dominating the AC behaviour.
Anyway, some thoughts on the "78XX/79XX replacement" project. By cutting out the series pass transistor the whole regualtor becomes much more capable within the dissipation limit of the Op-Amp.
Sayonara
Re: The answer maybe 42
Hi Andy,
The answer is extremely simple:
I build these regulators for AUDIO so the SOUND quality is what only matters to me.
I could build a sofisticated circuit wit 100 parallelled transistors kept at liquid helium temperature and claim some extreme noise spec. but will this approach sound better??? I guess not!
I try all kinds of circuits and try to learn from the practical results rather than copying a Jung regulator. I realise I have come a long way but will NEVER reach my goal! That's life!
ALW said:...but have you asked the right question
Rather than "what regulator specifications and ps "quality" are sufficient" how about "what regulator specifications and ps "quality" are relevant"
Aren't we a helpful bunch?
Bit like the guy asking the Irishman for directions "Well I wouldn't start from here, if I were you"
Andy (with a degree of fun).
Hi Andy,
The answer is extremely simple:
I build these regulators for AUDIO so the SOUND quality is what only matters to me.
I could build a sofisticated circuit wit 100 parallelled transistors kept at liquid helium temperature and claim some extreme noise spec. but will this approach sound better??? I guess not!
I try all kinds of circuits and try to learn from the practical results rather than copying a Jung regulator. I realise I have come a long way but will NEVER reach my goal! That's life!
Re: Re: The answer maybe 42
I build these regulators for SOUND quality is what only matters to me.
---------------------------------------------------------
Problem is: what is good sound quality; it's system, room and hearing dependent. There is a well known loudspeaker reviwer who has 3-4 cm of hair covering his ears. No wonder he likes Naim equipment!
In another Forum, there are people who claim that their clock is the only and that that those who don't use it are 'idiots' and years behind time. Again, the only evidence is that it sounds best to them.
How does one adjudicate?
I build these regulators for SOUND quality is what only matters to me.
---------------------------------------------------------
Problem is: what is good sound quality; it's system, room and hearing dependent. There is a well known loudspeaker reviwer who has 3-4 cm of hair covering his ears. No wonder he likes Naim equipment!
In another Forum, there are people who claim that their clock is the only and that that those who don't use it are 'idiots' and years behind time. Again, the only evidence is that it sounds best to them.
How does one adjudicate?
Re: Re: Re: The answer maybe 42
Yes that's the problem, as said in a application note by Burr-Brown: "soundquality is subjective"
(can not be measured by objective criteria)
I have to live by that too!
No problem for me!
BTW I have very short hair now but very long hair 25 years ago. Does it really matter?
Fmak If you like the KWAK-CLOCK or not does not bother me at all just as if you think I am a quirk or a quack.......
Hi Fred Mak,fmak said:
Yes that's the problem, as said in a application note by Burr-Brown: "soundquality is subjective"
(can not be measured by objective criteria)
I have to live by that too!
No problem for me!
BTW I have very short hair now but very long hair 25 years ago. Does it really matter?
Fmak If you like the KWAK-CLOCK or not does not bother me at all just as if you think I am a quirk or a quack.......
Regulators for poets
"If you use 78XX/79XX regs and their fairly high noise and poor ripple rejection is no issue (it would be for a clock oscillator or a DAC chip with very poor PSRR, in most other circuits it is not) you can simply stick a really large value low impedance electrolytic on their output and I find most 78XX/79XX (in fact all I tried) to sound very similar with a few 1,000uF after the Reg. Most PCB's have enough space."
This whole post is the funniest thing I have read this week.......
Did you know audio supply design was so simple Andy? Boy have we wasted some time. I am so embarrased.
"I build these regulators for SOUND quality is what only matters to me."
No problem with that but some of kind of want to know why circuits sound different and how to design for that goal. Good engineering does not often negate sound quality. Some people absolutey refuse to believe this. They usually seem to be people that are not engineers for some mysterious reason.
"If you use 78XX/79XX regs and their fairly high noise and poor ripple rejection is no issue (it would be for a clock oscillator or a DAC chip with very poor PSRR, in most other circuits it is not) you can simply stick a really large value low impedance electrolytic on their output and I find most 78XX/79XX (in fact all I tried) to sound very similar with a few 1,000uF after the Reg. Most PCB's have enough space."
This whole post is the funniest thing I have read this week.......
Did you know audio supply design was so simple Andy? Boy have we wasted some time. I am so embarrased.
"I build these regulators for SOUND quality is what only matters to me."
No problem with that but some of kind of want to know why circuits sound different and how to design for that goal. Good engineering does not often negate sound quality. Some people absolutey refuse to believe this. They usually seem to be people that are not engineers for some mysterious reason.
Linear regulators (for non-edible circuits)
I realize that asking that question in this forum is only a little less dangerous than if I were to bring up 5-pin regulators.
Fred: Yes, I have heard the salesman joke. Have you ever heard that to a man with a good hammer every problem looks like a nail? The answer to that one is that if you are the type of person who routinely orders a bottle of wine which costs more than the meal, I probably will not like your advice.
ALW: As for the answer being 42, I would hope that this group had not yet reached the stage where the only interesting question is, "where do we go for lunch?"
My question was about linear supplies using a 3-pin regulator, or a reasonably facsimily thereof, with a pass transistor and a reference, which have to cope with ripple as well as some remnants of switching noise, and do not have to source more than 1 amp or more than 12V.
It is not so much which part is a few uV better, or how many voltage references will fit on the head of a pin, but what is the goal? What kind of output does a linear supply used in a digital audio design need to avoid degrading what should be a good analog representation of the digitized signal.
It's "only" audio, right?
So happy to be the fun for the day Fred, even more honored to be part of the entertainment of the week.
I realize that asking that question in this forum is only a little less dangerous than if I were to bring up 5-pin regulators.
Fred: Yes, I have heard the salesman joke. Have you ever heard that to a man with a good hammer every problem looks like a nail? The answer to that one is that if you are the type of person who routinely orders a bottle of wine which costs more than the meal, I probably will not like your advice.
ALW: As for the answer being 42, I would hope that this group had not yet reached the stage where the only interesting question is, "where do we go for lunch?"
My question was about linear supplies using a 3-pin regulator, or a reasonably facsimily thereof, with a pass transistor and a reference, which have to cope with ripple as well as some remnants of switching noise, and do not have to source more than 1 amp or more than 12V.
It is not so much which part is a few uV better, or how many voltage references will fit on the head of a pin, but what is the goal? What kind of output does a linear supply used in a digital audio design need to avoid degrading what should be a good analog representation of the digitized signal.
It's "only" audio, right?
So happy to be the fun for the day Fred, even more honored to be part of the entertainment of the week.
Koinichiwa,
Yes. With optimised local decoupling I found no reliable improvements in sound over several much simpler solutions I prefer.
NOTE that I AM aware of the very good technical performance of these regulators, it simply fails to translate into material improvement in sound, in my experience, ESPECIALLY considering the not inconsiderable effort and cost involved for these.
Of course, there are circuits out there that are VERY senstive to PSU Line noise (Naim Preamp's come to mind, they have no PSRR to speak of - hence you hear anything happening on the supply, equally, Master clocks for CD-Players are picky about PSU quality) where the differences may very well be audible.
For many Op-Amp's and similar sturcture discrete circuits there is plenty of PSRR which helps relativate the advantages of such a circuit as the Jung/Sulzer Reg over the 78XX/79XX.
One thing I personally have noted however, is that in "audio" circuits regulators that can sink current as well as source it (Shunt regulators, Bipolar Output Regs such as I proposed) offer often quite material improvements in sound over pretty good quality "series" regulators such the LT1085/1033, something I have also found in CMOS in digital circuits.
Anyway, just some thoughs. And for those that have not tried, stick a 3,300uF/6.3V or larger value high quality Cap (Panasonic FC/HFQ) onto the output of a 7805 supplying a bit of digital circuitry with good local decoupling. You may yet be surprised...
Sayonara
ALW said:
Yes. With optimised local decoupling I found no reliable improvements in sound over several much simpler solutions I prefer.
NOTE that I AM aware of the very good technical performance of these regulators, it simply fails to translate into material improvement in sound, in my experience, ESPECIALLY considering the not inconsiderable effort and cost involved for these.
Of course, there are circuits out there that are VERY senstive to PSU Line noise (Naim Preamp's come to mind, they have no PSRR to speak of - hence you hear anything happening on the supply, equally, Master clocks for CD-Players are picky about PSU quality) where the differences may very well be audible.
For many Op-Amp's and similar sturcture discrete circuits there is plenty of PSRR which helps relativate the advantages of such a circuit as the Jung/Sulzer Reg over the 78XX/79XX.
One thing I personally have noted however, is that in "audio" circuits regulators that can sink current as well as source it (Shunt regulators, Bipolar Output Regs such as I proposed) offer often quite material improvements in sound over pretty good quality "series" regulators such the LT1085/1033, something I have also found in CMOS in digital circuits.
Anyway, just some thoughs. And for those that have not tried, stick a 3,300uF/6.3V or larger value high quality Cap (Panasonic FC/HFQ) onto the output of a 7805 supplying a bit of digital circuitry with good local decoupling. You may yet be surprised...
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.