...at what cost silence (i.e. reduced ripple)?

[davide256]

1 hour ago, the_doc735 said:

errr! - £2299! - "No!"

reading about how a quality sports car is made better doesn't mean you have to buy one... it just helps you understand what makes better engineering design

for a performance vehicle. Same thing applies for audio.

[the_doc735]

54 minutes ago, 4est said:

Then how so? Most i2s sources start off as USB, and they really are no different than a USB DAC.

this pink faun i2s PCIe riser card

[the_doc735]

33 minutes ago, davide256 said:

reading about how a quality sports car is made better doesn't mean you have to buy one... it just helps you understand what makes better engineering design

for a performance vehicle. Same thing applies for audio.

ok

[sandyk]

3 hours ago, Abtr said:

If the objective is output noise reduction, then according to the LT1963 datasheet 40uV is obtained with 10uF output capacitance; not 1000uF.

 Get real !

 I didn't design this PCB, and it has been reported to work well in practice, which is something you appear to have very little of, judging by your previous comments about fuses always being able to protect against supply rail short circuits !!

 Every device to be powered already has an Input capacitor across it's D.C. input of a value  that we have no control over.

 This means that the optimistic figures that many voltage regulator manufacturers quote is highly unlikely to be achieved in the real world unless the voltage regulator is an integral part of the original design and it's requirements for highest performance are fully taken into account.

A 100uF Panasonic FC or FM electrolytic across the input , and using the series bridge rectifier diode instead of the D.C. input terminals as I have shown, will help to reduce the amount of SMPS HF rubbish going into the voltage regulator.

 

[4est]

3 hours ago, the_doc735 said:

this pink faun i2s PCIe riser card

Ok, TY, just wondered.

[the_doc735]

5 hours ago, sandyk said:

 Get real !

 I didn't design this PCB, and it has been reported to work well in practice, which is something you appear to have very little of, judging by your previous comments about fuses always being able to protect against supply rail short circuits !!

 Every device to be powered already has an Input capacitor across it's D.C. input of a value  that we have no control over.

 This means that the optimistic figures that many voltage regulator manufacturers quote is highly unlikely to be achieved in the real world unless the voltage regulator is an integral part of the original design and it's requirements for highest performance are fully taken into account.

A 100uF Panasonic FC or FM electrolytic across the input , and using the series bridge rectifier diode instead of the D.C. input terminals as I have shown, will help to reduce the amount of SMPS HF rubbish going into the voltage regulator.

 

???

[the_doc735]

2 hours ago, 4est said:

Ok, TY, just wondered.

?

[sandyk]

52 minutes ago, the_doc735 said:

???

 

Hi Doc

 The attached diagram from the LT1963 Data Sheet shows that a 10 x higher value Tantalum (100uF) capacitor plus extra parallel on board  low value capacitors can further improve ripple rejection at >100KHZ over just a 10uF Tantalum.

 In addition, the 3 x  1,000uF capacitors used are Rubycon YXF 35V types which have a Low ESR (100KHZ)

(Effective Series Resistance)

 

Regards

Alex

 

 

 

[the_doc735]

6 minutes ago, sandyk said:

 

Hi Doc

 The attached diagram from the LT1963 Data Sheet shows that a 10 x higher value Tantalum (100uF) capacitor plus extra parallel on board  low value capacitors can further improve ripple rejection at >100KHZ over just a 10uF Tantalum.

 In addition, the 3 x  1,000uF capacitors used are Rubycon YXF 35V types which have a Low ESR. (100KHZ)

 

Regards

Alex

 

 

 

...don't understand this level of electronics, but thank you for this information, very grateful to you! Bless!

 

...just wish I could get more direct answers to my initial question, that is understandable to the average consumer.

[4est]

6 hours ago, the_doc735 said:

...don't understand this level of electronics, but thank you for this information, very grateful to you! Bless!

 

...just wish I could get more direct answers to my initial question, that is understandable to the average consumer.

To sorta answer your question then, the importance is not just the 5mV or 10mV ripple in the power supply, but it's rejection of high frequency noise and its ability to respond to the instantaneous current changes of the powered devices. Does that make sense?

[Abtr]

16 hours ago, sandyk said:

 Get real !

 I didn't design this PCB, and it has been reported to work well in practice, which is something you appear to have very little of, judging by your previous comments about fuses always being able to protect against supply rail short circuits !!

I said elsewhere that a short in a power cable will/should only blow the fuse (or trip a protection circuit) of the power supply, whereas you were talking about a stray wire strand possibly connecting to sensitive parts of some experimental circuit, which was completely irrelevant in a discussion about connecting a simple plug to a DC power cable, just as it is irrelevant to bring this up here.

 

Anyway, the parallel diodes protect the large input and output caps against reversed DC voltage. Not sure what a diode at the output does in terms of reverse bias leakage current and noise. Rather unusual to see a diode across the output of a low noise voltage regulator. But then again it may not be a low noise regulator circuit at all.

 

16 hours ago, sandyk said:

 Every device to be powered already has an Input capacitor across it's D.C. input of a value  that we have no control over.

 This means that the optimistic figures that many voltage regulator manufacturers quote is highly unlikely to be achieved in the real world unless the voltage regulator is an integral part of the original design and it's requirements for highest performance are fully taken into account.

Exactly. So why install an additional 1000uF(!) at the output? Cout=100uF may slightly reduce HF ripple rejection but will most likely increase output noise relative to Cout=10uF.

 

16 hours ago, sandyk said:

A 100uF Panasonic FC or FM electrolytic across the input , and using the series bridge rectifier diode instead of the D.C. input terminals as I have shown, will help to reduce the amount of SMPS HF rubbish going into the voltage regulator.

 

How and why? There are already 3 caps at the input and reducing (HF) rubbish is what the LT1963 is supposed to do.

[the_doc735]

6 hours ago, 4est said:

To sorta answer your question then, the importance is not just the 5mV or 10mV ripple in the power supply, but it's rejection of high frequency noise and its ability to respond to the instantaneous current changes of the powered devices. Does that make sense?

yes, definitely! cheers!

[sandyk]

5 hours ago, Abtr said:

How and why? There are already 3 caps at the input and reducing (HF) rubbish is what the LT1963 is supposed to do.

 The LT1963 is not only used with SMPS. The extra filtering will also reduce mains frequency related ripple before it gets into the regulator.

 You appear to be confusing noise out with the slew rate of the regulator.

 A single strand of stray wire IS sometime capable of blowing up a regulator or associated components, especially the lower powered voltage regulators. BTW, You previously stated WILL not WILL/SHOULD blow the fuse. Even Ralf11 disagreed with you there. .

5 hours ago, Abtr said:

Cout=100uF may slightly reduce HF ripple rejection but will most likely increase output noise relative to Cout=10uF.

Are you even able to fully understand Data Sheets and terminology ?

 The last paragraph of your previous reply where you also mentioned ESR  suggests otherwise.

B.TW., using a series diode with a DC supply is an old trick to further isolate sections of the PSU

 

 I have better things to do than play silly games with you ! 
 

[look&listen]

On 7/9/2018 at 11:54 PM, the_doc735 said:

 ...at what cost silence

Maybe time for you to research Paul Hynes SR-7 LPSU(s). Has rails & amps for you application, & cost is education.

[the_doc735]

2 hours ago, sandyk said:

 The LT1963 is not only used with SMPS. The extra filtering will also reduce mains frequency related ripple before it gets into the regulator.

 You appear to be confusing noise out with the slew rate of the regulator.

 A single strand of stray wire IS sometime capable of blowing up a regulator or associated components, especially the lower powered voltage regulators. BTW, You previously stated WILL not WILL/SHOULD blow the fuse. Even Ralf11 disagreed with you there. .

Are you even able to fully understand Data Sheets and terminology ?

 The last paragraph of your previous reply where you also mentioned ESR  suggests otherwise.

B.TW., using a series diode with a DC supply is an old trick to further isolate sections of the PSU

 

 I have better things to do than play silly games with you ! 
 

???

[the_doc735]

2 hours ago, look&listen said:

Maybe time for you to research Paul Hynes SR-7 LPSU(s). Has rails & amps for you application, & cost is education.

will look into it, many thanks!

[look&listen]

2 hours ago, sandyk said:

A single strand of stray wire IS sometime capable of blowing up a regulator or associated components, especially the lower powered voltage regulators. BTW, You previously stated WILL not WILL/SHOULD blow the fuse.

 

[Abtr]

3 hours ago, sandyk said:

 The LT1963 is not only used with SMPS. The extra filtering will also reduce mains frequency related ripple before it gets into the regulator.

Yes, but in your modified schematics, there already is 2 x 1000uF (+ 1uF + 10nF) low ESR capacitance at the input, so why the additional 100uF Panasonic?

 

Quote

 You appear to be confusing noise out with the slew rate of the regulator.

I don't think so..

 

Quote

 A single strand of stray wire IS sometime capable of blowing up a regulator or associated components, especially the lower powered voltage regulators. BTW, You previously stated WILL not WILL/SHOULD blow the fuse. Even Ralf11 disagreed with you there. .

If a short blows the regulator before the fuse, then the fuse has the wrong spec, otherwise, what's the point of a fuse? And IME the fuse generally does blow first.

 

Quote

Are you even able to fully understand Data Sheets and terminology ?

 The last paragraph of your previous reply where you also mentioned ESR  suggests otherwise.

Sorry for the typo. I meant to say that a Cout of 100uF (instead of 10uF) may improve HF ripple rejection (as per the graph you posted), but at the same time it may increase output noise. Why do you think all the noise specs in the LT1963 datasheet are based on a Cout of 10uF? And I really don't see any reason for 1000uF at the output.  

 

Quote

B.TW., using a series diode with a DC supply is an old trick to further isolate sections of the PSU

Ah, thank you.

 

Quote

 I have better things to do than play silly games with you ! 

I'm only asking you some questions, basically, but it's of course your prerogative not to answer..

 

[sandyk]

As Ralf11 agreed, a fuse does not blow instantaneously. This is especially so when the  PSU is not capable of far greater current than actually needed.

The 100uF FM  Panasonic will further reduce SMPS ripple before it gets into the regulator section, and may also help a little for other devices which may be connected from the same group of Molex 4 pin connectors in a P.C.

In addition, the forward resistance of the conducting Bridge Rectifier  diode, plus the 2  parallel 1,000uF Low ESR Rubycon YXF capacitors will also act as an additional RC filter when used with a D.C. supply as I described .

This will make life easier for the LT1963 with HF rubbish.

 

 With say a relatively low powered Stereo amplifier, (e.g. I.C. type) the use of a series diode will prevent one channel of the amplifier which may have a sustained low frequency note from also dragging down the voltage of the other channel , assuming adequate value filter capacitors are used.

 

Perhaps the Asian guy who designed this ebay Regulated PSU put a lot more thought into it than at first glance, with his choice of the YXF electrolytics and the carefully chosen values of the smaller parallel capacitors ?

 If the voltage regulator hadn't been designed for such a wide variation in Voltage output, he MAY  have further refined the ease of voltage adjustment by replacing the 330 ohms with say 120 ohms, and the 10k TRIMPOT WITH A 2kOhm 20turn type as I have done with mine ?

[Ralf11]

"even" ???

[barrows]

Just because one uses a voltage regulator with a spec of xx µV output noise does not mean that is how it performs in the circuit.  The specification is for self noise of the regulator, the actual output noise is a function of the input ripple, the power supply rejection ratio, and the self noise of the regulator.  Just throwing a LT 3045 at a "problem" will not "fix" everything.

With Sonore's top model, the Signature Rendu SE, we went to great lengths with power supplies to make power supply noise a non-issue.  The main power supply has output noise in the two digits of µV range (under high load conditions), and then we use ultra low noise, linear, regulators for all of the local supplies on the main board.  For an Ethernet Renderer, we indeed found improved performance by reducing power supply noise as low as possible-these are advantages only available when one gets rid of standard computer mother boards (with their onboard DC/DC converters (AKA switching regulators) from the components in the audio system.  Additionally, output impedance over a wide bandwidth is critical for high speed digital circuitry, so focusing on noise (under a constant load) alone only addresses some of the issues of power supply design for high speed circuits.

 

Suffice it to say, power supply noise in the mV level is not all that impressive these days!

 

On shunt regulation: this style of regulator can offer superb performance both on noise and output impedance, but for digital circuits such as computer processors, whose power requirements vary over a very wide margin (for example, boot current might require 2A, while the average running current is only 400 mA), a shunt regulator will require very high shunt current, and extensive heat sinking, as such it will be highly inefficient, but it can be done.  Better in almost every case to use excellent linear regulation for computer processor type circuits, being much more efficient and save the shunts for more constant loads where they make more sense. 

[the_doc735]

25 minutes ago, barrows said:

Just because one uses a voltage regulator with a spec of xx µV output noise does not mean that is how it performs in the circuit.  The specification is for self noise of the regulator, the actual output noise is a function of the input ripple, the power supply rejection ratio, and the self noise of the regulator.  Just throwing a LT 3045 at a "problem" will not "fix" everything.

With Sonore's top model, the Signature Rendu SE, we went to great lengths with power supplies to make power supply noise a non-issue.  The main power supply has output noise in the two digits of µV range (under high load conditions), and then we use ultra low noise, linear, regulators for all of the local supplies on the main board.  For an Ethernet Renderer, we indeed found improved performance by reducing power supply noise as low as possible-these are advantages only available when one gets rid of standard computer mother boards (with their onboard DC/DC converters (AKA switching regulators) from the components in the audio system.  Additionally, output impedance over a wide bandwidth is critical for high speed digital circuitry, so focusing on noise (under a constant load) alone only addresses some of the issues of power supply design for high speed circuits.

 

Suffice it to say, power supply noise in the mV level is not all that impressive these days!

 

On shunt regulation: this style of regulator can offer superb performance both on noise and output impedance, but for digital circuits such as computer processors, whose power requirements vary over a very wide margin (for example, boot current might require 2A, while the average running current is only 400 mA), a shunt regulator will require very high shunt current, and extensive heat sinking, as such it will be highly inefficient, but it can be done.  Better in almost every case to use excellent linear regulation for computer processor type circuits, being much more efficient and save the shunts for more constant loads where they make more sense. 

???

"we indeed found improved performance by reducing power supply noise as low as possible-these are advantages only available when one gets rid of standard computer mother boards (with their onboard DC/DC converters (AKA switching regulators) from the components in the audio system."

[sandyk]

 

I basically agree with everything Barrows has said here, but as I stated originally, this LT1963 PSU from $30 ebay can give a modest, but worthwhile improvement when powering an OS SSD for around only $30 when used in conjunction with an existing +12V SMPS supply in the computer, or further improve the power from a big Linear PSU such as the HDPlex for .areas such as an  SSD or internal USB card.. 

 

 Yes, there is far more to it than simply throwing  .8uV noise  LT3045s at the problem areas as some members are trying to do.

[the_doc735]

8 minutes ago, sandyk said:

 

I basically agree with everything Barrows has said here, but as I stated originally, this LT1963 PSU from $30 ebay can give a modest, but worthwhile improvement when powering an OS SSD for around only $30 when used in conjunction with an existing +12V SMPS supply in the computer, or further improve the power from a big Linear PSU such as the HDPlex for .areas such as an  SSD or internal USB card.. 

 

 Yes, there is far more to it than simply throwing  .8uV noise  LT3045s at the problem areas as some members are trying to do.

???

[look&listen]

38 minutes ago, the_doc735 said:

this LT1963 PSU from $30 ebay

$30? Same one I used by itself, then as 1st stage of new DIY LPSU build was only $19 from eBay. Maybe because year or 2 ago, but eBay prices on Chinese electronic parts highly variable. Wise to search eBay for description & see variety of sellers & prices.