Hi folks.
A small update.
Beside supporting Tim. (kernel/squeezelite/system mods)
I've also been in touch with the PcP folks (Paul).
I've been looking at a very preliminary PcP version supplied by Paul based on the new kernel.
Looks really good/promising to me.
What's still required is the mapping of the device tree (new audio device names) on the web interface. Tim has done it already inside Moode.
I don't know about Tims or Pauls schedule when to launch the related updates.
I think especially Moode will make a nice step ahead. Especially with squeezelite now on board.
On the kernel side both projects would run pretty much the same kernel config though.
From a kernel perspective we agreed that we'll probably be able to maintain Clives kernel patches for a while - if no major flaws or changes pop up.
A small update.
Beside supporting Tim. (kernel/squeezelite/system mods)
I've also been in touch with the PcP folks (Paul).
I've been looking at a very preliminary PcP version supplied by Paul based on the new kernel.
Looks really good/promising to me.
What's still required is the mapping of the device tree (new audio device names) on the web interface. Tim has done it already inside Moode.
I don't know about Tims or Pauls schedule when to launch the related updates.
I think especially Moode will make a nice step ahead. Especially with squeezelite now on board.
On the kernel side both projects would run pretty much the same kernel config though.
From a kernel perspective we agreed that we'll probably be able to maintain Clives kernel patches for a while - if no major flaws or changes pop up.
Kali Reclocker + MamboBerry
Another update on my HW setup.
Allo sent me the high frequency clock Kali reclocker ( it's rather a I2S stream regenerator) for test.
Inmate DimDim made a comprehensive introduction. Therefore I'll spare more details.
From what I read most people agree that the jitter performance is pretty outstanding.
The MamboBerry was expected and has shown already to improve on a higher quality I2S signal.
I thought the KALI is well worth trying.
All I can say is on the first glance the board IMO looks highly professional made.
There's also a DIY-A thread about the board. I actually didn't read much of it.
I received my unit just recently. I hooked it up to a separate mravcla regulator.
At 100mA consumption there's no big PS challenge.
Mambo and Kali are powered by a single LiFePO4 pack.
The power jumper was removed and the micro switch turned to "HAT".
To the RPI the Kali is transparent. That's good.
I ran 384kHz against it. No issues. Actually, there's just a small one. There are onboard samplerate LED indicators. Max frequency is 192kHz.
If you go above 192kHz LED starts blinking. OK. Not long ago nobody was expecting samplerates >192kHz on PI I2S.
Still I don't like blinking LEDs! I find it annoying in a dark listening room.
Slowly but surely my PI combo looks like a flashy Christmas tree. Christmas is near though.
Blue, yellow, green, red, partially blinking LEDS. Ok. Let's not talk about cosmetics.
Each LED also consumes power. I'm not sure if all these LEDs are really required.
Another small issue I experience is the 3mm diameter of the mounting holes.
The PI comes with smaller holes (2.9mm) that require 2.5mm screws. The 3mm
Kali screws won't match, if you don't drill up the PI holes.
I think it's rather a very poor choice by the PI guys to go for 2.9 mm instead of 3.2 in the first place. Anyhow -- PI was first.
Functionality/Issues:
Some issues had been reported on early day units. Allo promised to have these fixed.
Most of them, if not all, were firmware issues as far as I read the comments.
For now I havn't experienced any issues.
Unfortunately there is no way to upgrade the firmware yourself.
If anything would happen again, the unit would have to be send back for patching.
Luckily my unit - in my setup - seems to be stable.
Sound:
As software I'm using my "modified/tweaked" PiCorePlayer 3.2 setup. It's still my reference setup.
The Kali does what you'd expect from a reclocker. It just cleans everything further up.
The 3D image in front of you gets more solid. The instruments get a little more 3Dimensionality and separation.
Everything just gets more energy, thus more presence. This will also carve out more details.
Bottom line I 'm looking at an even more clean, detailed, dynamic, rock solid, 3D performance. Very nice!
I also read that more and more people seem to start trying/using Kali as I2S interface towards
external DACs to get rid of the USB interfaces. I think DimDim mentioned that his Amanero USB
wouldn't do any better.
Interesting, though expected very nice developments.
Regarding noise. Some people call for isolators. Hmmmh. I can't hear or experience obvious noise issues in my setup.
If noise would be a problem such a low jitter ( claims are around 3ps, if I recall it correctly ) would probably not be possible to achieve.
Of course I do not know how my real world noise/jitter situation looks like. I need to trust my ears.
Bottom line.
At the current pricepoint I consider the Kali a must have device for those who want to make a PI a solid I2S source. ( Don't forget to use high quality powersupplies!!)
For sure this setup will get me happily over the winter season 2016/17.
However. We all know there's always space for improvement. Progress is what makes us happy on the long run! Looking forward to further developments.
Time to listen to some good music (and to bake some new kernels ).
Enjoy.
Another update on my HW setup.
Allo sent me the high frequency clock Kali reclocker ( it's rather a I2S stream regenerator) for test.
Inmate DimDim made a comprehensive introduction. Therefore I'll spare more details.
From what I read most people agree that the jitter performance is pretty outstanding.
The MamboBerry was expected and has shown already to improve on a higher quality I2S signal.
I thought the KALI is well worth trying.
All I can say is on the first glance the board IMO looks highly professional made.
There's also a DIY-A thread about the board. I actually didn't read much of it.
I received my unit just recently. I hooked it up to a separate mravcla regulator.
At 100mA consumption there's no big PS challenge.
Mambo and Kali are powered by a single LiFePO4 pack.
The power jumper was removed and the micro switch turned to "HAT".
To the RPI the Kali is transparent. That's good.
I ran 384kHz against it. No issues. Actually, there's just a small one. There are onboard samplerate LED indicators. Max frequency is 192kHz.
If you go above 192kHz LED starts blinking. OK. Not long ago nobody was expecting samplerates >192kHz on PI I2S.
Still I don't like blinking LEDs! I find it annoying in a dark listening room.
Slowly but surely my PI combo looks like a flashy Christmas tree. Christmas is near though.
Blue, yellow, green, red, partially blinking LEDS. Ok. Let's not talk about cosmetics.
Each LED also consumes power. I'm not sure if all these LEDs are really required.
Another small issue I experience is the 3mm diameter of the mounting holes.
The PI comes with smaller holes (2.9mm) that require 2.5mm screws. The 3mm
Kali screws won't match, if you don't drill up the PI holes.
I think it's rather a very poor choice by the PI guys to go for 2.9 mm instead of 3.2 in the first place. Anyhow -- PI was first.
Functionality/Issues:
Some issues had been reported on early day units. Allo promised to have these fixed.
Most of them, if not all, were firmware issues as far as I read the comments.
For now I havn't experienced any issues.
Unfortunately there is no way to upgrade the firmware yourself.
If anything would happen again, the unit would have to be send back for patching.
Luckily my unit - in my setup - seems to be stable.
Sound:
As software I'm using my "modified/tweaked" PiCorePlayer 3.2 setup. It's still my reference setup.
The Kali does what you'd expect from a reclocker. It just cleans everything further up.
The 3D image in front of you gets more solid. The instruments get a little more 3Dimensionality and separation.
Everything just gets more energy, thus more presence. This will also carve out more details.
Bottom line I 'm looking at an even more clean, detailed, dynamic, rock solid, 3D performance. Very nice!
I also read that more and more people seem to start trying/using Kali as I2S interface towards
external DACs to get rid of the USB interfaces. I think DimDim mentioned that his Amanero USB
wouldn't do any better.
Interesting, though expected very nice developments.
Regarding noise. Some people call for isolators. Hmmmh. I can't hear or experience obvious noise issues in my setup.
If noise would be a problem such a low jitter ( claims are around 3ps, if I recall it correctly ) would probably not be possible to achieve.
Of course I do not know how my real world noise/jitter situation looks like. I need to trust my ears.
Bottom line.
At the current pricepoint I consider the Kali a must have device for those who want to make a PI a solid I2S source. ( Don't forget to use high quality powersupplies!!)
For sure this setup will get me happily over the winter season 2016/17.
However. We all know there's always space for improvement. Progress is what makes us happy on the long run! Looking forward to further developments.
Time to listen to some good music (and to bake some new kernels
).Enjoy.
Last edited:
For someone whose first language is not English, that is very well described.
Hi soundcheck,
My hunch is the main benefit of an isolator before the DAC (Kali or no Kali) would be to separate the DAC ground from the RPi ground. If the grounds on all connected circuits were infinitely low impedance at all ponts there would be no issue, but in the real world things are rather different, especially as frequency rises. The $64,000 dollar question is how much of an issue is it?
Fortunately we probably won't have long to find out with two RPi isolators on the way.
I do think that the direct ground connections through the pin headers between PI, DAC and Kali do not leave much room for nasty groundloops or antenna effects.
That'll be much better than any USB connection.
EMI/RFI is probably a much bigger issue.
Just to mention it: I don't have an ethernet shield connected to the pi.
That'll be much better than any USB connection.
EMI/RFI is probably a much bigger issue.
Just to mention it: I don't have an ethernet shield connected to the pi.
Vasilis, great replies... thanks!
Though I doubt the I2S and clock signal routing is exactly identical between the Mamboberry (MB) and Mamboberry LS (MBLS), I suspect they are very similar because when I replaced the series resistors in my MB with the 51R value from the MBLS, I felt it was a small, but noticable improvement, lowering the level of high-frequency emphasis a bit. So I think your choice of value from the MBLS was a good one and I am very happy to copy it in my MB mods. THANKS!
Also, I tried the next step in my signal path mods, that of jumpering the 51R on the clock signal feed to 0R. I quickly decided that was a step backward and removed the jumper to put it back to 51R, further confirming Vasilis' choice here.
Before I go on, I do need to clarify a few things. First, I am not picking on the Mamboberry because I think it is deficient. I only decide to modify a piece if I think it shows promise... and if you've read my comments on modifying my HiFiBerry DAC+ Pro (HFBD+P), it took a LOT to get the HFBD+P up to the level of my mostly stock Mamboberry and some SW and configuration tweaks from Soundcheck to make it pull ahead. Modifying a poor piece of gear is a bad investment in time and parts... As I've said earlier, the Mamboberry is a very good piece of gear and so far modifying the Mamboberry is proving to be a great investment in my efforts!
Also, thanks to Soundcheck for reminding me of 3 critical prerequisite mods to the Mamboberry (and the MBLS too). They are:
1. Put an Allo.com Kali between the R-Pi and the MB or MBLS. At the price, this is a no-brainer for anyone wanting to get the best results from their R-Pi DAC HAT!
2. Use good power on the Kali. I've started with a DIY linear supply with an oversized transformer feeding a 47,000uf Jensen 4-pole filter capacitor. I then fed the 13v from this through either a 5v Dexa or Belleson regulator (marginally preferred the Belleson) to feed the Kali through the barrel jack. And I added a 100uf/800v CDE polyprop cap to the output of the Belleson by connecting it to the header (I saw this trick in the main SDTran384 thread on DIYAudio).
3. Use good power on the MB. Feeding appropriate AC to it as designed would be very good, I used a 2nd supply just like the one above and feed it to the output of the rectifier diodes on the MB board to bypass them.
IMHO, these steps are a must to get the best out of an MB (and MBLS too, I bet!).
Running out of typing time tonight, I'll continue with my next set of mods when I get more time. These are to the power supply filtering at the DAC and are closely modeled on the setup of EUVL's ES9022 DAC card (and if you go hunt down what he did, you can get a preview). Though not to keep everyone in total suspense, my initial assessment after a couple of 2 hour or so sessions with a day of warm-up/break-in between, is these mods hit the ball out of the park... gobs of detail, texture, dynamics, and bass, I am so overjoyed with the sound now that I am re-considering any further mods to the card because it is so good! Instead I am planning to re-focus getting farther on some of my higher-end DAC projects to again widen the gap between them and the formerly lowly (but high-achieving) Mamboberry which is now pushing up against their performance!
More in the next day or so.
Greg in Mississippi
P.S. And I just recently replaced the two linear supplies powering the Kali and Mamboberry with a pair of Uptone Audio LPS-1 Ultracapacitor supplies... these upped the sound quality quite a bit from the linear supplies and I can highly recommend them if you want THE BEST, but are a bit over the top in powering a $110USD R-Pi DAC HAT with $800 of power supplies!
Though I doubt the I2S and clock signal routing is exactly identical between the Mamboberry (MB) and Mamboberry LS (MBLS), I suspect they are very similar because when I replaced the series resistors in my MB with the 51R value from the MBLS, I felt it was a small, but noticable improvement, lowering the level of high-frequency emphasis a bit. So I think your choice of value from the MBLS was a good one and I am very happy to copy it in my MB mods. THANKS!
Also, I tried the next step in my signal path mods, that of jumpering the 51R on the clock signal feed to 0R. I quickly decided that was a step backward and removed the jumper to put it back to 51R, further confirming Vasilis' choice here.
Before I go on, I do need to clarify a few things. First, I am not picking on the Mamboberry because I think it is deficient. I only decide to modify a piece if I think it shows promise... and if you've read my comments on modifying my HiFiBerry DAC+ Pro (HFBD+P), it took a LOT to get the HFBD+P up to the level of my mostly stock Mamboberry and some SW and configuration tweaks from Soundcheck to make it pull ahead. Modifying a poor piece of gear is a bad investment in time and parts... As I've said earlier, the Mamboberry is a very good piece of gear and so far modifying the Mamboberry is proving to be a great investment in my efforts!
Also, thanks to Soundcheck for reminding me of 3 critical prerequisite mods to the Mamboberry (and the MBLS too). They are:
1. Put an Allo.com Kali between the R-Pi and the MB or MBLS. At the price, this is a no-brainer for anyone wanting to get the best results from their R-Pi DAC HAT!
2. Use good power on the Kali. I've started with a DIY linear supply with an oversized transformer feeding a 47,000uf Jensen 4-pole filter capacitor. I then fed the 13v from this through either a 5v Dexa or Belleson regulator (marginally preferred the Belleson) to feed the Kali through the barrel jack. And I added a 100uf/800v CDE polyprop cap to the output of the Belleson by connecting it to the header (I saw this trick in the main SDTran384 thread on DIYAudio).
3. Use good power on the MB. Feeding appropriate AC to it as designed would be very good, I used a 2nd supply just like the one above and feed it to the output of the rectifier diodes on the MB board to bypass them.
IMHO, these steps are a must to get the best out of an MB (and MBLS too, I bet!).
Running out of typing time tonight, I'll continue with my next set of mods when I get more time. These are to the power supply filtering at the DAC and are closely modeled on the setup of EUVL's ES9022 DAC card (and if you go hunt down what he did, you can get a preview). Though not to keep everyone in total suspense, my initial assessment after a couple of 2 hour or so sessions with a day of warm-up/break-in between, is these mods hit the ball out of the park... gobs of detail, texture, dynamics, and bass, I am so overjoyed with the sound now that I am re-considering any further mods to the card because it is so good! Instead I am planning to re-focus getting farther on some of my higher-end DAC projects to again widen the gap between them and the formerly lowly (but high-achieving) Mamboberry which is now pushing up against their performance!
More in the next day or so.
Greg in Mississippi
P.S. And I just recently replaced the two linear supplies powering the Kali and Mamboberry with a pair of Uptone Audio LPS-1 Ultracapacitor supplies... these upped the sound quality quite a bit from the linear supplies and I can highly recommend them if you want THE BEST, but are a bit over the top in powering a $110USD R-Pi DAC HAT with $800 of power supplies!
I followed iancanadas project in the early days.
As soon as it was clear that "Mr. Reclock" just works on an isolator-HAT, the project was dead from my perspective. It was clear from the very beginning that a reclocker would be required.(Did the project scope change btw.?)
Right now I have separate power supplies in place. That alone makes a difference.
(That'll lead to the question how much gain is caused by a separate supply or the isolation
of the isolator HAT )
Beside that - as discussed - there's a nice common ground between PI and HATs.
Proper grounding is the key to success in the first place.
However. Looking forward to read some feedback about these isolators.
You've gotta be kidding us.
First. That pretty much implies somehow that you believe your linears were close to the best. You don't know that.
What you know is that the Ultrcacap supply is a little better then your stuff.
Talking about THE BEST is what the pricetag implies. It's gotta be THE BEST at $800.
That supply delivers just 1A. That alone is a NoGo.
We have been playing around with UltraCap supplies for quite a while in the past.
On certain applications they did have a small effect (very low ESR) on others they didn't show any improvement.
I'd suggest, instead of fiddling around with 51R resistors in the clock path, try to focus
and get your PS on the level of that Ultracap supply.
Beside that I do believe, that with the PI+Kali+up2date AudioImage we now have an "I2S" streamer at hand that
beats the USB streamer (micoRendu) coming from that very same person at a fragment of cost.
Beside that we're ten times more flexible with our solution and skip the entire USB crap.
( Just mentioning this to protect you from the next weird buying decision.
)Enjoy.
Last edited:
One thing I dont understand abouLPS1 ..
Is the ground also isolated between the 2 capacitor banks ?
Is the ground also isolated between the 2 capacitor banks ?
"Vasilis great replies"
I'd like to comment that.
I don't think his reply is great at all!!!!
First of all it looks pretty emotional to me.
2nd he has made a design choice to stick with Sabre DACs. Of course he has to sell this decision. Obviously he's got to be preoccupied.
3rd. Sabre DACs are not everybody's taste either.
My 2 cents:
While working on the HifiBerry DAC+ Pro project, several modifications (HW and SW) had been applied.
I then had a chance to compare a MamboBerry and a modified HifiBerry DAC+ Pro side by side. (I'm running electro stat speakers and Abletec amps as main system btw)
My comparison ended up more or less into a draw.
Bottom line. I basically can't confirm anything Vasilis is saying.
I even sent the 1st. gen MamboBerry back.
My conclusion was that what really matters is the implementation.
When looking at the stock units. The MamboBerry was in the lead. After modifying the DAC+ Pro that wasn't the case anymore. At one point things just got very similar.
Later this year Vasilis offered me the MamboBerry LS for test as soon as it was released.
The reason why I jumped on that train was 1. a now attractive price-tag and 2. The new 5V supply option 3. I got annoyed by the crippled view of my modified DAC+ Pro.
4. the mistake to think the Mambo would support DSD and finally 5. a slight chance the new version might have improved a little over the old one.
Let's put it this way. I'm quite happy as it is right now. Though I'm not saying that I'm lightyears ahead of my modified Hifiberry installation. Not at all. ( Kali is not considered in this statement) )
But. There is no black or white in all this. It's always grey out there.
Enjoy.
I'd like to comment that.
I don't think his reply is great at all!!!!
First of all it looks pretty emotional to me.
2nd he has made a design choice to stick with Sabre DACs. Of course he has to sell this decision. Obviously he's got to be preoccupied.
3rd. Sabre DACs are not everybody's taste either.
My 2 cents:
While working on the HifiBerry DAC+ Pro project, several modifications (HW and SW) had been applied.
I then had a chance to compare a MamboBerry and a modified HifiBerry DAC+ Pro side by side. (I'm running electro stat speakers and Abletec amps as main system btw)
My comparison ended up more or less into a draw.
Bottom line. I basically can't confirm anything Vasilis is saying.
I even sent the 1st. gen MamboBerry back.
My conclusion was that what really matters is the implementation.
When looking at the stock units. The MamboBerry was in the lead. After modifying the DAC+ Pro that wasn't the case anymore. At one point things just got very similar.
Later this year Vasilis offered me the MamboBerry LS for test as soon as it was released.
The reason why I jumped on that train was 1. a now attractive price-tag and 2. The new 5V supply option 3. I got annoyed by the crippled view of my modified DAC+ Pro.
4. the mistake to think the Mambo would support DSD and finally 5. a slight chance the new version might have improved a little over the old one.
Let's put it this way. I'm quite happy as it is right now. Though I'm not saying that I'm lightyears ahead of my modified Hifiberry installation. Not at all. ( Kali is not considered in this statement) )
But. There is no black or white in all this. It's always grey out there.
Enjoy.
Last edited:
Nikkem, the Mamboberry is only designed to operate in ESS Async mode using the on-board clock as the reference. So all of my listening of the Mamboberry, both stock and after mods, has been in Async mode so far. OTOH, the next thing I intend to try is putting it in Sync mod by removing the 51R resistor in series with the clock signal and adding a jumper to take the clock signal as passed through on the Kali to the DAC chip side of that 51R resistor... will report after I do that.
On the sound, I really can't add anything to what Soundcheck wrote.... it really improves the sound top to bottom in every parameter with no drawbacks that I've been able to find yet.
It is. The LPS-1 switches between the two banks, with the one on charge not connected to the load in any way (neither power or ground) and the one connected to the load ONLY connected to the load (both power and ground) and not connected to the charging side in any way (neither power or ground).
You could get a similar effect with a good battery (I recommend LiPOFe4 cells) and a good regulator (they use the TPS7A4700 which is very good, I've personally also had good results with the ADM715x series and the LT3042). But only connect it to the load. I suspect that would sound very similar to the LPS-1. But then you have to deal with charging.
SoTM makes a dual switched battery setup that functionally works in a similar manner:
SOtM mBPS-d2s
But it costs a bit more than an LPS-1 and I know of one who has tried both and prefers the LPS-1:
UltraCap™ LPS-1 LISTENING IMPRESSIONS thread - Page 3
Vinnie Rossi uses a similar scheme with his new line of components (Lio) and is coming out with similar products, with higher voltage and current limits at much higher prices:
Ultracapacitor Power Supplies are coming (PURE-DC-4EVR: the MINI and the MEGA) !
and
Ultracapacitor Power Supplies are coming (PURE-DC-4EVR: the MINI and the MEGA) !
Greg in Mississippi
P.S. More responses later when I have some time.
Last edited:
I just seen this..
Clive and I had our share of disagreements . We also have some things that we agree on.
A week before , I get an email from Clive and I got mad. No excuse , but that was the week where we had a bug in Kali firmware , we had to arrange for replacements etc..so I get mad and told him to stop lying .
This is my admission that I was wrong and my apology to Clive.
You are a real asset to this community . I hope to see you back..
Clive and I had our share of disagreements . We also have some things that we agree on.
A week before , I get an email from Clive and I got mad. No excuse , but that was the week where we had a bug in Kali firmware , we had to arrange for replacements etc..so I get mad and told him to stop lying .
This is my admission that I was wrong and my apology to Clive.
You are a real asset to this community . I hope to see you back..
From here:
Mystery revealed: UpTone Audio "UltraCap™ LPS-1" - Page 13
"From a sound aspect there is no reason for going with a linear charger, remember the output is isolated from the charging circuit. The only way any of the noise from the charging circuit can get to the output is through the transistors that switch the bank from charge to output. That can only happen through the capacitance of the transistors, so I choose very low capacitance transistors. They are hard to drive so it took a lot of work in the design, but they are working great now. "
Vinnie Rossi USED to use relays, but is moving towards a similar setup and offering upgrades for customers with the relay version. See more info here:
Announcing new LIO ?SS? upgrade (Silent Switching)
Greg in Mississippi
You've gotta be kidding us.
First. That pretty much implies somehow that you believe your linears were close to the best. You don't know that.
What you know is that the Ultrcacap supply is a little better then your stuff.
Talking about THE BEST is what the pricetag implies. It's gotta be THE BEST at $800.
That supply delivers just 1A. That alone is a NoGo.
We have been playing around with UltraCap supplies for quite a while in the past.
On certain applications they did have a small effect (very low ESR) on others they didn't show any improvement.
I'd suggest, instead of fiddling around with 51R resistors in the clock path, try to focus
and get your PS on the level of that Ultracap supply.
Beside that I do believe, that with the PI+Kali+up2date AudioImage we now have an "I2S" streamer at hand that
beats the USB streamer (micoRendu) coming from that very same person at a fragment of cost.
Beside that we're ten times more flexible with our solution and skip the entire USB crap.
( Just mentioning this to protect you from the next weird buying decision.
Enjoy.
Soundcheck,
First I believe my linear supplies to be very good. A lot of my power supply philosophy comes from PS Audio's 'HCPS' where they used oversized transformers and Peter Daniels' DIY Audio DIY amp, CD drive, DAC, and phono stage supplies.
I use:
1. I-E core transformers to limit primary to secondary noise coupling
2. Good well-over-spec'd soft recovery diodes
3. Jensen 4-pole capacitors which use the internal inductance and resistive elements of the capacitor construction to provide additional filtering (they are sorta like C-L-C filters)
4. Good connectors at the output side (Hirose mil-spec).
And while mine are good, I know there are better ones, at a cost in both $ and size. I know mine are better than most EBay Chinese supplies with standard capacitors and normally-sized transformers. The ones that are better are using higher-cost transformers (R-Core) and compound filtering (C-L-C, C-R-C, etc.).
BTW, in raw $, one of my AC-powered linear supplies as used with the Kali and Mamboberry (providing 2 output rails) costs about $250-$275 in raw parts not including mounting HW or cases, wire, and connectors. Upgrading these to C-L-C supplies would add about $120 per 2 output rails AND a doubling of the space needed.
The Uptone Audio LPS-1, at $395 per supply rail, provides a good power source with very low-impedence source devices (the ultracaps) that are disconnected from the charging source or any AC power or ground side connection. As I said above, I believe you can get as good of a supply using LiPoFE4 batteries and a good regulator, but once you connect it to the a charger you have re-introduced a connection to the AC with the attendant AC leakage loops and the noise that introduces. And you have to deal with both charging the cells AND ensuring you don't discharge them too low and ruin them.
Another SOTA scheme promoted on the Tir Na Hifi forums is using a linear supply to 'float charge' LiPoFE4 batteries that are closely coupled to the load circuits. I have not tried this yet, but people I trust say it is the best thing out there. The main limitations I see of this setup are:
1. It works best in multiples of 3.3v (the raw battery voltage level). They have it working for things like 5v, but that takes some care to achieve and keep running.
2. It stays connected to the AC line, which means that AC leakage loops and the attendant noise is present. The LiPoFE4 cells DO seem to do a good job of minimizing this though, based on reports.
3. It is a DIY-project that many won't or even shouldn't undertake.
OTOH, if I did not have the LPS-1 supplies available to me, I'd be using the float charged LiPoFE4 setups. Battery supplies without the float charging are just too much trouble and risk IMHO.
OTOH, I really want to try combining LPS-1's with float charged LiPoFE4 batteries for possibly the best of both worlds!
If you want >1A, buy Vinnie Rossi's supplies... for $995 per supply rail.
I think we agree that power supplies are critical to the sound. What I heard going from my AC-powered linear supplies (that are very good, but admittedly not the best available) to 2 LPS-1's showed me just how critical they are and what else is available beyond what most of us have tried and/or heard.
Finally, on I2S streamer versus a great USB one like the Sonore uRendu... I am a committed I2S connection person for my computer music playback setups. I sort of lucked into this when I decided to go down the cMP/cPlay path which used a PCI Juli@ as the output device AND it had easily accessable I2S pins. Several iterations of using either that I2S interface (with modified Juli@ analog cards and other add-on I2S-connected DAC cards) versus using the S/PDIF output from the Juli@ had me convinced that I2S was a better interface AS LONG AS one could use it (which then required extreme DIY abilities). AND removing the USB capabilities out of my cMP/cPlay's operating system showed me how much the overhead of USB causes noisy processing and damages the sound, EVEN if it is not being used!
Ok, so I am a died-in-the-wool I2S affectionado who CAN DIY a DAC to use that. No problem for me, but most other audio enthusiasts use USB-output from their playback computer into a USB-input DAC or interface. For them, the uRendu is a significant step up as it is a device built as an audio device, with all low-noise regulators and digital-audio quality power distribution networks and grounding designed by someone who's day job is doing that in a very competitive and highly technical industry (power network design for server farm environments). I worked with the same designer on an I2S-based design for a time and was impressed with his knowledge and design chops. I was very sad when that project died due to limitations in the base computer HW. AND I see that he has gone beyond the target design for that HW in the uRendu. If that I2S-based design HAD come to fruition, we'd be able to directly compare it to what we can do with consumer-level computer HW and our reclocker and DAC HATs. AND I think we are achieving incredible results. The explosion of DAC HATS along with the Kali reclocker allows us to achieve results far beyond what we could get a couple of years ago. But I still firmly believe that an equivalent purpose-designed R-Pi equivalent I2S digital audio source by someone who really knows what they are doing would take the results even further.
BTW, you don't have to worry about me buying a uRendu, not in my future. I am FIRMLY committed to I2S... it would take something making a quantum leap beyond USB to move me from that.
My 2 cents!
Greg in Mississippi
Last edited:
Guys we forget the simple modification of LPF out , easy and effective , if you want replace the through hole resistors and capacitors with better quality ( this the reason to choice TH in this place )
I prefer DALE ( the brown ) , and for capacitors silver mica 4700 pf , ( yes i now its the half cost of the board ,, but worth the last penny .. )
I prefer DALE ( the brown ) , and for capacitors silver mica 4700 pf , ( yes i now its the half cost of the board ,, but worth the last penny .. )
I once wrote you in a private mail about my view about JS postings:
Man, I do have a strong feeling you could join that brotherhood.
I think that's what I suggested in the first place.
Your old school linear supplies are obviously not the best option. I mean - you experienced it yourself.
LiFePO4 supplies and Ultracaps are discussed over here for years.
Svensson and Rosso just jumped on that train much later (Rosso quite a bit earlier).
Who of them is building a more insane 1A supply doesn't really matter to me !!!
I wouldn't buy any of them.
And I can tell you that a PI IMO don't need an overexpensive 1A supercap supply
feeding some mediocre switching regs from the outside! You don't use a sledgehammer to crack a nut.
As I said, what matters is the application.
I'd also suggest...
...if you want to discuss that 100% commercial supply option any further, I'd appreciate if you'd open a new thread in the power supply forum. Or you join the fanclub over at CA.
Enjoy.
Man, I do have a strong feeling you could join that brotherhood.
I think that's what I suggested in the first place.
Your old school linear supplies are obviously not the best option. I mean - you experienced it yourself.
LiFePO4 supplies and Ultracaps are discussed over here for years.
Svensson and Rosso just jumped on that train much later (Rosso quite a bit earlier).
Who of them is building a more insane 1A supply doesn't really matter to me !!!
I wouldn't buy any of them.
And I can tell you that a PI IMO don't need an overexpensive 1A supercap supply
feeding some mediocre switching regs from the outside! You don't use a sledgehammer to crack a nut.
As I said, what matters is the application.
I'd also suggest...
...if you want to discuss that 100% commercial supply option any further, I'd appreciate if you'd open a new thread in the power supply forum. Or you join the fanclub over at CA.
Enjoy.
Last edited:
Thx for the hint.
On the HifiBerry I removed the output LPF altogether. (I never received any feedback if somebody else tried that)
And I also removed the RCA jacks. That's a must! (Even a $200 audiophile jack/plug combo would not beat a 2ct soldering joint.)
Bottom line.
You can't have it cheaper than that.
In any case, for those afraid of catching some HF junk by removing the output LPF , you should try to find out if your amp has a LPF filter on the input!!
That would make the DAC output LPF redundant anyhow!!
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.