I pretty much finalized my setup.
Meanwhile I bought a rework station. With hot air it's been fun to remove pretty much all remaining parts from the MamboBerry.
I'm now running 3.6V from Ales reg. There I also removed all sockets, rectifier and the Pana FCs (replaced with 2*BGs 1000uF) . All power rails of the "DAC formerly known as MamboBerry" are now sitting idle.
Then I added a coax ( taken from a WLAN antenna) for the MCLK.
The LiFePo4 batteries I replaced with a 2nd iFi iPower and another Rifa 47k which I had around.
To the Kali main power rail I attached 1650uF (5*330uF) Oscons.
Then I also had to tweak my Anaview AMS0100 amp.
I replaced the rectifier with Hexfreds. ( A hint from an inmate who uses an Anaview ALC amp)
And I also replaced the onboard fuse with a piece of wire. (No Risk no Fun).
Overall result. Real nice sound. Music flows.
No harshness at all anymore. Better resolution than ever. Very natural. Applause and S sounds sound just right on spot.
What more do I want???
Ah. I introduced dithered volume control to squeezelite. Many quality audio players offer that feature, even on 24 bit data.
Especially those of us running extensive digital volume control should have a look at it. I'm still testing it.
However. Below an example - quick and dirty hack:
( the e-weighted filter applies to 44k1 only, "dither -S -p23" would serve all samplerates)
Would be nice to have a dither option (after the volume control) build into squeezelite.
Happy holidays.
Meanwhile I bought a rework station. With hot air it's been fun to remove pretty much all remaining parts from the MamboBerry.
I'm now running 3.6V from Ales reg. There I also removed all sockets, rectifier and the Pana FCs (replaced with 2*BGs 1000uF) . All power rails of the "DAC formerly known as MamboBerry" are now sitting idle.
Then I added a coax ( taken from a WLAN antenna) for the MCLK.
The LiFePo4 batteries I replaced with a 2nd iFi iPower and another Rifa 47k which I had around.
To the Kali main power rail I attached 1650uF (5*330uF) Oscons.
Then I also had to tweak my Anaview AMS0100 amp.
I replaced the rectifier with Hexfreds. ( A hint from an inmate who uses an Anaview ALC amp)
And I also replaced the onboard fuse with a piece of wire. (No Risk no Fun).
Overall result. Real nice sound. Music flows.
No harshness at all anymore. Better resolution than ever. Very natural. Applause and S sounds sound just right on spot.
What more do I want???
Ah. I introduced dithered volume control to squeezelite. Many quality audio players offer that feature, even on 24 bit data.
Especially those of us running extensive digital volume control should have a look at it. I'm still testing it.
However. Below an example - quick and dirty hack:
Code:
squeezelite -n sl1 -b 10000:660000 -c pcm,flac,mp3 -a 32 -W -o - | sox -q -t raw -c 2 -r 44100 -b 32 -e signed - -t alsa -b 32 hw:0,0 dither -s -f improved-e-weighted -p 23( the e-weighted filter applies to 44k1 only, "dither -S -p23" would serve all samplerates)
Would be nice to have a dither option (after the volume control) build into squeezelite.
Happy holidays.
Last edited:
I pretty much finalized my setup.
Meanwhile I bought a rework station. With hot air it's been fun to remove pretty much all remaining parts from the MamboBerry.
I'm now running 3.6V from Ales reg. There I also removed all sockets, rectifier and the Pana FCs (replaced with 2*BGs 1000uF) . All power rails of the "DAC formerly known as MamboBerry" are now sitting idle.
Then I added a coax ( taken from a WLAN antenna) for the MCLK.
The LiFePo4 batteries I replaced with a 2nd iFi iPower and another Rifa 47k which I had around.
To the Kali main power rail I attached 1650uF (5*330uF) Oscons.
Then I also had to tweak my Anaview AMS0100 amp.
I replaced the rectifier with Hexfreds. ( A hint from an inmate who uses an Anaview ALC amp)
And I also replaced the onboard fuse with a piece of wire. (No Risk no Fun).
Overall result. Real nice sound. Music flows.
No harshness at all anymore. Better resolution than ever. Very natural. Applause and S sounds sound just right on spot.
What more do I want???
Ah. I introduced dithered volume control to squeezelite. Many quality audio players offer that feature, even on 24 bit data.
Especially those of us running extensive digital volume control should have a look at it. I'm still testing it.
However. Below an example - quick and dirty hack:
Code:squeezelite -n sl1 -b 10000:660000 -c pcm,flac,mp3 -a 32 -W -o - | sox -q -t raw -c 2 -r 44100 -b 32 -e signed - -t alsa -b 32 hw:0,0 dither -s -f improved-e-weighted -p 23
( the e-weighted filter applies to 44k1 only, "dither -S -p23" would serve all samplerates)
Would be nice to have a dither option (after the volume control) build into squeezelite.
Happy holidays.
Oh no, what has happened to this thread? No more updates? I love this information sharing!
Yep. For a month -- I at least -- was not doing much. 
I let that dither thing go.
I also tried server based volume control and dithering.
It didn't work out that well.
There's a reason why volume control is done as pretty much last activity before the stream gets send to the DAC.
Controllers like e.g. iPeng wouldn't even support that approach at all.
I also checked with the squeezelite folks.
They didn't seem to be interested to look into it.
Usually it takes a volunteer to indroduce a new function such as dithered
volume control. Not many people are willing (or able) to contribute stuff.
Right now I'm looking into
a. Reviewing and potentially improving the grounding scheme of the system.
Proper grounding is key!
Basically all different supplies and regulators and cabling need to be reviewed. That includes the amplifier - the entire system.
One question is - what's the best position or best implementation for a starground on my specific system?? You might want to read this
.
b. DAC power
Temporarily I'm powering the Sabre with 3.6V from Ales reg supplied from my 7.2V
LiFePO4 2*3 battery pack. Basically I'm running 3 supplies now.
That's actually a little better than having one iPower powering both, the Kali and the
DAC. Is it worth it ??? Hmmh. The difference is not earth shattering.
For tweakers like me though, going back even a tiny step usually is a NoGO.
I still do find the battery handling more than annoying.
The Q - which annoyance is bothering more !?!?
Inmate Eldam proposed to have a look at a R-core transformer for the Ales reg as a nice alternative running LiFePO4s. Hmmh. At least $30 for just a try.
And, as I said earlier, the linear supplies I used in the past were not convincing me.
c. DAC reg
Another option would be to look at a better 3.6V regulator then the TPS7A4700 for the DAC.
I have no idea though what that could be.
Twisted Pear offers 3.5V shunt regs. Not sure if that's worth it. And if it works. It's just 3.5V.
Any feedback/advise is welcome here.
I let that dither thing go.
I also tried server based volume control and dithering.
It didn't work out that well.
There's a reason why volume control is done as pretty much last activity before the stream gets send to the DAC.
Controllers like e.g. iPeng wouldn't even support that approach at all.
I also checked with the squeezelite folks.
They didn't seem to be interested to look into it.
Usually it takes a volunteer to indroduce a new function such as dithered
volume control. Not many people are willing (or able) to contribute stuff.
Right now I'm looking into
a. Reviewing and potentially improving the grounding scheme of the system.
Proper grounding is key!
Basically all different supplies and regulators and cabling need to be reviewed. That includes the amplifier - the entire system.
One question is - what's the best position or best implementation for a starground on my specific system?? You might want to read this
.
b. DAC power
Temporarily I'm powering the Sabre with 3.6V from Ales reg supplied from my 7.2V
LiFePO4 2*3 battery pack. Basically I'm running 3 supplies now.
That's actually a little better than having one iPower powering both, the Kali and the
DAC. Is it worth it ??? Hmmh. The difference is not earth shattering.
For tweakers like me though, going back even a tiny step usually is a NoGO.
I still do find the battery handling more than annoying.
The Q - which annoyance is bothering more !?!?
Inmate Eldam proposed to have a look at a R-core transformer for the Ales reg as a nice alternative running LiFePO4s. Hmmh. At least $30 for just a try.
And, as I said earlier, the linear supplies I used in the past were not convincing me.
c. DAC reg
Another option would be to look at a better 3.6V regulator then the TPS7A4700 for the DAC.
I have no idea though what that could be.
Twisted Pear offers 3.5V shunt regs. Not sure if that's worth it. And if it works. It's just 3.5V.
Any feedback/advise is welcome here.
On a Master Clock with Crystek 957, I find there is a big difference with an Àles reg with 2 cm wire between the board plugs and the reg and a 10 cm wire !
Both has to be imho tried, firstly to hear how is sensible your DAC to the supply !, of course the 10 cm wire has to be twisted ! Also different thickness gave me different result ?! I assume only thr ground return thickness is important, but I didn't make more experiments ! This is mainly based on optimisation at ears ! Nothing scientific. But simple to experiment. YMMV.
I believe only one A123 cell at 3.4 V even with a 3.6 V load should give better results (drop the outputt voltage to 3.2 V on the À les reg) for some reasons ! If you make the test, I will be interested by what you find on this particular board as the PS is feeding more than a clock !
regards
Both has to be imho tried, firstly to hear how is sensible your DAC to the supply !, of course the 10 cm wire has to be twisted ! Also different thickness gave me different result ?! I assume only thr ground return thickness is important, but I didn't make more experiments ! This is mainly based on optimisation at ears ! Nothing scientific. But simple to experiment. YMMV.
I believe only one A123 cell at 3.4 V even with a 3.6 V load should give better results (drop the outputt voltage to 3.2 V on the À les reg) for some reasons ! If you make the test, I will be interested by what you find on this particular board as the PS is feeding more than a clock !
regards
Anyway here this is the longer length which gave me the best result !
But maybe the length was not the only factor ! (Traffo farer ??)
Your scope is not hearing music... to say it differently : the pleasure you have with each side of your brain at looking at the result MAY BE different sometimes ! You can enjoy good result at looking the scope, but is it the good thing which is measured ? One thing is sure ! my ears decide, because this is the goal ! (left brain ?) !
Did you make the diesel/gazoline setup at Volkswagen ? Lol !
IIRC, Àles showed scopes results : not bad ! But my memorie is the one of a red fisch! (that's why I listen to music... same album for 30 years )
!But maybe the length was not the only factor ! (Traffo farer ??)
Your scope is not hearing music... to say it differently : the pleasure you have with each side of your brain at looking at the result MAY BE different sometimes ! You can enjoy good result at looking the scope, but is it the good thing which is measured ? One thing is sure ! my ears decide, because this is the goal ! (left brain ?) !
Did you make the diesel/gazoline setup at Volkswagen ? Lol !
IIRC, Àles showed scopes results : not bad ! But my memorie is the one of a red fisch! (that's why I listen to music... same album for 30 years
)
Last edited:
Let's not start the debate again that listening is the best way to evaluate compared to measuring. Please. It is alway started by people who don't have measuring equipment
Still, 2 or 10 cm wire making a difference is something quite strange which might indicate things going on. The scope is not listening but looking at DC. Clean DC I hope.
Still, 2 or 10 cm wire making a difference is something quite strange which might indicate things going on. The scope is not listening but looking at DC. Clean DC I hope.
What to measure is the question and find the correlation (coloration?) !
I have also better result with the R-Core than the A123 cell : tested (measured by ears) with the same length of wire taking into acount the width of the cell !) ! (more slam & clearness than with the quiet LiPoFe4 cell ! I couldn't believe it... so ...)
I have also better result with the R-Core than the A123 cell : tested (measured by ears) with the same length of wire taking into acount the width of the cell !) ! (more slam & clearness than with the quiet LiPoFe4 cell ! I couldn't believe it... so ...)
... near from you ? One sourced at Singapor or maybe this one ? : DIYCLUB
Mine is bought from Selectronic France (6v/30VA), normal quality (sourced in India if what they said are true !
The best are from Japan. (made in China?)
J-P made a supply with the same type of traffo, but not the same reg chip than Àles Mavrica fellow if I remember ! What I like with the T7A chip Àles choosed is the voltage programation just by straping with a drop of solder !
I sourced my A123 cell from a battery sailer in France... hopping they are not fake ! Voltage is correct but as you know the A123 LiPoFe4 has the lowest ESR ! That's also why they are copied with fake ! So was my experiment reliable ? Or Something else is oscillating elswhere on the DAC ? Who knows ?
But the level of clearness I have is going far further than a ES9023 chip with spidf inputt... (sorry didn't test it with an I2S direct source or a FIFO either!)... but the reg is supplying a Master Clock, not the DAC chip itself. So best to follow SoundCheck experiments as he owns the DAC tested in this topic.
IanCanada seems to prefer an A123 cell to a torroid transformer though !
Mine is bought from Selectronic France (6v/30VA), normal quality (sourced in India if what they said are true !
The best are from Japan. (made in China?)
J-P made a supply with the same type of traffo, but not the same reg chip than Àles Mavrica fellow if I remember ! What I like with the T7A chip Àles choosed is the voltage programation just by straping with a drop of solder !
I sourced my A123 cell from a battery sailer in France... hopping they are not fake ! Voltage is correct but as you know the A123 LiPoFe4 has the lowest ESR ! That's also why they are copied with fake ! So was my experiment reliable ? Or Something else is oscillating elswhere on the DAC ? Who knows ?
But the level of clearness I have is going far further than a ES9023 chip with spidf inputt... (sorry didn't test it with an I2S direct source or a FIFO either!)... but the reg is supplying a Master Clock, not the DAC chip itself. So best to follow SoundCheck experiments as he owns the DAC tested in this topic.
IanCanada seems to prefer an A123 cell to a torroid transformer though !
R-Core transformers from Jingru seem to be of good quality, here's the link on TaoBao:
https://world.taobao.com/item/524355674309.htm?spm=a312a.7700714.0.0.35wTY7#detail
https://world.taobao.com/item/524355674309.htm?spm=a312a.7700714.0.0.35wTY7#detail
Eh !? There are good Rcore transformers and there are less good Rcore transformers. The japanese are often very good (also their OEM ones made in China). The indian types I have seen are less good.
Kitamura Kiden is the inventor of the Rcore, these are the best.
http://kitamura-kiden.co.jp
http://www.pnxcorp.co.jp
* The wire you mention is the earth screen. It MUST be connected to PE (just as the chassis) and definitely NOT to audio GND !!!! Please do not call PE (Protective Earth) GND (Ground). GND is the 0V in your device. GND is at the secondary side of the PSU. PE is at the mains side and it is for safety when a wire comes loos or other failures that can be lethal. In some countries audio GND is connected to PE but it leads to ground loops. In many devices audio GND is either floating or "lifted" in a safe way. Using both terms for the wrong item makes the already "hard enough to understand" matters even more hard to understand
Kitamura Kiden is the inventor of the Rcore, these are the best.
http://kitamura-kiden.co.jp
http://www.pnxcorp.co.jp
* The wire you mention is the earth screen. It MUST be connected to PE (just as the chassis) and definitely NOT to audio GND !!!! Please do not call PE (Protective Earth) GND (Ground). GND is the 0V in your device. GND is at the secondary side of the PSU. PE is at the mains side and it is for safety when a wire comes loos or other failures that can be lethal. In some countries audio GND is connected to PE but it leads to ground loops. In many devices audio GND is either floating or "lifted" in a safe way. Using both terms for the wrong item makes the already "hard enough to understand" matters even more hard to understand
Last edited:
Yes I assume the one you mention are better, having also one from a Sony SACD player : very good quality.
What I think is most of them (China, India) are good enough & better in relation to the others classic traffo : torroid, etc !
Thanks for the tip about the screening (yes, sorry for the English, I meaned Earth, thanks for correction.. ) ! When I used tha A123 LiPoFe4 with and without ferrite bid I was surprised to have a less good sound (muddier and less clear) than with a R-core (feeding the Clock II/Laptech 957 from Ian) ! Is it a correct conclusion, I don't know, but the experiment was at iso perimeter : i.e. : with ales TS7A reg & Ian Clock 2 + same length of wires each time !
For what it worths...
P.S : byt the way is it the screen between primary and secondary or an external shield around the windings ?
What I think is most of them (China, India) are good enough & better in relation to the others classic traffo : torroid, etc !
Thanks for the tip about the screening (yes, sorry for the English, I meaned Earth, thanks for correction.. ) ! When I used tha A123 LiPoFe4 with and without ferrite bid I was surprised to have a less good sound (muddier and less clear) than with a R-core (feeding the Clock II/Laptech 957 from Ian) ! Is it a correct conclusion, I don't know, but the experiment was at iso perimeter : i.e. : with ales TS7A reg & Ian Clock 2 + same length of wires each time !
For what it worths...
P.S : byt the way is it the screen between primary and secondary or an external shield around the windings ?
Last edited:
Just an off-topic question -- since have the "linear PS" experts around here.
Before I might waste time and money on the R-Core.
I have a Talema 2*7V in the workshop. These are not the worst transfomers on the planet. I can easily build a supply with Ales regulator, which I also still have around, in 10 minutes.
The question is what to look for to make it right from the beginning !?!?
How about mains filter? I have some Schaffners around.
Shall I parallel the two secondaries??
Any hints are welcome.
The PS would just feed the AVCC 3.6V of the Sabre DAC.
THX
Before I might waste time and money on the R-Core.
I have a Talema 2*7V in the workshop. These are not the worst transfomers on the planet. I can easily build a supply with Ales regulator, which I also still have around, in 10 minutes.
The question is what to look for to make it right from the beginning !?!?
How about mains filter? I have some Schaffners around.
Shall I parallel the two secondaries??
Any hints are welcome.
The PS would just feed the AVCC 3.6V of the Sabre DAC.
THX
What VA is the transformer? I used a couple of similar (7v + 7v secondary at almost 4A!) Amveco 50VA transformers in my latest build.
I configured each in a 2-diode full-wave configuration, putting the 2 secondaries together to create a center-tapped secondary, putting a diode in series with the outer secondary ends to the + of the filter cap, and putting the center tap to the - of the filter cap. Some of my audio friends prefer that configuration to a full-wave bridge.
I used IXYS DSEP30-06A (DSEP30-06A-ND) 30A diodes. Overkill, but I know they sound good AND they will NEVER fail in this application.
I used a 47,000uf/16v Jensen 4-pole capacitor as the filter cap. They are the best, but expensive. Consider a pair to a quad of United Chemi-Con 5600uf/16v (565-3876-ND) in parallel as a lower cost alternative (these are in the OPC paralleled LT3042 regulator boards I used on my modified HFBD+P).
I put a single Epcos (TDK) 2.2uf/450v polyprop (495-4717-ND) across the AC input for filtering. I might use a more sophisticated filter for a fully-digital circuit like the R-Pi, but in a mixed circuit like the ES9023, I find minimal filtering to be better.
And I use a John Swenson-recommended 330R in series with a .022uf capacitor transformer secondary ringing snubber across each secondary.
Mine are wired point-to-point, ugly, but minimizes wire length AND solder joints.
They put out about 12v unloaded... and with that currrent output capability and the type of loads for applications such as this, they don't sag much at all.
BTW, I agree that a good R-Core would be better, but I am just trying them out and don't have any sources I could recommend yet.
Greg in Mississippi
P.S. the ....-ND numbers are the Digikey part numbers.
I configured each in a 2-diode full-wave configuration, putting the 2 secondaries together to create a center-tapped secondary, putting a diode in series with the outer secondary ends to the + of the filter cap, and putting the center tap to the - of the filter cap. Some of my audio friends prefer that configuration to a full-wave bridge.
I used IXYS DSEP30-06A (DSEP30-06A-ND) 30A diodes. Overkill, but I know they sound good AND they will NEVER fail in this application.
I used a 47,000uf/16v Jensen 4-pole capacitor as the filter cap. They are the best, but expensive. Consider a pair to a quad of United Chemi-Con 5600uf/16v (565-3876-ND) in parallel as a lower cost alternative (these are in the OPC paralleled LT3042 regulator boards I used on my modified HFBD+P).
I put a single Epcos (TDK) 2.2uf/450v polyprop (495-4717-ND) across the AC input for filtering. I might use a more sophisticated filter for a fully-digital circuit like the R-Pi, but in a mixed circuit like the ES9023, I find minimal filtering to be better.
And I use a John Swenson-recommended 330R in series with a .022uf capacitor transformer secondary ringing snubber across each secondary.
Mine are wired point-to-point, ugly, but minimizes wire length AND solder joints.
They put out about 12v unloaded... and with that currrent output capability and the type of loads for applications such as this, they don't sag much at all.
BTW, I agree that a good R-Core would be better, but I am just trying them out and don't have any sources I could recommend yet.
Greg in Mississippi
P.S. the ....-ND numbers are the Digikey part numbers.
Last edited:
Greg, et al,
Pretty sage advice based on the comparisons I've made over the years. Only exception I might make is the R-core transformer. I mean, I can understand how someone might "like" or "prefer" it; but, I've never seen anything to technically explain it's superiority to my satisfaction.
Pretty sage advice based on the comparisons I've made over the years. Only exception I might make is the R-core transformer. I mean, I can understand how someone might "like" or "prefer" it; but, I've never seen anything to technically explain it's superiority to my satisfaction.
Hi.
Update.
I took the batteries out again.
And changed the ground-routing a little.
It can stay like this. I think it is even better then before.
I also gave up on the linear supply idea.
On reason is that the LDO @ 3.6V has to burn too much excess voltage
while using a (my) 7V transfomer. At 30mA DAC-load that shouldn't be such a serious problem though.
Still. I rather try the 4.5V from the Meanwell.
Bottom line.
I'll now focus a bit on getting the ($9) Meanwell RS-15-5 shaped up.
First I need to figure out what's wrong with it on comparison to the iFi.
Update.
I took the batteries out again.
And changed the ground-routing a little.
It can stay like this. I think it is even better then before.
I also gave up on the linear supply idea.
On reason is that the LDO @ 3.6V has to burn too much excess voltage
while using a (my) 7V transfomer. At 30mA DAC-load that shouldn't be such a serious problem though.
Still. I rather try the 4.5V from the Meanwell.
Bottom line.
I'll now focus a bit on getting the ($9) Meanwell RS-15-5 shaped up.
First I need to figure out what's wrong with it on comparison to the iFi.
Hi soundcheck,
in the meantime i again tried Batteries/Accus and Meanwell PSU.
But still i like the LDO (ANTX old PCB-Version) better. As transformer is use
an talema torriod 2*7VAc followed by schotky-bridge, CRC and then die LDO
boards. After those i use LR-Filter and then one path to the Mambo and from
the other 7VAc to the RasPi.
what did you modify at the groundplane?
Greetings
ulf
in the meantime i again tried Batteries/Accus and Meanwell PSU.
But still i like the LDO (ANTX old PCB-Version) better. As transformer is use
an talema torriod 2*7VAc followed by schotky-bridge, CRC and then die LDO
boards. After those i use LR-Filter and then one path to the Mambo and from
the other 7VAc to the RasPi.
what did you modify at the groundplane?
Greetings
ulf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.