continue to work on improving the set up. So I realized that I was mistaken and that there are 3 pairs of wires but actually 1 rail. However with only one of my 3 supplies based on the LM317, I get 100mv ripple instead of 20mv.
So I may try a LM350. I am trying a Mic29300-3.3 but I am having issues with the batch I have. Pin 1 is supposed to be Input, pin 2 the ground and pin 3 the output. At this time whatever voltage I put on pin 1 I get it on pin 3 without regulation.
Ideally I would need a very quiet 3.3v supply that can output 3 Amps peak.
Also I am trying a to put a new clock instead of the stock crystal.
Bellow you can see the stock wave:
and the wave from the TCXO (0.5ppm model from connor winfield D75F):
I must say that I am not satisfied with this wave, I would like it to be cleaner. I removed all caps at the input of the Marvell 88F series processor, but still not very nice.
Sound wise I can hear gains, more details, more "air" but also a beginning of harshness on the loud female voices.
Last but not least, I looked at the spdif signal and unloaded it looks near perfection:
with the load, i.e connected to the DAC (AES/EBU input) it looks like this:
Note I removed the Lundal impedance transformer as it made the wave worse.
What could I use to keep the perfect wave? 2 transistors? Any schematic?
Any input welcome.
Note: In my latest investigations I found that the Marvell processor is connected to a DAC from CIRRUS Logic CS4344 via I2S.
So I may try a LM350. I am trying a Mic29300-3.3 but I am having issues with the batch I have. Pin 1 is supposed to be Input, pin 2 the ground and pin 3 the output. At this time whatever voltage I put on pin 1 I get it on pin 3 without regulation.

Ideally I would need a very quiet 3.3v supply that can output 3 Amps peak.
Also I am trying a to put a new clock instead of the stock crystal.
Bellow you can see the stock wave:
An externally hosted image should be here but it was not working when we last tested it.
and the wave from the TCXO (0.5ppm model from connor winfield D75F):
An externally hosted image should be here but it was not working when we last tested it.
I must say that I am not satisfied with this wave, I would like it to be cleaner. I removed all caps at the input of the Marvell 88F series processor, but still not very nice.
Sound wise I can hear gains, more details, more "air" but also a beginning of harshness on the loud female voices.
Last but not least, I looked at the spdif signal and unloaded it looks near perfection:
An externally hosted image should be here but it was not working when we last tested it.
with the load, i.e connected to the DAC (AES/EBU input) it looks like this:
An externally hosted image should be here but it was not working when we last tested it.
Note I removed the Lundal impedance transformer as it made the wave worse.
What could I use to keep the perfect wave? 2 transistors? Any schematic?
Any input welcome.
Note: In my latest investigations I found that the Marvell processor is connected to a DAC from CIRRUS Logic CS4344 via I2S.
Last edited:
I recommend using linear tech regulators, the LT317/337 are drop in replacements that are a significant upgrade, also for 317 the LT1085 and several others if you want the numbers i'll have a look for you.
cdfr, thank you for the information.
I've heard many times that 2705 is the limier itself, if you want to get digital out.
Did you change oscillator for N, or just changed power supply?
I'm going to do the same, however I don't know which oscillator to get and where from, also I still have hope to get good results with good short glass optical cable. Do you think it is possible? My DAC has only SPDIF, optical and coax.
For now CD-transport with coax output sounds little better than basic AE g. I would like to get them equal at least...
Do you think it is definitely necessary to change it go AE n version, before mod?
Thanks!
I've heard many times that 2705 is the limier itself, if you want to get digital out.
Did you change oscillator for N, or just changed power supply?
I'm going to do the same, however I don't know which oscillator to get and where from, also I still have hope to get good results with good short glass optical cable. Do you think it is possible? My DAC has only SPDIF, optical and coax.
For now CD-transport with coax output sounds little better than basic AE g. I would like to get them equal at least...
Do you think it is definitely necessary to change it go AE n version, before mod?
Thanks!
Last edited:
Vitalbenson,
The AE G modified can be quite good, but yes I believe that the limiting factor is the USB transmission and or pcm2705 even in spdif!
The AE N is a better start. Even the stock version sounds quite decent. The PSU change and if you want to use it as a drive the removal of the toslink gives significant improvements in my experience.
I am now experimenting with a clock to see if I can go further, this is what I relate in the post.
cheers.
The AE G modified can be quite good, but yes I believe that the limiting factor is the USB transmission and or pcm2705 even in spdif!
The AE N is a better start. Even the stock version sounds quite decent. The PSU change and if you want to use it as a drive the removal of the toslink gives significant improvements in my experience.
I am now experimenting with a clock to see if I can go further, this is what I relate in the post.
cheers.
Thank you!
What DAC, finally, you found in AE-N? I would like to see its datasheet to check how to get Coax out of it...
What DAC, finally, you found in AE-N? I would like to see its datasheet to check how to get Coax out of it...
cdfr, I went through homecinema-fr, and it is still unclear: finally, should it be 3 separate power sources, or you decided to come back to a single one, for all three inputs soldered together?
Thank you!
Thank you!
In my post on homecinema I show where the DAC is located. It is I think a CIRRUS Logic CS4344.
For the power supply, I believe that I was mistaken previously and that there is only 1 rail accessible easily. So my 3 rail power supply is not needed. However the design I use is I belive too limited in current, so I am looking for something that will spply up to 3 Amps.
That's why I am interested in Qusp post.
For the power supply, I believe that I was mistaken previously and that there is only 1 rail accessible easily. So my 3 rail power supply is not needed. However the design I use is I belive too limited in current, so I am looking for something that will spply up to 3 Amps.
That's why I am interested in Qusp post.
I see! Fortunately, CS doesn't give away digital out, but Marvell...
Thank you for your answer! I'm going to buy parts for power supply tomorrow and proceed.
Just have tested both N and G, both are stock. Even short test shown that N noticeable better (longer echoes, more aggressive, more active and solid on bass). I use optics, for the moment. Going to open it right now and get coax output to see how different is it vs optics. By the way, you removed jack module, but what about resistor that supposed to be before it, before the LED? I noticed you have took signal out straight from the module's contact, right?
As for power: one LM317 allow 1.5A peak. Did you measure real consumption? Don't know why, it seems the unit eat less than that. If it's more, maybe LT1085 could help? It give away up to 3A, with less noise on low freqs (5 db less than 317), and 5 db more on higher freqs. And it's not as expensive as LT1083 (which is the same but 7A current).
Thank you for your answer! I'm going to buy parts for power supply tomorrow and proceed.
Just have tested both N and G, both are stock. Even short test shown that N noticeable better (longer echoes, more aggressive, more active and solid on bass). I use optics, for the moment. Going to open it right now and get coax output to see how different is it vs optics. By the way, you removed jack module, but what about resistor that supposed to be before it, before the LED? I noticed you have took signal out straight from the module's contact, right?
As for power: one LM317 allow 1.5A peak. Did you measure real consumption? Don't know why, it seems the unit eat less than that. If it's more, maybe LT1085 could help? It give away up to 3A, with less noise on low freqs (5 db less than 317), and 5 db more on higher freqs. And it's not as expensive as LT1083 (which is the same but 7A current).
P.S. Stock PSU has two separate lines! If you measure resistance of the stock PSU you will see that g-g-g (ground is united) — 3.3-3.3-3.3 are done like this: g-g-g — 3.3(separate)-3.3-3.3(both united, 0 Ohms between).
Voltage itself is 3,37 - 3,38-3.38. About 20 oHms is between contacts 2 and 3, in accordance to your photo at homecinema.
UPD. Sorry, if you measure it at the AE, it's all equal. Zero Ohms between all 3.3V inputs. You were right: only single power supply required.
Voltage itself is 3,37 - 3,38-3.38. About 20 oHms is between contacts 2 and 3, in accordance to your photo at homecinema.
UPD. Sorry, if you measure it at the AE, it's all equal. Zero Ohms between all 3.3V inputs. You were right: only single power supply required.
Last edited:
Let me share first results.
LM317 is enough: unit takes just about 0.9A! I used Li-Ion battery 7.4v 2200 mAh.
I used optics, not coaxial yet. This is actually a very mid level cable, not even glass, and I thought this is what will prevent me from hearing real difference, but...
I switched power while testing, from original PSU to the "silent" one, few times, listening to the same tracks. Those tracks are awesome recorded, I know them well as often use while test something.
So what's changed. With clear supply I can easily hear that sound stage became more readable. More volume, more space, more air. Some sharp "end-notes" on saxophones appeared, same, as it was with direct USB connection to the DAC. They are missing with stock PSU - just absent. Vocals has longer echoes, which are so pleasant to hear, which - also - were missing with stock power.
So I would like to thank you so much for your help in this!
Now I'm waiting for glass Silflex cable, before that, I would like to try coaxial connection.
Do you think is it absolutely necessary to remove jack module, or I can leave it there and connect to its pins? Or LED will take all the power?
Also, can short good quality RF-antenna cable be used as a coax cable?
Thanks!
LM317 is enough: unit takes just about 0.9A! I used Li-Ion battery 7.4v 2200 mAh.
I used optics, not coaxial yet. This is actually a very mid level cable, not even glass, and I thought this is what will prevent me from hearing real difference, but...
I switched power while testing, from original PSU to the "silent" one, few times, listening to the same tracks. Those tracks are awesome recorded, I know them well as often use while test something.
So what's changed. With clear supply I can easily hear that sound stage became more readable. More volume, more space, more air. Some sharp "end-notes" on saxophones appeared, same, as it was with direct USB connection to the DAC. They are missing with stock PSU - just absent. Vocals has longer echoes, which are so pleasant to hear, which - also - were missing with stock power.
So I would like to thank you so much for your help in this!
Now I'm waiting for glass Silflex cable, before that, I would like to try coaxial connection.
Do you think is it absolutely necessary to remove jack module, or I can leave it there and connect to its pins? Or LED will take all the power?
Also, can short good quality RF-antenna cable be used as a coax cable?
Thanks!
I am glad to see that you like the improvements.
Indeed on a lab power supply I read 0.8 - 0.9 Amps. Somebody though measured peaks up to 2 amps.
For the LM317, how much caps after it? Did you measure the ripple?
Also how do you plan to recharge the battery. The ideal would be to have 2 batteries wth auto-charge and auto-switch.
For the dig output, I removed the jack combo to have no load. I did pay attention if there was a component before on which to tap the the spdif. What I'd like to have is a buffer to drive the cox cable to keep the beautiful spdif coming out of the Marvell chip. May be a 74HC14?
Indeed on a lab power supply I read 0.8 - 0.9 Amps. Somebody though measured peaks up to 2 amps.
For the LM317, how much caps after it? Did you measure the ripple?
Also how do you plan to recharge the battery. The ideal would be to have 2 batteries wth auto-charge and auto-switch.
For the dig output, I removed the jack combo to have no load. I did pay attention if there was a component before on which to tap the the spdif. What I'd like to have is a buffer to drive the cox cable to keep the beautiful spdif coming out of the Marvell chip. May be a 74HC14?
At the moment I have just 4700 uF, LM, resistors and small 0,1 uF before it and after, plus 47 uF at the end.
I don't plan to use battery all the time: I'll use good transformer, add Elna low ESR caps. I can't measure ripple, but before I did same PSU for my DAC, couple, separate for digital and analogue parts, and when I power analogue part with batteries instead, I can't find any difference in sound, that means my PSU is good enough. Same will be this one.
I'll try today a very short piece of cable like 10 cm maximum, and see how it works. Removing jack combo isn't easy thing required special made remover, to have all contacts hot at one time... Did you see any SPDIF out schematics? I guess the answer is there.
Consumption: if unit take 2A, LM will shut off. You'll notice that, no prob.
I don't plan to use battery all the time: I'll use good transformer, add Elna low ESR caps. I can't measure ripple, but before I did same PSU for my DAC, couple, separate for digital and analogue parts, and when I power analogue part with batteries instead, I can't find any difference in sound, that means my PSU is good enough. Same will be this one.
I'll try today a very short piece of cable like 10 cm maximum, and see how it works. Removing jack combo isn't easy thing required special made remover, to have all contacts hot at one time... Did you see any SPDIF out schematics? I guess the answer is there.
Consumption: if unit take 2A, LM will shut off. You'll notice that, no prob.
Done, I simply connected spdif wire to the pin on jack, leaving jack at the place, via 75 oHm resistor.
Wire to DAC is short - just about 7 cm.
Coax sounded same as optics or barely better, but almost identical in my case. Now I'm going to try the unit at the very unique high resolution system of a friend of mine.
Wire to DAC is short - just about 7 cm.
Coax sounded same as optics or barely better, but almost identical in my case. Now I'm going to try the unit at the very unique high resolution system of a friend of mine.
Simple (stupid) direct coax out with resistor (and LED at the same time) didn't work good at very high level system - transport beat it. A little bit.
Wrong coaxial output scheme. Awaiting for good glass cable.
Wrong coaxial output scheme. Awaiting for good glass cable.
Yes I would assume that the led would affect the signal a little bit.
For the trace I was showing apparently the 'sloped square signal' would be due to reflection, i.e. matching impedance problem. However with the Lundahl 1574 matchig impredance it does not look better!?
For the trace I was showing apparently the 'sloped square signal' would be due to reflection, i.e. matching impedance problem. However with the Lundahl 1574 matchig impredance it does not look better!?
As long as we take signal from led contact, we take it affected by resistor before it. Is it possible to trace that resistor?
Are you talking about R522?
I think that the signal is pretty much direct from the processor, but I must say that I did not investigate.
I think that the signal is pretty much direct from the processor, but I must say that I did not investigate.
Possible solution that keep you off the transformer and signal and coax and so on.
Let's route D-out signal (ok, through resistor maybe) straight to DAC's Digital IN. No coaxial. How long can be digital wires without risk of getting noise? AE anyway can be placed close do DAC?
Let's route D-out signal (ok, through resistor maybe) straight to DAC's Digital IN. No coaxial. How long can be digital wires without risk of getting noise? AE anyway can be placed close do DAC?
- Status
- Not open for further replies.
- Home
- Source & Line
- Digital Line Level
- Music Server with Improved Apple Airport Express