Just for curiosity I got one for myself. Prepared a small attachment with I/V and AVCC's on top of it. In current mode dac is playing a way smoother, background goes a bit more forward. The filter I made is rolling off a tiny bit too early I hade to rework it after some listening tests.
So far so good
Hi borys,
Thank you for sharing. Interesting approach you took there. No surprise it sounds better. Looks like you decided to pretty much completely go your own way seemingly without any of the advice provided by ESS. I know with more work you can get get to much better sound quality if you care to, but for some people the fun may be in just doing one's own experiments more freely. Only one question for you, not sure what you meant by saying the dac is smoother, which I understand, but then say, 'background goes more forward,' which seems less common to me. Do you mean the small details are more forward? IME, any forwardness at all is one sign of a bit of remaining distortion. But, I don't usually hear people describe it exactly as you did.
Last edited:
It helps a lot to be able to verify expected signals are making it to the dac chip. If so, then the MCU might be putting the dac chip into an incompatible state. Lot of work to get in and check that, if it were suspected. If not, then one would look back towards the USB to I2S board, and connections between it and the dac board. Pretty hard to do at some level without being able to see signals. If you are going to be doing much of this kind of thing, eventually a scope would be a very appropriate thing to pick up, even a low cost used one. Not sure what else to say.
Markw4
I would say the sound is more neutral/natural in current mode, the top octave have a less ''sparkle/harshness''.
The background is more ''readable'' for me.
I have made the board from what I had in stock, as I did not have any 3,3V regulator so I have biased two leds in series (3,52V) with constant current source as a reference and made 1/2 Vref and analog VCC of it.
I think it is worth of spending a bit more time on this project.
Will have to read this thread from the first to last post.
Regards
I would say the sound is more neutral/natural in current mode, the top octave have a less ''sparkle/harshness''.
The background is more ''readable'' for me.
I have made the board from what I had in stock, as I did not have any 3,3V regulator so I have biased two leds in series (3,52V) with constant current source as a reference and made 1/2 Vref and analog VCC of it.
I think it is worth of spending a bit more time on this project.
Will have to read this thread from the first to last post.
Regards
The average propagation velocity of the air in an average listening environment at LF is far worse than that. Add an amp/speaker/microphone to the jitter measurement loop.
Pico second signal correlated jitter, lets see at ~1000m/sec that's holding your speakers to 1nm displacement while they play.
Scott,
I sure don't hear it as any kind of time difference. Just something that sounds like low-level harmonic distortion, in part (a bit artificially bright). Including in mono, nothing to do with stereo. The other thing I notice is besides what I mentioned about low level reverb tails, is more accurate reproduction of low level cymbal tail details. I don't know why those things seem to be correlated with measured jitter. No idea. I would love to figure it out though.
Just thinking out loud, we are upsampling to 11.2MHz. If jitter as a fraction of sample rate has some influence on the amplitude vs time values of individual samples, maybe it affects reconstruction accuracy. Seems to me I have seen someone make a more detailed case for something like that, have to see if I can find it.
As always, interested parties would be welcome to meet over at my place for listening training and to see if anyone else hears it like I hear it. Someone, or various people can take turns making everyone else blinded. IMHO, having the Benchmark AHB2 power amp here helps a lot to hear the small details clearly.
EDIT: IIRC, I think somewhere or other at some point ESS said something to the effect that it is better to set DPLL bandwidth to the lowest stable setting for best sound quality. So far as I can tell, the ability to get stable operation at lower DPLL bandwidth settings is a function of incoming jitter level. The lower the incoming jitter, the more stable operation seems to be a lower bandwidth settings. You might ask your friends over at ESS what is going on there, they don't tell people like me that stuff. That might result in satisfaction of your curiosity about jitter effects much more quickly than anything else.
I sure don't hear it as any kind of time difference. Just something that sounds like low-level harmonic distortion, in part (a bit artificially bright). Including in mono, nothing to do with stereo. The other thing I notice is besides what I mentioned about low level reverb tails, is more accurate reproduction of low level cymbal tail details. I don't know why those things seem to be correlated with measured jitter. No idea. I would love to figure it out though.
Just thinking out loud, we are upsampling to 11.2MHz. If jitter as a fraction of sample rate has some influence on the amplitude vs time values of individual samples, maybe it affects reconstruction accuracy. Seems to me I have seen someone make a more detailed case for something like that, have to see if I can find it.
As always, interested parties would be welcome to meet over at my place for listening training and to see if anyone else hears it like I hear it. Someone, or various people can take turns making everyone else blinded. IMHO, having the Benchmark AHB2 power amp here helps a lot to hear the small details clearly.
EDIT: IIRC, I think somewhere or other at some point ESS said something to the effect that it is better to set DPLL bandwidth to the lowest stable setting for best sound quality. So far as I can tell, the ability to get stable operation at lower DPLL bandwidth settings is a function of incoming jitter level. The lower the incoming jitter, the more stable operation seems to be a lower bandwidth settings. You might ask your friends over at ESS what is going on there, they don't tell people like me that stuff. That might result in satisfaction of your curiosity about jitter effects much more quickly than anything else.
Last edited:
Hi borys,
Unless you really want to read the whole thread, I posted some summary information recently with some links to other bits of possibly useful information. There is also a downloads page at ESS website where there are a few old documents including an application note with tips on getting the best sound out out Sabre dacs. It is here: ESS Technology :: Downloads
The app not is this one:
http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
The way they developed Vref is only shown in an evaluation board schematic on the downloads page.
Some summary posts of mine are here:
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-320.html#post5609058
and here:
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-301.html#post5577671
Schematics where posted last by MrSlim in post #3003.
Since there is audio when the microcontroller is off im assuming the signals are getting to the DAC and its in incompatable state like you mention, so what about taking control of the DAC chips? the audio of the volume cant be controlled when the MC is off because the MC controls the volume, its at fixed very low level, this is really the only thing in the way of having basic functioning dac (but it probably isnt simple with the lossless digital volume these 9038 dac chips use).
Sorry for the loose questions, its hard to find info on this subject.
Maybe good to look at information in this thread. What we have is spread out over several posts. After that you can see if you have more questions:
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-72.html#post5408825
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-78.html#post5412858
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-142.html#post5452170
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-288.html#post5565005
https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-280.html#post5557160
redjr,
Do you by any chance have custom page lengths set for the forum? I ask because that can cause post links to fail. Post #3301 is on page 331 here.
The post numbers should be (reverse order):
2792
2873
1412
780
716
Just noticed bit more in post #2862
ES9038Q2M Board
Do you by any chance have custom page lengths set for the forum? I ask because that can cause post links to fail. Post #3301 is on page 331 here.
The post numbers should be (reverse order):
2792
2873
1412
780
716
Just noticed bit more in post #2862
ES9038Q2M Board
Last edited:
While working on the modded dac today, I decided to remove the bank of film caps and try them with Katana and the modest linear power supply setup to see if any of the graininess I reported previously could be smoothed out. My initial impression is that it seems to be helping. I will listen again tomorrow with and without the film caps to see if my impression still holds then. (As a reminder, a previous test using 10,000uf, .053 ohm electrolytics on the +-15v rails didn't really seem to help at all. They might help some though if used in parallel with the film caps, a combination which I haven't tried so far with Katana.)
The film cap bank is made out of whatever large value film caps I could find in stock. There are some Wima 33uf, 22uf, and 10uf caps of same types I used for LTC6655 output filtering and AVCC opamp output filtering. There is about 110uf for each of the 15v power rails (220uf in total).
Using that many film caps in those sizes would probably add a fair amount to the cost of the dac, but the linear power supply I am using is very modest indeed. It is one of these with a few simple mods: Amazon.com: Elenco AC/DC Triple Output Power Supply Kit: Electronics
The mods are that all the rectifier diodes where replaced with Hexfred ultra soft recovery rectifiers, and the + 15v LM317 regulator was replaced with an LT1086. That's it.
Total cost of the film caps might be a little more than the cost of the linear supply.
The other thing, of course, is that the linear supply is powered from a very good AC line power conditioner. Don't know how much difference that is making at the moment, but it can help a lot in some cases.
The RPi and that half of the Katana isolator are powered by an IFI 5v wall wart supply, just for the record.
The film cap bank is made out of whatever large value film caps I could find in stock. There are some Wima 33uf, 22uf, and 10uf caps of same types I used for LTC6655 output filtering and AVCC opamp output filtering. There is about 110uf for each of the 15v power rails (220uf in total).
Using that many film caps in those sizes would probably add a fair amount to the cost of the dac, but the linear power supply I am using is very modest indeed. It is one of these with a few simple mods: Amazon.com: Elenco AC/DC Triple Output Power Supply Kit: Electronics
The mods are that all the rectifier diodes where replaced with Hexfred ultra soft recovery rectifiers, and the + 15v LM317 regulator was replaced with an LT1086. That's it.
Total cost of the film caps might be a little more than the cost of the linear supply.
The other thing, of course, is that the linear supply is powered from a very good AC line power conditioner. Don't know how much difference that is making at the moment, but it can help a lot in some cases.
The RPi and that half of the Katana isolator are powered by an IFI 5v wall wart supply, just for the record.
Last edited:
Katana Update - Good News
Okay, I listened to Katana again with the film caps. The graininess is gone. It's the best I have ever heard Katana sound. The only thing I notice is maybe some slight thickening in the midrange, a tiny bit of muddiness that kind of sounds like it may be some low level 2nd harmonic distortion characteristic of the discrete output stage. Don't know for sure. But, aside from that minor little effect the sound is extremely detailed, smooth, and very pleasant to listen to. It is good news for people wanting a great RPi dac at very reasonable cost. I think this is it.
However, in a couple of forums some people have started asking it Katana can work with a USB input, or maybe SPDIF. In other words, some people still want a general purpose dac, not just an RPi music player. They just want that general purpose dac to sound great too, and be low-ish cost. Katana isn't that type of dac, not yet anyway. The modded dac we are working on is a general purpose dac, but we need to get that jitter and DPLL bandwidth stuff worked out. That's what I was doing some work on earlier today, but no listening tests to report yet. Its still on the bench getting some little test mods done to try out later.
Okay, I listened to Katana again with the film caps. The graininess is gone. It's the best I have ever heard Katana sound. The only thing I notice is maybe some slight thickening in the midrange, a tiny bit of muddiness that kind of sounds like it may be some low level 2nd harmonic distortion characteristic of the discrete output stage. Don't know for sure. But, aside from that minor little effect the sound is extremely detailed, smooth, and very pleasant to listen to. It is good news for people wanting a great RPi dac at very reasonable cost. I think this is it.
However, in a couple of forums some people have started asking it Katana can work with a USB input, or maybe SPDIF. In other words, some people still want a general purpose dac, not just an RPi music player. They just want that general purpose dac to sound great too, and be low-ish cost. Katana isn't that type of dac, not yet anyway. The modded dac we are working on is a general purpose dac, but we need to get that jitter and DPLL bandwidth stuff worked out. That's what I was doing some work on earlier today, but no listening tests to report yet. Its still on the bench getting some little test mods done to try out later.
Last edited:
Hi Mark
its a pleasure being in the same league as modded DAC. I know all the efforts that you and community have poured into those boards.
We are developing very good dual rail LPS , but please allow some time for it. Since the beginning , we wanted Katana to become an alternative to hi end DACs and with good PSUs I think its possible to get real hi end sound from a 350$ streamer.
its a pleasure being in the same league as modded DAC. I know all the efforts that you and community have poured into those boards.
We are developing very good dual rail LPS , but please allow some time for it. Since the beginning , we wanted Katana to become an alternative to hi end DACs and with good PSUs I think its possible to get real hi end sound from a 350$ streamer.
It is still not as good as DAC-3. How close? Not sure how to answer that. Let me say this: if people have one of our Q2M dacs at its best they will probably have a better sounding dac than anyone they know, and they will be happy at how natural it sounds. That is, until they listen to DAC-3. Then they will know the sound can be clearly better. To equal DAC-3 requires using more technology than we do now. For one thing there is external interpolation filtering. Another thing is noise, and noise floor modulation. You can't completely get there from here where we are with our Q2M dacs.
The other point I want to make and that I will send by email is that I don't think power supply noise that you are working on is necessarily the right problem to be focusing on. It is probably power supply output impedance for step load current changes over the entire audio band that may be limiting sound quality. That's what the film caps help with more than helping with noise, I am thinking. A Jung Super Regulator power supply as Mikett has suggested may well be an appropriate solution, or maybe just lots of film caps. Voltage regulator low noise, by itself, is something else, not the same as output impedance for step loads in the presence of filter capacitor dielectric absorption (dynamic output impedance). Maybe this will help to explain: https://www.calex.com/pdf/3power_impedance.pdf
The other point I want to make and that I will send by email is that I don't think power supply noise that you are working on is necessarily the right problem to be focusing on. It is probably power supply output impedance for step load current changes over the entire audio band that may be limiting sound quality. That's what the film caps help with more than helping with noise, I am thinking. A Jung Super Regulator power supply as Mikett has suggested may well be an appropriate solution, or maybe just lots of film caps. Voltage regulator low noise, by itself, is something else, not the same as output impedance for step loads in the presence of filter capacitor dielectric absorption (dynamic output impedance). Maybe this will help to explain: https://www.calex.com/pdf/3power_impedance.pdf
Last edited:
I took a picture of the Katana power supply setup. Its still early in the morning here and kind of dark in that area. Didn't want to use a flash, so this is what I have.
EDIT: Unfortunately, I only have one block of film caps, so can't compare Katana and the modded dac directly right now. If I order more caps, maybe I could next week.
EDIT: Unfortunately, I only have one block of film caps, so can't compare Katana and the modded dac directly right now. If I order more caps, maybe I could next week.
Attachments
Last edited:
thanks, its probably too advanced for me. There is small chance it could be the old USB interface still communicating with the MC, even causing it to malfunction in its broken state since only IS2 pins were cut, cutting old interface power pins is the last hope of an easy fix.
On a seperate note about power supplies, Im using LT3045 regulator to provide 5V to multiple 3.3V internal regulators in the DAC, would it be a bad idea to bypass these internal regulators and use the LT3045 set to 3.3V to power everything?
In normal circumstances these internal regulators would be receiving power from noisy SMPS, like a PC, but with the LT3045 these internal regulators are probably adding noise. there is multiple though, so im not sure if they are providing some isolation between different parts of the DAC chips.
I agree , on PSU not only noise but impedance seems to be very important.
Regarding q2m , yeah that makes sense. Yes those benchmark have clever digital filtering but in my opinion plain old school hardware tweaks (noise jitter reduction) can bring it on an equal footing. We have a few tricks up our sleeve.. since we have a discreete opamp we can tweak a few parameters. Will keep you updated.