Hi,
I have some isssues with my Hifiberry Digi and a Denon DA 500 DAC. The DA-500 can lock to the signal and plays via Toslink and Koax APDIF equally. The strange thing is that the DA-500 is not recognizing the CLock somehow and therefore the Frequenz indicators are not lighting up (32,44.1,48 khz).
I tried this with Volumio & Raspbian and I have a Rasberry PI 2.
I checked this with Airplay and Direkt Play or mp3's and Waf files with all 3 different Sample rates. The Frequencies are never indicated (a Denon x4000 proves though that the actual signal is according to the source files).
A usb dac, atv3, and an OLD Denon cd Player with spdif do get recognized properly.
So here my questions:
-Is the i2s clock of the PI so inacurate and the Wolfson 8804 is following this pace via PPL that the DA-500 cant recognize it completly?
-Is actually the Wolfson 8804 buffering and reclocking the signal (it has a Quartz)? And is maybe slightly to fast or slow?
-Are there specific bits set/unset within the SPDIF/PCM stream. Copy bit etc. which the DAC maybe expects?
-Are there Known Issues with the DA-500 or a yamaha YM3623 spdif receiver?
its really a strange behavior which leaves my the flavor that my setup is having issues and might not sound as good as possible.
with kind regards,
--thomas
Date
Votes
50 comments
-
HiFiBerry team Dear Thomas,
the Digi is not using the clock from the Raspberry Pi, but creates its own clock using an onboard PLL. Therefore the clock is quite stable.
We don't know about incompatibilities with your specific DAC. We have seen rare issues with older DACs (mostly from the 1990s). However there are so many DACs out there and we don't get much feedback. Therefore I can't say if there are other users with the same DAC having similar problems.
The YM3623 is a very old device. I don't know if it supports all SPDIF sample rates and bit configuration.
Best regards
Daniel -
Thomas Interesting then I have a few more questions for my understanding.
The Hifiberry Digi uses its own PLL and reads via i2S PCM stream Buffers it and reclocks the signal? What if the Pi is slower than the Hifiberry clock...? Thill now I dont fully understand what the 8804 is actually doing.
Its hard to imagine that it can be a clock issue. A quarz derived clock should be extremly stable and accurate. Especially if we are talking about the standard frequencies derived from such high frequency osscillator.
I will try within the next 2 weeks with my other Denon DA-500 to see if the behavior is the same.
I it possible to influence the possible bit configurations via parameter or property files which the driver is reading out? I would like to drill down into this issue a bit to figure out what actually is the problem.
Have you experienced in the past tolerances in the clock depending on the actual quartz used on the board? Or are these difference not measurable?
best regards,
--thomas -
HiFiBerry team Hi Thomas,
the Digi doesn't reclock the signal. The Raspberry Pi is a clock slave to the Digi. That means it does not use its own clock, but the clock from the Digi.
There are no parameters that can be adjusted by external parameters. This would need patching the wm8804 driver and recompiling the kernel.
We did not do specific clock measurements. In general the Digi provides the master clock to the DAC. Even if the clock is completely "wrong", the DAC should sync onto this clock. Usually a PLL at the input stage of a DAC is used for this.
Best regards
Daniel -
Thomas Thanks Daniel,
so actually the PI will sync onto the Hifiberry's clock and sends via i2s the stream. But through this design actually jitter should not be an visible issue with the hifiberry Digi+.
Unfortunatly it looks like that the driver of it is quiet a black box then. I think i will first try my second Denon DA 500 and after that maybe a second Digi+ from a local dealer.
I will check the source code of the Driver in a while and see if it is feasable for me to understand and to recompile it.
Are there different possible Formats for the same audi setting possible (e.g. 16bit/44.1khz)? Or is the s/pdif stream for a given format always the very same?
And I have 2 more strategic questions:
1.) Are you interested in also supporting your products for the upcomming Windows 10 version?
2.) Will you maybee provide a HAT compatible board (were the cables for camera and screed have a space to be guided through the board. Maybee with an added nice high quality Headphone amplifier, which can drive 600 Ohm Headphones... This could cover quiet a few audiophile office desk needs ;-).
thanks a lot and a nice evening,
--thomas -
HiFiBerry team Hi Thomas,
yes, the I2S clock comes from the Digi and has much less jitter than the Raspberry Pi internal clock.
At the moment, we have no plans to support Windows 10. This would need a lot of software development and I don't think there will be enough users that really want to run Windows 10 for an audio system. Note that the hardware is still a Raspberry Pi, which is much less powerful than a normal computer. I think running Windows 10 on this for fun is ok, but I don't see audio related applications.
In general our boards are HAT compatible. The cutouts or the camera are optional for a HAT. We don't see a huge need for a camera on a audio player. But if you need it, it should be possible to connect the camera and have the cable on the side of the board.
About the headphone amplifier. This is something we have looked into. However a real good headphone amplifier needs a symmetrical power supply which needs to be provided externally. This makes it hard to implement a real audiophile headphone amplifier. I think the best solution here is an external amplifier.
Best regards
Daniel -
Thomas I understand Daniel,
We will see with the time what this Mocrosft Press release will bring... On the other DAC+ Board you provide with the PCM5122 from BurrBrown I have a question. The PCM5122 is theoretically also allowing to work as Master and the Pi could also work then in slave mode as with the DIGI+.
Are you operating on the DAC+ in this mode as described above or is the DAC+ board working as Slave as the PCM5102A before?
Is this on the DAC+ board fully software configurable by the driver or do you use the so called "Hardware" configuration?
Sorry for my late posting ...
Best Regards,
--thomas -
HiFiBerry team Hi Thomas,
the DAC is working as a slave. The board is configured by software, but that doesn't mean it is fully configurable by the user. Also with the DAC+, most of the logic is handled internally in the driver.
Best regards
Daniel -
Thomas Good Morning Daniel,
I see. But this means if the driver implementation would change the same Hifiberry which works as slave could work then as a master as well or are these major configurations done on the board via the layout?
kind regards,
--thomas -
HiFiBerry team Hi Thomas,
no, this wouldn't work as the DAC doesn't have a clock generator on board. I can tell you that we are working on something, but it is too early to say more.
Best regards
Daniel -
Thomas Thanks Daniel,
Ahh thats great to have in mind... will it be again based on the 5122 or maybe another BB/TI based DAC....? The PCM1794 is also a great device ... (but I dont know if it can setup in such master scenario...)
kr,
--thomas -
HiFiBerry team I won't say anything before we have something ready.
Best regards
Daniel -
minko Hi,
I have the same problem with parasoud dac1000: it does not recognize the sampling frequency of digi+. I use volumio + raspberry pi b and digi+ + spdif. Setup with usb to spdif convertor is working fine, dac recognizes correctly sampling frequencies. Another strange issue is that I can play music at 96khz sampling rate, Parasound dac1000 does not support it, how it is possible if digi is bit perfect ? Do you have any idea ?
Thanks in advance -
HiFiBerry team Hi Minko,
your application might do some sample rate conversion. Just the fact that the Digi does not change the signal does not mean that the signal wasn't modified before.
Your Parasound DAC does not support 24bit output. As the driver uses 24bit mode for everything with more then 16bit, this might cause the problems with your DAC. You can try to limit the bits per samples to 16bit within your application.
Best regards
Daniel -
minko Hi Daniel,
Thank you for the quick response. In the described scenario above I disabled sample rate converter in volumio. However I tried to use the converter to set 44.1 khz x 16 bits and 48 khz x 16 bits, but DAC still did not recognize the sampling rate. Next think that I'll try is different distribution do you have anything in mind ?
Regards,
Minko -
HiFiBerry team I don't think a different distribution will change anything here. As far as I understand the output works, the DAC just doesn't display the sample rate. If this is the case, I wouldn't care for this.
Best regards
Daniel -
minko I'm worried because I think that sample frequency is not precise enough. In addition I expected better quality than the setup with the usb to spdif converter, however that's not the case, usb-spdif sounds better.
Regards,
Minko -
HiFiBerry team If you think, your USB adapter sounds better and you bought the Digi from us, you can return the Digi within 30 days.
https://www.hifiberry.com/shop/money-back-guarantee/
Best regards
Daniel -
Thomas Hi Daniel,
here an update on my problem with the Denon DA500. I did test the Digi+ also with my second DA-500 which exactly shows the same behavior. That means no sample rate is going to be displayed. I also tested with volumio and fixed output as 16bit.
Daniel, you mentioned a new product previously in this thread. Will it also offer next to the audio output a digital output path? Because I am wondering if I should keep the Digi+ device.
with many greetings,
--thomas -
HiFiBerry team Hi Thomas,
we don't plan any replacement or upgrade to the Digi in the near future.
Best regards
Daniel -
Thomas Hi Daniel,
I were reading in the meantime about this Spdif topic and could not find a particular good reason why my dac is not recognizing the signal as expected.
You are using on the DIGI+ a 27mhz Quartz "others" seem to use a 12mhz and the spec. recommends 10-27mhz. What I do not completly understand is that the quartz clock is not a multiple of 44,1khz or 48khz etc.? The WM8804 seem to reclock the signal by using the external 27mhz oscillator.
But how close you get at the end to the 44.1khz??? Is it possible that my DAC can "lock on" with its internal PPL to the SPDIF stream but its out of its undocumented boundaries/limits and therefore its not displaying the clockspeed ...?
Could the variations/tolerances on the onboard oscillator be the reason vor my issue? Would it make sense to test a second sample of the DIGI+ or maybe i should try the wolfson audio card to see if this is an general issue or if it is specific to my board (do you know by chance if the wolfson card is also using a 27mhz oscillator?).
with kind regards,
--thomas -
HiFiBerry team Hi Thomas,
the chip used on the HiFiBerry Digi is using an internal PLL that can virtually create any frequency out of the 27MHz external clock.
As I don't have the Denon DAC available for tests, I can't say how it works and why its PLL isn't locking. Everything would just be a pure speculation. I also can't say how other products work.
If you want to test another board, you can order one and return it. But note the conditions of our money-back guarantee here:
https://www.hifiberry.com/shop/money-back-guarantee/
Best regards
Daniel -
Thomas Hi Daniel,
I see there is some "magic" within the WM8804. It is using any external clock to usi it for the internal PLL as well as for the generation of the clock for the different "output" signals... right?
Is there a particular reason to use 27mhz?
Because the Wolfson Audio Card is using 12mhz. I think I will try to get hands on one of these 12mhz cards to do a testing ...
with kind regards,
--thomas -
HiFiBerry team Yes, the chip does some "magic" (which is called fractional PLL) internally. The base frequency doesn't really matter as long as the driver selects the right coefficients.
Best regards
Daniel -
Thomas Hi Daniel,
Is this Driver/WM8804 "magic" a black box? What happens if we would change the Oscillator frequency? Do I need then a new driver? Or is the existing driver recognizing this and therefore picking new coefficients to match the new frequency situation?
with kind regards,
--thomas -
HiFiBerry team Hi Thomas,
the crystal frequency is hardcoded in the driver. To use another frequency you have to change the driver source code and recompile the kernel.
Best regards
Daniel -
Samuel Darby I see similar behaviour with Digi+ and Lindy DAC-Pro. The DAC always reads 44.1KHz regardless of the output frequency of the Digi+.
I tested the same FLACs on an Denon AV Amp and it displayed the correct frequencies (44.1KHz, 48KHz, 96KHz). I tested a squeezebox touch with the DAC-Pro and again the correct frequencies were displayed.
I'm wondering if the DAC-Pro is attempting to display the frequency derived from the SPDIF status block instead of the frequency it was locked onto. -
HiFiBerry team Interesting theory. The channel status should be set to "undefined", but I will have a look at the driver.
Best regards
Daniel -
HiFiBerry team Hi Samuel,
today I did some tests with the Yamaha DSP-A2. Its input does not even lock to a signal if the sample rate is not indicated in the SPDIF status bits. As our drivers does not indicate the sample rate, it doesn't sync at all. I'm working on an updated driver now that should fix this problem. Maybe this will also fix the display on the Lindy DAC.
Best regards
Daniel -
Thomas Hi Daniel,
this seem to be great news. Maybe my da-500 is syncing but for the proper indication it is reading the status bit's? Who knows what they had in Mind in the nineties :-). Please let us know when this fix got merged into the 3.18 kernel. I would love to test this fix.
many thanks + good night,
--thomas -
HiFiBerry team The fix already got merged. However, I can't say how long it will take until a new firmware will be released.
Best regards
Daniel
Please sign in to leave a comment.