My computers playing up so using a quote is a simple way to get in to the thread . I love the being banned bit in your tag . Move to the UK . Plumbers charge $100+ an hour if they admit it . They seem to be almost doctors .
A question . Adam who writes on this thread has bought this DAC . I want to use an external linear PSU for the Raspberry plus DAC . It solves so many problems to do this . Ideally I don't want the ready made Raspberry devices . For pennies more an old fashioned linear can be tried . My question is the Raspberry power in cables seem to cost more than budget PSU's . I object to cutting the PSU off to get the cable . Is it possible to power the Raspberry via the DAC ( easier soldering to the 5 V rail that way ? ) . Or is there a convenient larger power point on the Raspberry ? Sorry to be so lazy asking others . Time poverty and a full head asks for this help . I will use a 2.5 mm power in socket as used 50 years ago as my preference ( we called them Roker I think ?) . The practical bit is the place where power goes in might have consequences . Perhaps an opportunity to find ideal 0 V .
I phoned a friend Paul about CD into the Raspberry ( who want to be a Millionaire style ) . His remark was if it will play DVD it should play CD . Then said some Pioneer transports for DVD are the best of all for CD . Daft questions sometimes have good answers . Pioneer best transport !!! Not the Blue Ray he said .
Adam if you read this Paul helped bring back metal cone speakers . He and I don't agree on that . Acoustic Energy are well respected all the same . BTW my ex wife thinks your amp sounds great . Seeing as she seldom will give anything I do the time of day that is praise beyond anything anyone at DIY Audio could provide . She is not wrong . With metal cones ? I doubt it .
Parts and sound . In my experience the parts are 20 % of the story . They can be the 20% one needs . For example the much disliked NE5534 op amp is not too bad . Most people who do not like it never tried to make it work properly is my guess . It can sound more transparent than most capacitors . They didn't like it because it is the girl next door is my guess ? Surely the exotic is better ? In context of a strong RFi field a J FET op amp might work better ( OPA 604 , 2604 are very cheap and better than TL0 series ) . One could buy a bit of Mu-metal to help the 5534 ( 2 ) for less money . 99% wouldn't do that as it is almost mechanical engineering at that point .Op amps were rejected and declassified in about 1962 ( Many have not been credited . H C Lin is one . LM 741 existed in 1963 it is said ) . Their Rfi performance was not suitable for military use . The Soviets even used valves as they were very OK and needed no improvement except for the weight penalty . J FET's whilst not valves are better suited . I say that becasue it is too easy to say they are the same . The lack of high voltage ability in op amps to me is the problem . Hitachi had a SIL ( most are DIL ) op amp once which looked to show the way . Alas it was not widely used . I think it was mistaken as for phono use only .
http://www.analog.com/static/imported-files/tutorials/MT-096.pdf
Last edited:
Send my regards to your Ex - she has great taste
I do hope it plays well with metal cones as I've just invested in some Mark Audio Alpair 10's and my reference pair at home are Monitor Audio SE20's - metal everywhere you look!
That said the Alpair's are supposed to be very neutral for any driver let alone metal, worst case is that I can swap them out for paper - the Alpair 10p.
Encouraging about power in via the DAC, lets hope that leads to a simple solution within the amp.
Please can you elaborate? Are you saying that for best sound quality you should use a wireless usb connection rather than the Ethernet port on the Raspberry Pi?
Yes, that's what I'm saying. Check out the measurements here:
Tracking down noise sources on a Raspberry Pi | Crazy Audio
We're not talking about extreme levels of noise. You may not even hear an audible difference. However it can make a difference between 0.003% THD+N and 0.01% THD+N.
The link above shows a single measurement in a small home network. Depending on the network equipment, network utilization, layout, cabling this may vary a lot. Maybe there is not even a minimal impact in some environment.
What's shown here is not noise injected from the power supply. In these measurements, the Raspberry and the DAC were completely unpowered. That means a better power supply will not improve this.
Tracking down noise sources on a Raspberry Pi | Crazy Audio
We're not talking about extreme levels of noise. You may not even hear an audible difference. However it can make a difference between 0.003% THD+N and 0.01% THD+N.
The link above shows a single measurement in a small home network. Depending on the network equipment, network utilization, layout, cabling this may vary a lot. Maybe there is not even a minimal impact in some environment.
What's shown here is not noise injected from the power supply. In these measurements, the Raspberry and the DAC were completely unpowered. That means a better power supply will not improve this.
Thanks for the explanation. I bought some powerline adapters to make the setup simpler, but perhaps I will try using a wireless adapter again and see if I can notice a difference.
I presume the reason for picking the PCM5122 is because it has the ability to generate SCK from BCK itself ?
Im looking into the possibility of having a PLL generate SCK, dividing that to get a BCK, and slaving the Pi to this clock. My project is a complete audio streaming player using the Pi, and providing both an analogue and SPDIF output using Cirrus Logic parts (because theyre available in SOIC which I can work with at home).
Im looking into the possibility of having a PLL generate SCK, dividing that to get a BCK, and slaving the Pi to this clock. My project is a complete audio streaming player using the Pi, and providing both an analogue and SPDIF output using Cirrus Logic parts (because theyre available in SOIC which I can work with at home).
interesting that people are actually paying money for the very flawed pcm5102.
it has a known problem (gibbs) when the peaks stay too close to 0db for 'too long'. I built a dac using this chip and had to abandon it due to this very bug. its in the chip and nothing can be done about it. and when the engineer or marketing guy at TI spoke up (on diyaudio, in another thread), I asked about this flaw and suddenly, he stopped posting and was silent as can be. I think he knew the chip was a piece of crap.
maybe you guys have not noticed the problem, but with the loudness war putting audio so close to 0, its not hard to notice the bug.
there are much better dac chips that don't cost that much more. this one is a non-starter for anything other than 'mobile phone audio' (so to speak).
it has a known problem (gibbs) when the peaks stay too close to 0db for 'too long'. I built a dac using this chip and had to abandon it due to this very bug. its in the chip and nothing can be done about it. and when the engineer or marketing guy at TI spoke up (on diyaudio, in another thread), I asked about this flaw and suddenly, he stopped posting and was silent as can be. I think he knew the chip was a piece of crap.
maybe you guys have not noticed the problem, but with the loudness war putting audio so close to 0, its not hard to notice the bug.
there are much better dac chips that don't cost that much more. this one is a non-starter for anything other than 'mobile phone audio' (so to speak).
You mean Rochey as the engineer? On another forum he answered like this to question about the clipping om PCM5102(A?):
"For the clipping output, try switching the I2C programmable parts to use VCOM instead of VBandgap. We use a bandgap reference on the PCM5102 to generate the reference voltage, instead of the usual resistor divider with an external cap. That gives it great Power supply noise rejection, at the cost of requiring a relatively steady 3.3V or higher. If your power supply sags too much, then the bandgap reference continues to stay steady at 1.65V (and is no longer necessarily half of the rails).
PCM512x and PCM514x (and newer derivatives) have the option to switch between both. My personal preference is to use the bandgap, but then, I usually have a nice 3V3 regulator to look after the device. "
Maybe it's better have large caps close to power pins and use largest possible supply voltage to run the chip. You tried it already?
he's saying the power supply will affect gibbs?
uhm, no. I don't think so. I honestly don't see how on earth that could do anything.
the phenom was noticed by many of us (see the 'usb audio widget' thread). there was an ESS chip that had this bug, we abandoned it; we tried 5102, had the same basic bug and we blew this chip off, as well, as a cheap non-quality budget-only chip. like I said, in cellphones and low end audio systems, they could atten the signal digitally so to avoid gibbs issues. but in quality audio, that solution is NOT acceptable.
uhm, no. I don't think so. I honestly don't see how on earth that could do anything.
the phenom was noticed by many of us (see the 'usb audio widget' thread). there was an ESS chip that had this bug, we abandoned it; we tried 5102, had the same basic bug and we blew this chip off, as well, as a cheap non-quality budget-only chip. like I said, in cellphones and low end audio systems, they could atten the signal digitally so to avoid gibbs issues. but in quality audio, that solution is NOT acceptable.
AD1953 data sheet tells:
"A digital clipper circuit is used between the output of the DSP
core and the input to the DAC Σ-Δ modulators to prevent overloading
the DAC circuitry (see Figure 4). Note that there is a
gain factor of 0.75 used in the DAC interpolation filters, and
therefore signal values of up to 1/0.75 will pass through the DSP
without clipping."
Source: http://www.analog.com/static/imported-files/data_sheets/AD1953.pdf
That's to avoid the problem with Gibbs (- Is that really the problem, Gibbs effect means the overshoot and ringing in frequency domain as a result of truncation of the higher frequencies/harmonics which if not truncated would propably sum so that there were no overshoots and ringing)? Anyway, life is compromises - either attenuate the signal or be afraid of hearing the digital clipping, what ever sounds best for you.
Noticed just today that the XMOS USB Audio 2.0 Reference Design (firmware for the XMOS chips) implements volume control by attenuating the signal itself (guess it can be configured to use I2C and codec interface as well).
"A digital clipper circuit is used between the output of the DSP
core and the input to the DAC Σ-Δ modulators to prevent overloading
the DAC circuitry (see Figure 4). Note that there is a
gain factor of 0.75 used in the DAC interpolation filters, and
therefore signal values of up to 1/0.75 will pass through the DSP
without clipping."
Source: http://www.analog.com/static/imported-files/data_sheets/AD1953.pdf
That's to avoid the problem with Gibbs (- Is that really the problem, Gibbs effect means the overshoot and ringing in frequency domain as a result of truncation of the higher frequencies/harmonics which if not truncated would propably sum so that there were no overshoots and ringing)? Anyway, life is compromises - either attenuate the signal or be afraid of hearing the digital clipping, what ever sounds best for you.
Noticed just today that the XMOS USB Audio 2.0 Reference Design (firmware for the XMOS chips) implements volume control by attenuating the signal itself (guess it can be configured to use I2C and codec interface as well).
Last edited:
Powering the Hifiberry with external linear psu?
Hello,
I just received Hifiberry DAC and I wondering how to powering it up via EXTERNAL linear power supply (my GPIO port is quit busy and I would prefer the DAC to have his own psu). The DAC will be added on a raspberry pi with volumio1.2 beta.
I could imagine two solutions :
1)
- I do not add a pin on the 5v (from the p5 of the pi). Then only the 3.3V (from the p5 of the pi) will be delivering power to the DAC. 3.3 V is used for the digital part of the DAC.
- I connect +5V and ground cables (from a linear psu rail) to the “+ 5V -” slots on the DAC (right to the filer jumper of the DAC)
question : did I have to merge the ground cables (the one powering up the pi through micro USB and the one powering up the DAC ?)
2)
Since my pi is also powered by a 5V linear psu, I just do the normal thing (connect all pins of the P5 (including the 5V)) and plug the DAC on the pi. I’m not too much in favor of this solution since I already have a “busy pi” with a 4×20 LCD, IR and temperature sensor.
My 5V linear psu should theoretically output 5V under 2 amps (did mot measured it but this is what I’ve ordered).
Is there is a chance that I could hear a difference by picking up one or the other solution ? What would be the preferred solution ?
Thanks a lot for your inputs,
Francis
Hello,
I just received Hifiberry DAC and I wondering how to powering it up via EXTERNAL linear power supply (my GPIO port is quit busy and I would prefer the DAC to have his own psu). The DAC will be added on a raspberry pi with volumio1.2 beta.
I could imagine two solutions :
1)
- I do not add a pin on the 5v (from the p5 of the pi). Then only the 3.3V (from the p5 of the pi) will be delivering power to the DAC. 3.3 V is used for the digital part of the DAC.
- I connect +5V and ground cables (from a linear psu rail) to the “+ 5V -” slots on the DAC (right to the filer jumper of the DAC)
question : did I have to merge the ground cables (the one powering up the pi through micro USB and the one powering up the DAC ?)
2)
Since my pi is also powered by a 5V linear psu, I just do the normal thing (connect all pins of the P5 (including the 5V)) and plug the DAC on the pi. I’m not too much in favor of this solution since I already have a “busy pi” with a 4×20 LCD, IR and temperature sensor.
My 5V linear psu should theoretically output 5V under 2 amps (did mot measured it but this is what I’ve ordered).
Is there is a chance that I could hear a difference by picking up one or the other solution ? What would be the preferred solution ?
Thanks a lot for your inputs,
Francis
Francis, soon I hope you will be able to check with the Pi-DAC. I'll get back to you on your other questions shortly.
Gordon@IQaudIO.com
Gordon@IQaudIO.com
Hi, the IQaudIO.com Pi-DAC is available via the Tindie.com website.
Very best regards,
Gordon@iqaudio.com
Very best regards,
Gordon@iqaudio.com
@XCOMRider: The HiFiBerry DAC is available in our online shop HiFiBerry - High quality Raspberry Pi Audio | Audiophile 24/96, 24/192 sound cards (Audio DAC and SPDIF out) for the Raspberry Pi, audio out, audio cards, audio projects. The DAC on Tindie is NOT the HiFiBerry DAC.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.