I dont think you understood the question Ian; unless i've misunderstood it myself.
If I understand correctly, diyiggy wants to use the fifopi outputs for digital crossover, where the datastreams contain different data, not dual mono, where both the datastreams are the same and the independent L/R dacs only use what data they need.
As I understand it, the fifopi has 2 outputs, one stereo i2s output on GPIO and a duplicate of that same stereo i2s on µfl. if the crossover/DSP is after the fifo, processing each of those separately, you could use it that way, but I dont think thats not what diyiggy wants. Your MCFIFO does the job, but not fifopi.
Raspberry pi does not do multichannel i2s unfortunately, it only outputs multichannel via USB, or HDMI (could likely make it work over the PCIE bus too, but that's a different story)
Last edited:
We live in a age where people get triggered by a complete stranger canceling their order and trying impose their own purchasing decision philosophy. Only the person purchasing knows what's best for their system and safety. Advice appreciated, but beyond that let people decide what's best for them. Even if it's a mistake, let them learn from that mistake.
Might as well moderate the Group Buy thread too when people modify their GB orders. I don't agree that <username> changed their order. Why don't they just stay put with their original intentions. They are not allowed to adjust and make the proper game theory decision on my watch.
A complete waste of energy. I forgive as they are just trying to look out for my best interests, but it's a little too much to put so much emotion into worrying about someone else's order.
Let's just move on. Poster has like +30 positive points with me and only one -1 so it's a minor blemish. I'm leaving this DIY hobby soon too so all is good and well. He didn't know I was prioritizing safety first.
Might as well moderate the Group Buy thread too when people modify their GB orders. I don't agree that <username> changed their order. Why don't they just stay put with their original intentions. They are not allowed to adjust and make the proper game theory decision on my watch.
A complete waste of energy. I forgive as they are just trying to look out for my best interests, but it's a little too much to put so much emotion into worrying about someone else's order.
Let's just move on. Poster has like +30 positive points with me and only one -1 so it's a minor blemish. I'm leaving this DIY hobby soon too so all is good and well. He didn't know I was prioritizing safety first.
Last edited:
only person triggered here is you mate ...
If you are going to fill this public thread with your purchasing monologue, you better get used to people commenting on it ... apparently I wasn't the only one who noticed.
If you are going to fill this public thread with your purchasing monologue, you better get used to people commenting on it ... apparently I wasn't the only one who noticed.
Last edited:
... and anyway, not sure if you noticed; I spoke from personal experience with this. I have to work hard to keep it in check; personally. No reason to be upset, I was just making an observation and the reason I recognised it, is because i'm guilty of it.
Yes, if there was more information on 3000F solutions I would not share my thought process. I'm only trying to contribute to an optimal solution, not worrying about other people's orders.
It's better to try to provide a solution instead of getting so emotional. I guess because you had no answer for a 3000F solution, you turned your attention instead on the ordering process instead of the solution process.
It's okay if you don't have an answer for everything, no need to divert that attention to someone's order.
Many thanks for all your advice. In this case, no need to get emotional. There are many posts on forums that are much worse than someone trying to layout their purchasing decision that involves their own property loss / personal safety. It's okay to not have a solution.
It's better to try to provide a solution instead of getting so emotional. I guess because you had no answer for a 3000F solution, you turned your attention instead on the ordering process instead of the solution process.
It's okay if you don't have an answer for everything, no need to divert that attention to someone's order.
Many thanks for all your advice. In this case, no need to get emotional. There are many posts on forums that are much worse than someone trying to layout their purchasing decision that involves their own property loss / personal safety. It's okay to not have a solution.
Okay, I'm going to leave this thread early so people don't get triggered on which direction I want to pursue on my Ultra Cap orders. I guess I'm taking up space trying to find a 3000F solution.
Good luck all. I'm happy with my Ian system and just need to upgrade the power solution over time. Goal accomplished besides the UcPure hurdle, but I should have that handled by sharing my thought process here. Sharing that process helped with making a better buying decision.
Good luck all. I'm happy with my Ian system and just need to upgrade the power solution over time. Goal accomplished besides the UcPure hurdle, but I should have that handled by sharing my thought process here. Sharing that process helped with making a better buying decision.
A123, lets just say it wasnt the first time .... lol it was a trend i'd noticed and it made me giggle, so I mentioned it. apparently you didnt find it as funny as I did.
Emotional? who got emotional?
I'm not here to solve all your problems. If I have a solution based on work i've done before, or if something is obvious, fine i'll talk from experience, but as far as doing your research for you, when there is a good chance you'll just turn around and do something else anyway ... yeah nah.
I honestly don't even remember the question. So let me go check.
OK, so yes, the reason I didnt answer before going back to work, is because I didn't have the schematic for Ian's board ...
As I mentioned in previous posts in that conversation, a single mosfet gate has higher resistance than these Ucaps, even the worlds lowest RDSON fets, so frankly asking for protection at the same time as searching out the absolute lowest resistance µcaps is kinda counterproductive, so yeah I didnt bother, especially without the information I suspected, but didnt have, from Ian above. You may have infinite time to throw at solutions to problems that dont exist; I do not.
Emotional? who got emotional?
I'm not here to solve all your problems. If I have a solution based on work i've done before, or if something is obvious, fine i'll talk from experience, but as far as doing your research for you, when there is a good chance you'll just turn around and do something else anyway ... yeah nah.
I honestly don't even remember the question. So let me go check.
OK, so yes, the reason I didnt answer before going back to work, is because I didn't have the schematic for Ian's board ...
iancanada said:
As I mentioned in previous posts in that conversation, a single mosfet gate has higher resistance than these Ucaps, even the worlds lowest RDSON fets, so frankly asking for protection at the same time as searching out the absolute lowest resistance µcaps is kinda counterproductive, so yeah I didnt bother, especially without the information I suspected, but didnt have, from Ian above. You may have infinite time to throw at solutions to problems that dont exist; I do not.
Last edited:
Many thanks @InspectorGadget.
I'm enjoying your A123 26650 w/ tabs battery solution almost everyday. It's amazing to have so much power in such a small footprint. It makes me want to not upgrade to a desktop Andrea Mori, but I wouldn't complete my DIY arc without a Andrea clock.
I don't want to bore people with my "purchasing decisions" so I'm out early. I was planning EOY, but it's a good time as any since I already finished laying out my 3000F thought process. So I have nothing left to contribute and time to just implement.
I need to focus on Black Friday sales as there may be deals I cannot resist.
I'm enjoying your A123 26650 w/ tabs battery solution almost everyday. It's amazing to have so much power in such a small footprint. It makes me want to not upgrade to a desktop Andrea Mori, but I wouldn't complete my DIY arc without a Andrea clock.
I don't want to bore people with my "purchasing decisions" so I'm out early. I was planning EOY, but it's a good time as any since I already finished laying out my 3000F thought process. So I have nothing left to contribute and time to just implement.
I need to focus on Black Friday sales as there may be deals I cannot resist.
Last edited:
All good A123, enjoy your system without worrying about whether or not you can hear 500µΩ vs 1mΩ 😉 pro tip ... you cant.
Just a side note, there isnt really an end to the DIY ARC, so i've found it best to split projects up into achievable performance goals and clearly identify what those are, before beginning. Otherwise it is difficult to know whether youve succeeded.
I'm more focused on speakers and CNC at the moment, as far as research is concerned
Just a side note, there isnt really an end to the DIY ARC, so i've found it best to split projects up into achievable performance goals and clearly identify what those are, before beginning. Otherwise it is difficult to know whether youve succeeded.
I'm more focused on speakers and CNC at the moment, as far as research is concerned
Guidance of how to assemble the SMT finished SinePi
Here is the guidance of how to assemble the SMT finished SinePi:
DocumentDownload/AssembleSmtFinishedSinePi.pdf at master * iancanada/DocumentDownload * GitHub
Please note that the SinePi has a 4 layers PCB with two inner layers of whole pieces of ground/shield plate. So, you have to use high power (80W or higher) solder iron to solder the pins and the SMA connectors. It could be not as easy as you thought to make a high quality soldering job for it.
Regards,
Ian
Here is the guidance of how to assemble the SMT finished SinePi:
DocumentDownload/AssembleSmtFinishedSinePi.pdf at master * iancanada/DocumentDownload * GitHub
Please note that the SinePi has a 4 layers PCB with two inner layers of whole pieces of ground/shield plate. So, you have to use high power (80W or higher) solder iron to solder the pins and the SMA connectors. It could be not as easy as you thought to make a high quality soldering job for it.
Regards,
Ian
A great story from Audio Creative on a highly pimped DDDAC where the FiFoPi and Ian ReclockPI are featured as well…
berrystreamer-xl-ultimo/
berrystreamer-xl-ultimo/
@dddac
That's really nice 🙂
Thank you for sharing the link
Dit keer verkennen we de bovengrenzen van de BerryStreamer/ DDDAC1794. Waar liggen die? Wat wel zeker is dat de laatste loodjes aan kwaliteitswinst de meeste bloed, zweet en geld kosten. Maar wat blijft dit een boeiende ontdekkingsreis. Met uiteindelijk toch prachtige resultaten. - BerryStreamer XL Ultimo
Ian
That's really nice 🙂
Thank you for sharing the link
Dit keer verkennen we de bovengrenzen van de BerryStreamer/ DDDAC1794. Waar liggen die? Wat wel zeker is dat de laatste loodjes aan kwaliteitswinst de meeste bloed, zweet en geld kosten. Maar wat blijft dit een boeiende ontdekkingsreis. Met uiteindelijk toch prachtige resultaten. - BerryStreamer XL Ultimo
Ian
Attachments
Thanks.
In my scenario, the convolution of the crossover is made by a computer that output on a single USB. I assume maybe the MCDual could outputt the splitted digital signal, low pass on an I2S output and High pass on the second I2s output? I noticed such multichannel DAC have a standalone usb input for Roon, HQ player, etc , that accept crossover convolution from a computer on their standalone usb input (oktodac pro, hexasound dacs) Hence the question to Ian.
Last edited:
Yes, those units have a second USB input, or rather they provide access to the rpi usb port as well as the one used for the xmos USB input. if using the rpi for convolution in oktodac, the usb output is connected to the xmos usb input via an external patch cable IIRC. if using a different, external computer to perform the XO, you connect to the xmos USB directly.
anyway, as mentioned, rpi does not support multichannel i2s, nor does fifopi, only MCFifo accepts multichannel i2s input and output, but you would need to feed it from an xmos or similar USB->multichannel i2s convertor.
anyway, as mentioned, rpi does not support multichannel i2s, nor does fifopi, only MCFifo accepts multichannel i2s input and output, but you would need to feed it from an xmos or similar USB->multichannel i2s convertor.
thanks again. I didn't realize than the USB Xmos board in an OKTODAC 8 channel dac or an HexaSound Dac, was making the splitting channels job towards the dac chips for an active multi dacs filter for active loudspeakers .
Quite a mystery on how it's done. But seems the Xmos chip is programed to see the convolution filters in which there are the crossover channels (one per dac chip channel or dac chip) if i understand what you mean?
Didn't realize that the MCFifo was outputting the same I2S on all the outputs it sees at its input if that one is a standalone one in th etypical USB to I2S board à la Amanero and all.
So MCFIFO ok for a multiple amplification with passive filter in the loudspeakers but not active filter made before the dacs ! We still need something that split the input standalone I2S towards several I2S outputs, each being the crossover band (low pass, high pass, etc) seen in the convolution made by the playback-soft before (Roon, HQ-Player, Acourate, Audirvana, etc).
Of course the goal is to avoid the Adc job before the Dac job, by staying in the numeric domain. Don't know if that can be done with old PCM dac chips or if it needs some digital front end in the Xmos or the dac chips itselves (modern Cirus, AK, ESS chips ?)
Not only some want good PCM dac chips and good digital timing front-ends, they also want sometimes the help of a good PC -NUC style, etc- powerfull enough for multiple active amp filtering, room correction, phase tracking between drivers, with a lot of taps à la Chord but via convolutions sent by the PC through the Fifos/DACs/amp per loudspeaker channel for active filtering.
Asking myself if those moderns multichannels DAC devices have good Fifo and good crystals inside. But off topic here, just wanted to know if I could make 2 channels convolution through a DualMC clock board with its Fifo or the FifoPi. Goal being to have the same good reference of the async clock after the Fifo gate.
Quite a mystery on how it's done. But seems the Xmos chip is programed to see the convolution filters in which there are the crossover channels (one per dac chip channel or dac chip) if i understand what you mean?
Didn't realize that the MCFifo was outputting the same I2S on all the outputs it sees at its input if that one is a standalone one in th etypical USB to I2S board à la Amanero and all.
So MCFIFO ok for a multiple amplification with passive filter in the loudspeakers but not active filter made before the dacs ! We still need something that split the input standalone I2S towards several I2S outputs, each being the crossover band (low pass, high pass, etc) seen in the convolution made by the playback-soft before (Roon, HQ-Player, Acourate, Audirvana, etc).
Of course the goal is to avoid the Adc job before the Dac job, by staying in the numeric domain. Don't know if that can be done with old PCM dac chips or if it needs some digital front end in the Xmos or the dac chips itselves (modern Cirus, AK, ESS chips ?)
Not only some want good PCM dac chips and good digital timing front-ends, they also want sometimes the help of a good PC -NUC style, etc- powerfull enough for multiple active amp filtering, room correction, phase tracking between drivers, with a lot of taps à la Chord but via convolutions sent by the PC through the Fifos/DACs/amp per loudspeaker channel for active filtering.
Asking myself if those moderns multichannels DAC devices have good Fifo and good crystals inside. But off topic here, just wanted to know if I could make 2 channels convolution through a DualMC clock board with its Fifo or the FifoPi. Goal being to have the same good reference of the async clock after the Fifo gate.
Last edited:
I cant say for sure about the exa, I havent checked their site in a while, but that is the case for the oktodac8. 2 channel, but also multichannel, since it connects by USB, not the i2s/gpio (which is also in general higher jitter than usb)
regarding the mcfifo, No, sorry you have misunderstood. the MCFIFO is capable of multichannel i2s input and output (with the DSP processing happening before, for best performance) it is the rpi and fifopi that do not allow multi channel i2s input, or output. rpi only allows multichannel output via USB or hdmi, its gpio/i2s is only capable of stereo.
you will find that rpi4 is plenty powerful for dsp, at least using Camilladsp, Henrik's experiments appear to show it is capable of a fair bit before struggling. sure, you could punish it with total OTT settings, but for a stereo 4 way (8 channel) I believe it should be powerful enough. That being said, if you arent fixated on using rpi, a NUC would certainly be more powerful and only take up a marginally larger area.
anyway that's a bit OT, I just wanted to clear that up. the MCfifo is what you need, if you insist on using a fifo, but the okto does appear well designed, so perhaps you dont.
regarding the mcfifo, No, sorry you have misunderstood. the MCFIFO is capable of multichannel i2s input and output (with the DSP processing happening before, for best performance) it is the rpi and fifopi that do not allow multi channel i2s input, or output. rpi only allows multichannel output via USB or hdmi, its gpio/i2s is only capable of stereo.
you will find that rpi4 is plenty powerful for dsp, at least using Camilladsp, Henrik's experiments appear to show it is capable of a fair bit before struggling. sure, you could punish it with total OTT settings, but for a stereo 4 way (8 channel) I believe it should be powerful enough. That being said, if you arent fixated on using rpi, a NUC would certainly be more powerful and only take up a marginally larger area.
anyway that's a bit OT, I just wanted to clear that up. the MCfifo is what you need, if you insist on using a fifo, but the okto does appear well designed, so perhaps you dont.
Last time I looked at a pic of an okto dac PCB, looked like only one clock near the dac chip. If so, it means the dac chip is always resampling the audio. Not ideal in my book. However, feeding it lower jitter I2S input might allow the dac chip DPLL_Bandwidth setting to be reduced to a good minimum, assuming I2S input and DPLL adjustment are both available to the user.
hmmm, really? I agree, that's not ideal. I haven't scrutinized the design that hard, just see some results. Its a fairly vanilla, but well performing IV. many wont care about the single clock, but I do also prefer synchronous operation myself. That being said, using a well performing OSF to upsample to a single, high rate, is perhaps more suitable for uncomplicated DSP, so the DSP is not having to refresh a different set of filters and restart every time it changes sample rate. that is not the way i'm going, but it does avoid some extra complexity and in my experience the power of DSP speaker crossovers and room correction far outweighs synchronous clocking.
Last edited:
If not mistaken the I/V & buffer in the oktodac is a simple all in one modern op1646 or 1648?
I believe in both the Okto Dac Pro and the Hexa which goals is to process active filter multi amp outputs, you can switch off the upsampling via the screen board, then you can input a FIR filter with your own upsampling algos via its standalone USB input.
I'm glad the MCFIFO can read a 3 ways filter convolution via a single I2S input then split each way on its own I2S output. It's not clear to me though if the standalione USB to single I2S must be on a special type.
I don't know about the multi channel outputs of an Hexa Dac we often see with a NUC pc before that makes the filter channels splitting for active loudspeakers via a software... HQ player, Roon, etc !
it could be a dream to have such >Fified multichanned dac with one PCM dac chip channel per way instead multiple outputs on a standalone dac chip à la AK or ESS.
If the MC Fifo can read a digital multiplexed convolution on its of its standalone input and split it on its outputS, then for sure it can be done with simple PCM dac chips
I believe in both the Okto Dac Pro and the Hexa which goals is to process active filter multi amp outputs, you can switch off the upsampling via the screen board, then you can input a FIR filter with your own upsampling algos via its standalone USB input.
I'm glad the MCFIFO can read a 3 ways filter convolution via a single I2S input then split each way on its own I2S output. It's not clear to me though if the standalione USB to single I2S must be on a special type.
I don't know about the multi channel outputs of an Hexa Dac we often see with a NUC pc before that makes the filter channels splitting for active loudspeakers via a software... HQ player, Roon, etc !
it could be a dream to have such >Fified multichanned dac with one PCM dac chip channel per way instead multiple outputs on a standalone dac chip à la AK or ESS.
If the MC Fifo can read a digital multiplexed convolution on its of its standalone input and split it on its outputS, then for sure it can be done with simple PCM dac chips