Presumably, you could hack up a way to use any I2S source with the Katana. I doubt Allo would officially support such an arrangement, but I don't see why it wouldn't be doable. You'd need to wire I2S pins and power, that should be easy. The tricky part would be to get the driver functionality. Easiest way would be to use an RPi, but not attach to the Katana, just run jumper wires from the right GPIO pins to the DAC. Or you could read the Katana driver source, figure out how it works, then re-implement in a microcontroller (e.g. Arduino).
A really enterprising person could design a "bridge" PCB that mates to the Katana's 40-pin header and provides easy power, i2s, and uController hook-ups, thus turning the Katana into a general DAC. Again, probably frowned upon by Allo, but arguably well within the spirit of this forum.
Presumably, you could hack up a way to use any I2S source with the Katana. I doubt Allo would officially support such an arrangement, but I don't see why it wouldn't be doable. You'd need to wire I2S pins and power, that should be easy. The tricky part would be to get the driver functionality. Easiest way would be to use an RPi, but not attach to the Katana, just run jumper wires from the right GPIO pins to the DAC. Or you could read the Katana driver source, figure out how it works, then re-implement in a microcontroller (e.g. Arduino).
A really enterprising person could design a "bridge" PCB that mates to the Katana's 40-pin header and provides easy power, i2s, and uController hook-ups, thus turning the Katana into a general DAC. Again, probably frowned upon by Allo, but arguably well within the spirit of this forum.
Not sure how easy that will be , but yes its definitely in the spirit of this forum . We will try to assist with a doc...but don't forget , you will void the warranty .
We are testing the last EMI reduction component on primary side. We still have a spike on Mosfet (when turned off because of the parasitic inductance of transformer) but we reduced it 50% and completely eliminated the oscillations (almost nill)
Tomorrow we start checking the secondary side. No other update or info at this time.
Tomorrow has come
A few pages back Allo mentions the minimum impedance of the TPA3118 is 1.6ohm, and they have tested/measured with 4ohm. I believe these are the most recent numbers posted:
4R
5W 10W 15W 20W
0.006234% 0.006425% 0.005151% 0.005142%
8R
5W 10W 15W 20W
0.005873% 0.005180% 0.006097% 0.007637%
The infamous "tomorrow" has also finally arrived.. Volt+ D is up on the Allo and Volumio sites. Very much looking forward to hearing it in a couple days.
A few pages back Allo mentions the minimum impedance of the TPA3118 is 1.6ohm, and they have tested/measured with 4ohm. I believe these are the most recent numbers posted:
4R
5W 10W 15W 20W
0.006234% 0.006425% 0.005151% 0.005142%
8R
5W 10W 15W 20W
0.005873% 0.005180% 0.006097% 0.007637%
The infamous "tomorrow" has also finally arrived.. Volt+ D is up on the Allo and Volumio sites. Very much looking forward to hearing it in a couple days.
How would this compare (or simply: why would this be better?) than a USB connection directly from the RPi to a "good" DAC?
I have a NuPrime DAC-10 (US$1500) to which I hook the RPi with a USB-B Cable.
Would the jitter / noise be enough to make an internally isolated i2s connection a better alternative, even with a DAC costing a lot less?
Thanks for any insight, best regards,
Rafa.
I have a NuPrime DAC-10 (US$1500) to which I hook the RPi with a USB-B Cable.
Would the jitter / noise be enough to make an internally isolated i2s connection a better alternative, even with a DAC costing a lot less?
Thanks for any insight, best regards,
Rafa.
Hi Rafa
On the Katana there is no jitter since this is a "master dac" or rather there is only the jitter of the MCLK.
USB connection and audio...its a rather a long subject and there are many devices out there to clean the noise and jitter from USB (RPI USB is one of the worst)
I2S bus (especially in master) or a "recloaked" i2s has advantages and disadvantages .
Jitter is very low (master or reclocked) but you can only get 32/384Khz and DOP 128
If you look for higer sample rates , USB is the only way.
Katana is designed for highest sound quality with measurements that are pretty good (THD+N at -111.xx a weighted)
Project is on track. Expect some kind of announcement in 3 weeks..
On the Katana there is no jitter since this is a "master dac" or rather there is only the jitter of the MCLK.
USB connection and audio...its a rather a long subject and there are many devices out there to clean the noise and jitter from USB (RPI USB is one of the worst)
I2S bus (especially in master) or a "recloaked" i2s has advantages and disadvantages .
Jitter is very low (master or reclocked) but you can only get 32/384Khz and DOP 128
If you look for higer sample rates , USB is the only way.
Katana is designed for highest sound quality with measurements that are pretty good (THD+N at -111.xx a weighted)
Project is on track. Expect some kind of announcement in 3 weeks..
Pretty much all USB DACs, no matter what price, have shown that flaws
on the USB interface are translating into audible flaws.
As cdsgames says, and I also wrote to you in the other thread, there are numerous gadgets out there, fighting USB related noise, power, jitter issues.
"Good DAC" doesn't mean anything. 1500$ for a DAC doesn't mean anything either.
Again. It's not about more or less cost. Fact is, fighting the I2S related
flaws is much less trouble then fighting the USB induced troubles.
And we shouldn't forget that USB DACs translate the stream also to I2S. To a certain extent
they also have to cope with issues on that account.
I'd also like to mention:
The I2S of the PI is almost as bad as its USB interface!!! We shouldn't forget that either!
The I2S related flaws are slightly easier to handle though. It still requires
reclockers and isolators and separate power supplies to achieve highest quality performance!
If you'd buy a Katana, there is a lot of DIY involved. There's no nice case.
There's no huge marketing and sales chain attached. That alone makes
for a huge difference on the price tag.
Obviously with such a DAC you're 100% glued to a RPI.
And with all bells and whistles you'll probably also end up at around 400/500$. That'll be quite an invest for many people out there.
At such a (DIY-) cost Katana will have to compete with the digital big boys out there.
On the other hand meanwhile there are ES9038Pro (a better DAC then the Katana low-cost ES9038Q2M DAC) USB-DACs @250$ out there.
Pretty amazing considering the cost of that Pro chip alone.
However. As usual. The entire solution makes the music. I'd be very surprised if. e.g. that HifiMeDiy ES9038Pro USB solution would sound better then a Katana. No question, the ES9038Pro is the much better DAC chip. Outputstage, powering, clocking asf, do match that pricetag though.
You bet, there are no wizards out there.
However. Combining things in a different way might end up with very surprising results. You never know.Allo is well aware that they have to deliver an outstanding solution with the
Katana to remain in the game.
They are afaik the first who enter - commercially - the high quality HAT DAC segment.
Interesting developments.
Enjoy.
Last edited:
@soundcheck, Over in the 'ES9038Q2M Board' thread we are DYI modding a cheap DAC into a very good one with no jitter, I2S, or USB issues. Everything works quite well. Cost for all the parts is probably a little over $200, but the SQ is very good (however, it is a lot of work to fully mod!). The is a state of the art reference system consisting of Benchmark DAC-3 and Benchmark AHB2 power amp for subjective comparison purposes. Hope to post some measurements of the DIY DAC very soon.
Also, comparison of ES9038PRO and ES9038Q2M specs shows the Q2M distortion is only 2dB worse that PRO. Mainly they differ in that PRO has 8-channels and Q2M has only 2-channels. It is also possible to use external DSP processing for the reconstruction filter with PRO, rather than using the built-in filters. In addition, it is possible to get somewhat lower noise out of PRO as compared to Q2M, but Q2M noise is still down around -120dB.
The point would be that Q2M can serve as the basis for an excellent DAC, and most of the differences between DACs are in the implementations, not the DAC chips. The design, including schematic, parts selection, and layout around the DAC chip are all critical for best performance.
Also, comparison of ES9038PRO and ES9038Q2M specs shows the Q2M distortion is only 2dB worse that PRO. Mainly they differ in that PRO has 8-channels and Q2M has only 2-channels. It is also possible to use external DSP processing for the reconstruction filter with PRO, rather than using the built-in filters. In addition, it is possible to get somewhat lower noise out of PRO as compared to Q2M, but Q2M noise is still down around -120dB.
The point would be that Q2M can serve as the basis for an excellent DAC, and most of the differences between DACs are in the implementations, not the DAC chips. The design, including schematic, parts selection, and layout around the DAC chip are all critical for best performance.
Last edited:
I think we all agree on the above. Yes we are aware of the ess9038PRO dac at 250$ but in my opinion its playing on marketing (get the newest IC with the better specs.. no matter the rest). That DAC runs in asynch mode (so asrc and dpll enabled) , with an output stage based on on mono opamps (but prob not hi end))
We said this consistently , DAC sq depends on a small part on the DAC ic. Rest (majority ) is implementation .
Katana is a Master Dac that runs in syncronous mode (no asrc , no dpll) with one of the best clock for audio(with some special tweaking
. Ouput stage is discreete and class A. Voltages have been measured and cleaned ( for opamps) with filters and LDOs. Every single rails (in fact) has filtering (and countless measurements were taken to make sure we achived uV low noise , you have no idea guys how many times theory and practical results were different ) + LDOs, in fact critical components have 2 LDOS in series along with HF filters (and plenty of decoupling)Clean power , great clocks , great analog stage no asrc or dpll...thats Katana.
I think that our dac is the greatest since red wine...but I might be wrong. (not the first time). I keep thinking I forgot one thing, one more cap , one last change..
Too late, its already sent for production.
It will be your turn to give your opinion . I keep hearing less than 4 weeks for first units..
Last edited:
Thanks for answers cdsgames and soundchek.
Obviously, by price, I am talking about a 'target' sound. Not the price itself. That's why I said "good DAC", which is also relative, but lets assume that reviews and measurements back this claim. That is why I also included the model, just to disambiguate any cost and quality subjectivity.
That said, yes, the answers are interesting, and obviously, full of possibilities. I don't matter the DIY part, as a matter of fact I rather enjoy it. I also already have the big DAC.
So I'm wondering if I could improve sound by adding a DAC to the PI and then using the big DAC/PRE's analog input (which is not digitized), or if I am better served by a COAXIAL output from the PI to the digital inputs of the DAC, or if still USB would be the better way to solve the issue.
Ideally, one could test all these scenarios and just pick the better solution, alas, where I live, it's like Yoda explains: there is no try, do, or do not . We cannot test or have home trials. We either buy or stay away from things, so we are forced to overthink a little and test less.
That is why this forums are a godsend to us where we could do theoretical testing and ride on the back of knowledgeable people like yourselves.
Thanks again,
Rafa.
Obviously, by price, I am talking about a 'target' sound. Not the price itself. That's why I said "good DAC", which is also relative, but lets assume that reviews and measurements back this claim. That is why I also included the model, just to disambiguate any cost and quality subjectivity.
That said, yes, the answers are interesting, and obviously, full of possibilities. I don't matter the DIY part, as a matter of fact I rather enjoy it. I also already have the big DAC.
So I'm wondering if I could improve sound by adding a DAC to the PI and then using the big DAC/PRE's analog input (which is not digitized), or if I am better served by a COAXIAL output from the PI to the digital inputs of the DAC, or if still USB would be the better way to solve the issue.
Ideally, one could test all these scenarios and just pick the better solution, alas, where I live, it's like Yoda explains: there is no try, do, or do not
. We cannot test or have home trials. We either buy or stay away from things, so we are forced to overthink a little and test less.That is why this forums are a godsend to us where we could do theoretical testing and ride on the back of knowledgeable people like yourselves.
Thanks again,
Rafa.
Last edited:
FYI. Finally the RPI rt-kernel sources have also been updated.
It's now 4.14.39-rt29.
And it also comes with Katana support.
For all the integrators it's gonna be a pretty straightforward hazzle-free integration.
Now we just need to get the DAC on the table for testing.
Enjoy.
It's now 4.14.39-rt29.
And it also comes with Katana support.
Code:
Name: allo-katana-dac-audio
Info: Configures the Allo Katana DAC audio card
Load: dtoverlay=allo-katana-dac-audio
Params: <None>For all the integrators it's gonna be a pretty straightforward hazzle-free integration.
Now we just need to get the DAC on the table for testing.
Enjoy.