The AMB gamma 2, which uses the Wolfson (now Cirrus I think) WM8741/42, has an integrated SRC4192, which is an earlier version of one of the SRC chips that MarK has been working with. I also think that it was also part of the reason that the Gamma 2 sounded so good. It was a giant killer in it's time, before the ESS chips became so prevalent. I also expect that the latest AK chips work very well with the 4137, since they were designed by the same company
@Ian Greenhalgh
the greenm "smp" es9038q2m is the original design and the very 1st es9038q2m dac available on late 2016, the blue one you have come up one year later and just a cheap copy of the the greenm "smp" es9038q2m.
the greenm "smp" es9038q2m is the original design and the very 1st es9038q2m dac available on late 2016, the blue one you have come up one year later and just a cheap copy of the the greenm "smp" es9038q2m.
@eslei, I take it your point to Ian is that he purchased one of the cheap copies he was complaining about, because it was cheaper.
I'm still not convinced you can turn off the dac MCU with jumpers. The MCU on my dac does not turn off with J1,J2 installed. Since the dac MCU runs constantly, you may or may not be able to grab control the I2C bus. Please let us know if you are successful that way. Otherwise, maybe time to lift the MCU I2C pins.
EDIT: Oops! Looks like we cross posted. Good to see to decided to get an analyzer.
EDIT 2: Notice one of your scope traces looks tilted. Are you using 10x probes that are calibrated?
I'm still not convinced you can turn off the dac MCU with jumpers. The MCU on my dac does not turn off with J1,J2 installed. Since the dac MCU runs constantly, you may or may not be able to grab control the I2C bus. Please let us know if you are successful that way. Otherwise, maybe time to lift the MCU I2C pins.
EDIT: Oops! Looks like we cross posted. Good to see to decided to get an analyzer.
EDIT 2: Notice one of your scope traces looks tilted. Are you using 10x probes that are calibrated?
Last edited:
i have purchased 9 "smp" es9038q2m boards, the quality is consistency, modified avcc for my audiophile frend, all good.
i lift the attiny24 pin 2,3 at the beginning, couldn't get control of the dac but the driver is working, then add J1,J2, no luck, so time to get a logic analyzer.
you are right, that probe is a very old 10x, it dosn't have calibration features.
What I2C bus address are you using for the dac chip? It should be 0x90, if using 8-bit addressing format.
EDIT: No calibration ability for a 10x probe would be very unusual. If no little hole somewhere, I think some really old ones might have used a threaded barrel adjustment at the probe end. That's really stretching my memory though.
EDIT 2: Might there be an I2C bus sniffer or monitor for RPi? Could be somebody wrote an app for it.
EDIT: No calibration ability for a 10x probe would be very unusual. If no little hole somewhere, I think some really old ones might have used a threaded barrel adjustment at the probe end. That's really stretching my memory though.
EDIT 2: Might there be an I2C bus sniffer or monitor for RPi? Could be somebody wrote an app for it.
Last edited:
It looks like I made a mistake buying the blue version, I had no idea it was any different to the green one, oh well, live and learn. The blue one is so bad it's unlistenable, it sounds like a 1960's handheld transistor radio tuned to a weak station.
So I'd definitely encourage people to get the SMPCB green version.
So I'd definitely encourage people to get the SMPCB green version.
Really, is the green one that bad too? I hadn't imagined that was the case, even my ultra cheap CS4344 USB DAC sounds really pretty good in stock form, so I really didn't imagine the ESS board was so bad in stock form.
The mind really boggles how they could get things so wrong, I guess it just goes to show that the ESS Sabre chips are not at all simple to implement. That's the strength of the Cirrus chips - they are designed to be ultra easy to implement. I think ESS made the ES9023 variant to be similarly simple to implement.
BTW, if you have a good use for my blue board, you can have it, maybe you can use it to experiment with some other mod ideas, it's redundant for me as I've taken a different route.
The mind really boggles how they could get things so wrong, I guess it just goes to show that the ESS Sabre chips are not at all simple to implement. That's the strength of the Cirrus chips - they are designed to be ultra easy to implement. I think ESS made the ES9023 variant to be similarly simple to implement.
BTW, if you have a good use for my blue board, you can have it, maybe you can use it to experiment with some other mod ideas, it's redundant for me as I've taken a different route.
I didn't have any means of volume control plugged into the board, rather I connected the board via phonos to my Yamaha DSP-E800 amp and used the volume on the amp, the output of the board was at a fairly low level so the volume had to be turned up considerably more than from other sources. The distortion didn't seem to increase with volume, it was there at all volume levels.
I never tried plugging a set of headphones into the board's headphone socket as I don't own any headphones.
The source I used was my PC via a TOSlink cable, I never tried using the RCA input.
I tried powering the board with a wall wart and then from a battery I made for 18650 cells, the distortion was the same with both.
Really, it sounded so bad with so much distortion it had to be faulty.
I never tried plugging a set of headphones into the board's headphone socket as I don't own any headphones.
The source I used was my PC via a TOSlink cable, I never tried using the RCA input.
I tried powering the board with a wall wart and then from a battery I made for 18650 cells, the distortion was the same with both.
Really, it sounded so bad with so much distortion it had to be faulty.
Important is the digital volume going into DAC, no matter what you do at analog output side. If dgiital level is on 0dB the digital FIR filter in DAC could lead to inter sample clipping when by filter calculation the vol runs over 0dB. I guess that is why so many users here talk of such bad out of stock performance. maybe a small resistor on onboard vol. control poti connector could solve this issue.
Looks like Allo is going to stop selling Katana dacs and try to improve the design some more before bringing them back to market. Having put all the effort that they put into end up about where my my dac is, is not good enough. If they are successful maybe that will be a very good thing. After all, the dac I produced in this thread was made without using any test instruments to help shape the design. That was constraint we were trying to work with for a low-cost DIY project. Personally, I wish them the best possible outcome from their continuing efforts. Here is the story from cdsgames: http://www.diyaudio.com/forums/vendor-s-bazaar/294940-fifo-buffer-rpi-sbcs-218.html#post5509359
Last edited:
Ian, Deenoo would be correct that you can't use a phono input with a dac, if that is what happened. Maybe you know that it would probably get overloaded and distort, not to mention to RIAA EQ wouldn't be right for a dac. Hopefully, you meant to say to connected the dac to an aux or tape input of some kind on amplifier.
Also, to the point made by freezebox, the dac should be turned down some preferably by turning down the digital signal sent to it from your computer by maybe around 4dB. If using Windows on your computer we could explain how to do that from the Windows Sound settings in the Windows Control Panel.
Also, to the point made by freezebox, the dac should be turned down some preferably by turning down the digital signal sent to it from your computer by maybe around 4dB. If using Windows on your computer we could explain how to do that from the Windows Sound settings in the Windows Control Panel.
Last edited:
I feel sorry for them. It is frustrating but I think there is something going on that many might have missed and Markw4 might have been able to avoid somehow. That the DAC3 gets right. Jitter reduction/control of the source.
It has profound effect even on the op amps. I've come to realize for me it is a bigger issue than I thought. I think jitter possible creates a complex distortion that so far the end result for testing is simply listening. Careful changing of sources and transport mechanisms will show it up. The best so far has been delivery via ssync USB.
Folks, bits is not bits.
Mikett, Katana runs in master mode to control jitter and probably does a fine job of it. That works for a music player dac, but not as well for a general purpose dac.
What it really is, is what Dave Hill of Crane Song said, "Everything matters, even the things that the manufacturer says doesn't matter. It all matters."
Making a really good dac is like peeling away layers of an onion. You fix one thing only to reveal the next lower level problem you didn't know about before.
Allo can fix the analog stage so that it is not adversely affected by the low level RF leakage out of the dac. (Also, to make sure low level RF doesn't make it into a power amp.) They might be able to get the dac THD down to what ESS specifies as possible which is -120dB. But doing that will not necessarily make it sound like DAC-3. It might be peeling away another layer of the onion. At some point there is no avoiding getting into the interpolation filter, IMHO, if one wants to get it down to the best that is possible in terms of audible SQ improvements. That may not be apparent yet, or addressed ever, we will just have to wait and see.
One thing that I am not so sure about either is using a panel of audiophiles to judge SQ. My preference would be to use a panel of successful mastering engineers. Maybe some audiophiles too, but the problem with audiophiles is how do you vet their opinions? Some may hear accurately and some may claim they do. Hard to know without blind testing them, but we don't have well developed blind testing for the top 5% of listeners. At least success at mastering can be a type of vetting. If everybody in an industry respects someone's work, that may not be a bad proxy for blind testing. At least, it might be the best we can do for now.
What it really is, is what Dave Hill of Crane Song said, "Everything matters, even the things that the manufacturer says doesn't matter. It all matters."
Making a really good dac is like peeling away layers of an onion. You fix one thing only to reveal the next lower level problem you didn't know about before.
Allo can fix the analog stage so that it is not adversely affected by the low level RF leakage out of the dac. (Also, to make sure low level RF doesn't make it into a power amp.) They might be able to get the dac THD down to what ESS specifies as possible which is -120dB. But doing that will not necessarily make it sound like DAC-3. It might be peeling away another layer of the onion. At some point there is no avoiding getting into the interpolation filter, IMHO, if one wants to get it down to the best that is possible in terms of audible SQ improvements. That may not be apparent yet, or addressed ever, we will just have to wait and see.
One thing that I am not so sure about either is using a panel of audiophiles to judge SQ. My preference would be to use a panel of successful mastering engineers. Maybe some audiophiles too, but the problem with audiophiles is how do you vet their opinions? Some may hear accurately and some may claim they do. Hard to know without blind testing them, but we don't have well developed blind testing for the top 5% of listeners. At least success at mastering can be a type of vetting. If everybody in an industry respects someone's work, that may not be a bad proxy for blind testing. At least, it might be the best we can do for now.
Last edited:
There is no absolute and there will never be. DAC1, DAC2 DAC3....and the next one to come which I am sure Benchmark is working on now. At some point, given a price point, it is good enough for shipping has to suffice. I hope ALLO does not fall into the high end trap of always modifying and tweaking their current product to their own revenue detriment. Store up the revisions for a major rerelease of VII
I really hope they do not fall into the unreal expectations and perfectionist streaks of engineers. Engineers typically have a fault of not being able to come to grips with market realities. Their price point and performance level must be defined. Then they put out the design that meets those objectives with the simplest and least costly way to get there. Is it a reachable hence realistic goal to match the DAC3? Nice if they could BUT I hope they don't set themself an impossible task. They need to continue to exist hence ship. They can start peeling but getting deep enough to the inside may take two or three models to gather the experience and knowledge to know what still counts.
I really hope they do not fall into the unreal expectations and perfectionist streaks of engineers. Engineers typically have a fault of not being able to come to grips with market realities. Their price point and performance level must be defined. Then they put out the design that meets those objectives with the simplest and least costly way to get there. Is it a reachable hence realistic goal to match the DAC3? Nice if they could BUT I hope they don't set themself an impossible task. They need to continue to exist hence ship. They can start peeling but getting deep enough to the inside may take two or three models to gather the experience and knowledge to know what still counts.