Sure, but that's entirely unrelated to my question. I asked why does Kali output an I2S signal if there is no I2S at the input? Is this a bug or a feature?
Niravana project is advancing quite rapidly nowadays.
Today readings at 1Mhz bandwith /1A shows noise at 771uV with a noise floor (ground of our osciloscope ) default at 625uV...so we are at about 150uV pk/pk above ground noise
In the last few months we have tested a ton of SMPSs ....we never seen anything like this. No unit on the market comes close.
1Mhz bandwith vs 20Mhz bandwith....both wave forms are almost similar(with only a small spike of HF that is also observed on ground) , most SMPSs increase the noise significatly when you increase the bandwith.
Nirvana , like a linear PSU , shows the same noise no matter the bandwidth.
Next testing is casing and ground earthing , primary and secondary side...Noise will be reduced even further.
Today readings at 1Mhz bandwith /1A shows noise at 771uV with a noise floor (ground of our osciloscope ) default at 625uV...so we are at about 150uV pk/pk above ground noise
In the last few months we have tested a ton of SMPSs ....we never seen anything like this. No unit on the market comes close.
1Mhz bandwith vs 20Mhz bandwith....both wave forms are almost similar(with only a small spike of HF that is also observed on ground) , most SMPSs increase the noise significatly when you increase the bandwith.
Nirvana , like a linear PSU , shows the same noise no matter the bandwidth.
Next testing is casing and ground earthing , primary and secondary side...Noise will be reduced even further.
Have you checked to see of the noise is actually random or if it a pattern. Usually a pattern with switchers. Careful adjustment of triggering and waveform averaging may show more detail that what you can see now. Vertical amplifier bandwidth limiting to 20MHz is a setting many scopes have and can also help get noise down. Then again, maybe not to all that. Just some thoughts.
R U now considering to push out Nirvana and Katana at the same time?
...I'd be happy to see that.
You should even consider to offer both products as a combo!
cdsgames, for better waiting for Katana. Can you tell story, why Allo choosed ES9038Q2M as DAC chip. This thread have ESS9028Q2M in name , someone would like to see ES9038PRO on board. I also read some review of Allo Boss and inside he or she talked about next DAC with AKM AK 4497 chip. Can you explain your decision?
, someone would like to see ES9038PRO on board. I also read some review of Allo Boss and inside he or she talked about next DAC with AKM AK 4497 chip. Can you explain your decision? Ok get ready for a bedtime story..Why Katana ? Simple , I love Japan , its history and especially food (and got married many years ago to a beautiful asian woman) so I think of Katana Dac as an exquisite weapon .
We started with ak4497. That ic has the advantage of being in voltage output (IV stage is incorporated) but still it needs buffer/lpf opamp/trans stage..and while transfos are good, we think that a properly designed discreet stage + ultra low noise opamp PSU (dual rail 15V) is the correct direction (but much more complicated to get it right ). Problem is that Master Mode was not available..
Since we chose opamps stage and power, we started to look around at "other ICs"
Ess Sabre has lots of good reviews . So we chose the ess9028 (has master mode) and started the design and about halfway there , Ess Sabre advised us to take a look at ess9038q2m that had DOP capabilities . Ess9028 and 9038 differ in output impedance , but with a IV opamp stage (and virtual ground) it simply does not matter .
Now some people will say ess9038 sound better than ess9028 (or vice-versa). Its a bunch of b*** to me. They both sound great and what you hear is....
IMPLEMENTATION. (same implementation will sound different on ess9028 vs ess9038 so you must optimize for one)
Please dont think of Katana dac as 9038 Sabre DAC. Katana DAC is about "good" implementation (in my opinion) of a pretty good IC.
There are 4 main areas we focused.
1. Avcc has a Supercap decoupled on each channel (L and R) The same avcc was measured/tweaked over 75 times to arrive at noise in uV. Simply taking LDOs and datasheet info is far from being realistic. Using the best LDOs is (half) useful because measuring the output noise is never the same as datasheet. So we did the legwork..measured , tweaked ,remeasured . We changed the inductors at least 7 times because , again , datasheet and real measurement showed that s smaller inductor film was better than a bigger one (wirewound) (core matters) in lowering the noise. Now imaging doing this (measuring tweaking )on each important component...until we got it right (yeap I am a perfectionist..on some subjects)
2. Clocks and Master.
We are firm believers in master dacs. .since this is a master , BCLK and LRCLK lines are going out and noise present has no impact on the DAC IC. Data line itself is cleaned on our design.
Our clocks are cm away from the DAC IC. After we launched the Boss (master dac) and Kali (reclocker) we have developed a true art (if I may say so) for MCLK. MCLK was measured countless times and its the best we got so far in any of our designs. Oscilators (ultralow phase noise from NDK) have power cleaned using 2 LDOs in series..but it was very hard to find the right inductors in the LPF that did what datasheet said...(SRF in GHz) (yeap we have 2 LPF)
3 Analog Stage.
I said and will say it again. Sparkos Labs discreete opamps are something else. Amazing design by an analog wizard that far surpasses my knowledge . Class A output is the best in my books. While too power hungry for an AMP , for a DAC output...its perfect. Its hard to explain the sound, vocabulary , or words , are simply not made for it. So I will stop.
Still each Opamp stage has to be hand trimmed for offset voltages (6 of them) (we dont agree with DC servo) and then programed (mated to a specific controller board) for THD+N compensation.
All we did is to provide power to it. 15V dual rail , and we used LPF + LDOs on both rails to clean it to uV levels. Thats whats Opamp board drinks ... (me I prefer red wine)
There is one last area that we concentrated on, while not technical , its important. Allo designs have a great quality/price, money does not come from the sky and we understand that. Katana is more expensive but each cent of the 249$ (USA pricing) is well accounted for. SQ /price is better than the Boss.. (again my flawed personal opinion)
Now, you (all) tell me a story when you have the Katana in hand..
The story has one flaw.(and you guys know it)
No matter what we did, no matter how many filters , LDOs and caps we have in the path, you must thing about PSU.
POWER. (yes in capital)
There are 3 types of noises on each PSU
1. Differential noise
Simple. most of our filtering will take care of that. Still pay attention to it , do not bring PSUs over 20mV of DN. Internally the DAC iteself was desgined with each rail independent (low pass filters and LDOs , about 20 of them..)
2. Common mode noise
Internally most of our components are designed not to inject common mode noise back into the DAC, still your PSU has to be designed with CMN in mind. For example IFI psu has about 86mV-91 (20Mhz/2a testing) and a OEM about 100mV
3. Leakage current
Simple, if your PSU has no earthing (dc ground earthed) , your leakage current will be present on the DC output. No way around it and yes you can hear it.
So best design is not so great if your PSU is crappy.
No matter what we did, no matter how many filters , LDOs and caps we have in the path, you must thing about PSU.
POWER. (yes in capital)
There are 3 types of noises on each PSU
1. Differential noise
Simple. most of our filtering will take care of that. Still pay attention to it , do not bring PSUs over 20mV of DN. Internally the DAC iteself was desgined with each rail independent (low pass filters and LDOs , about 20 of them..)
2. Common mode noise
Internally most of our components are designed not to inject common mode noise back into the DAC, still your PSU has to be designed with CMN in mind. For example IFI psu has about 86mV-91 (20Mhz/2a testing) and a OEM about 100mV
3. Leakage current
Simple, if your PSU has no earthing (dc ground earthed) , your leakage current will be present on the DC output. No way around it and yes you can hear it.
So best design is not so great if your PSU is crappy.
Last edited: