RABAS Technologies - Allo products

[Tim Curtis]

I note that the new Rabas website only lists the 44/48 Mhz version of the Kali reclcoker, not the 22/24Mhz version required for multibit DACs like the TDA1387  used in the protoDAC.  Hopefully they will produce that version again too.

Are you sure about that?

I thought the 22/24 clock version was made for an OEM special order run and that the 44/48 clock version was the default shipping version.

I have a couple original Kali's which I believe use the 44/48 clocks and they work fine with ProtoDAC.

How would I determine which clocks are on the boards?

[zmk88]

That's the information stated on the Audiphonics website if you scroll down to compatability,

https://www.audiophonics.fr/en/accessori...11433.html

Regarding. determing which clocks are on the board, the only way, I think, is to read the text printed on the oscillators themselves.

[Tim Curtis]

The part number for the 44.1 clock on one of my boards is 45.158M which I think indicates the 44/48 clock set but I'm not an expert in electronics. Maybe @hifinet can have a look at his Kali board and determine what clocks are used.

What I would suggest is contacting Ravi via their "Contact Us" form and ask about the clock sets and whether they can produce a run of the 22/24 boards. 

45.158M
https://www.digikey.co.il/en/products/de...jRBKquG4QC

Manufacturer Product Number
NSC5083D-45.158M

Description
OSC XO 22.45.158MHZ CMOS 3.3V SM

Detailed Description
45.158 MHz XO (Standard) CMOS Oscillator 3.3V Standby (Power Down) 4-SMD, No Lead

   

[TheOldPresbyope]

It isn't clear to me what's going on with the 22/24 board and IMHO the Allo/Audiophonics documentation for both boards is terrible.There's folderol about 16/20/24 bit 192 kHz DACs vs 24/32 bit 394 kHz DACs but no specific explanation about why two different versions of the reclocker are necessary. 

I can play games with numbers and ratios, I know the i2S spec, and I have a general idea what the circuitry on the reclocker does with the I2S signals, but that's a long way from knowing why Allo felt it necessary to produce two different SKUs.

Regards,
Kent

[Tim Curtis]

It isn't clear to me what's going on with the 22/24 board and IMHO the Allo/Audiophonics documentation for both boards is terrible.There's folderol about 16/20/24 bit 192 kHz DACs vs 24/32 bit 394 kHz DACs but no specific explanation about why two different versions of the reclocker are necessary. 

I can play games with numbers and ratios, I know the i2S spec, and I have a general idea what the circuitry on the reclocker does with the I2S signals, but that's a long way from knowing why Allo felt it necessary to produce two different SKUs.

Regards,
Kent

Right, the Audiophonics documentation is a challenge to sift through but the OP suggests that ProtoDAC which is based on TDA1387 won't work with the 44/48 Kali and this is sort of is what Audiophonics is saying in their compatibility notes.

But it looks like my Kali has the 44/48 clock set and it works fine. I sent a note off to @hifinet to see if he can provide  some insights.

My memory is pretty hazy on the Kali clock discussion from many years back.

[hifinet]

I note that the new Rabas website only lists the 44/48 Mhz version of the Kali reclcoker, not the 22/24Mhz version required for multibit DACs like the TDA1387  used in the protoDAC.  Hopefully they will produce that version again too.

My Allo Kali version operates at 384kHz, so it must be the 44/48 version. It works with ProtoDAC.

ALLO KALI Reclocker I2S FIFO Sparky SBC and Raspberry Pi 22/24MHz 192kHZ 

ALLO KALI Reclocker I2S FIFO Sparky SBC and Raspberry 44/48MHz 32bit 384kHZ

Using Internet Archive, the first archived web page with Allo Kali appeared on September 22, 2016, Kali could be ordered in two versions 22/24MHz or 44/48MHz. Features state "I2S output: 44.1 KHz, 48 KHz, 88.2 KHz, 96 KHz, 176.4 KHz, 192 KHz - 16bit, 24bit or 32bit".

Allo Kali Sept 22, 2016

The clock option disappears in the next archive page on October 23, 2016 . The features change to: "With 44/48Mhz Oscillators".

[zmk88]

Thanks for debunking the misinformation on the net.  I specifically looked for a 22/24 version that should arrive tomorow.  I don't play files beyond 96/24 resolution so that suits me fine.  I look forward to trying it with the protoDAC kits you sent me via the US.

[hifinet]

The 22/24 version is reported to have lower phase noise compared to the 44/48 version. Jitter is higher with 44/48, but probably not audibly significant.

I came across this post regarding possible incompatibility of 22/24 with Moode 8.

[TheOldPresbyope]

I'll be happy to be shown I'm talking through my hat, but it seems to me that this is close to much ado about nothing.

I decompiled the protodac driver. Looks like the only thing it does is connect to I2S.

As for I2S, the only signals it communicates to the DAC, hence to the Kali Reclocker, are the serial clock (aka the bit clock), the word select (e.g., left or right channel), and the serial data (e.g., the bits).

AIUI, the reclocker strips out the serial clock from the incoming I2S stream. Its I2S output stream consists of a clean clock of the same frequency with the word select and serial data synced to it.

What this would mean is that the only thing which matters for the reclocker is that it can match the incoming serial clock frequencies. These are set by the PCM material being streamed---16 bit/44.1 kHz, etc.

By my calculations, a 45.1584 MHz oscillator can be the basis for all the 16 bit to 32 bit samples at 44.1 kHz to 358.2 kHz sampling rates if one divides the oscillator output by a factor from 2 to 32. Likewise, a 49.152 MHz oscillator would cover the multiples of 48 kHz sampling rates. I assume the 44/48 nomenclature is just marketing simplification.

Similarly, a 22.5792 MHz oscillator can be the basis for all the 16 bit to 32 bit samples at 44.1 kHz to 176.4 kHz sampling rates if one divides by a factor of 2 to 16. Likewise, a 24.576 MHz oscillator for multiples of 48 kHz sampling rates. Hence, the 22/24 nomenclature.

I see no reason why the 22/24 board would not function correctly with moOde on a Pi unless one selected incorrect options.

Regards,
Kent

[Tim Curtis]

That sounds correct.