Apple

Le 22 août 2016 à 15:25, Gordon Rankin <email@hidden> a écrit :

JC,

If you are using a TCXO then your best option is to go ASYNC mode and yes that will work with IOS and macOS as well as Windows and Linux. In ASYNC you are the total master and the rate will be perfect.

Knowing the TAS1020B, PCM2902 and some of the other older chips from products I made decades ago. These products only use the XO input to create a PLL Master Clock (and USB clock). The Adaptive rate is then based on the SOF from the HOST and that timing is matched to the internal clock which then changes the Master Clock and the sampling speed to match the host. So using a quality TXCO to accomplish a really good 48K sample rate is not really going to accomplish your goal. The true rate is going to be controlled from the host.

You could use the TAS1020B or TUSB3200, which goes out of production at some point. It is used in too many products to predict when. You would need the Kiel 8051/52 development system and probably the eval board. Probably a better idea would be a Microchip PIC32MX270 processor. The dev is cheap and they have sample code for some of the operation.

Thanks,
Gordon

On 8/21/16 3:57 PM, Jean-Charles Rousset wrote:
Thanks for your reply Gordon.

What we have been trying to do is to RECORD sounds in iOS/macOS with our USB device as the master clock (TCXO) and iOS/macOS devices as the slave in order to get perfect samples.
If I have understood correctly :
- It cannot be achieved with ASYNC IN, as the sink (iOS) will be master.
- Of course, it cannot be achieved by SYNC mode.
- In ADAPTIVE mode, iOS should adapt its rate to what the USB device send.

Maybe I understood it wrongly, but then, I have no idea how to send audio to iOS with iOS being clock slave.

As I’m not recording audio for music but for timing measurement tool, I don’t need an ultra-high-quality audio. 16 bits / 48KHz is totally ok. But these 48KHz must be absolutely perfect (48.0000KHz..)

Thanks !

JC

Le 21 août 2016 à 21:26, Gordon Rankin <email@hidden> a écrit :

JC,

Adaptive of course will add jitter to either the input or output device because of it's flow control being that of changing the Master Clock. Which in turn changes the speed of the clocked device (I2S, L/RJ whatever). Parts like the PCM2902 and really early TI parts were done (even early TAS1020 code) before anyone really could take a look the consequences of the protocols used.

The XMOS software only works in ASYNC mode. You can use input, output both etc...

Maybe a better approach maybe to tell us what you are trying to accomplish and maybe we could help you look for a better solution. Price or not, ASYNC or SYNC is a much better ADC way than Adaptive.

Brian is correct on the Windows front, unless you are only trying to accomplish Full Speed UAC1 type products.

Thanks,

Gordon

--
J. Gordon Rankin
Owner and Chief Scientist
====== Wavelength Audio, ltd ======

References:
	>Re: Coreaudio-api Digest, Vol 13, Issue 115 (From: Gordon Rankin <email@hidden>)
	>Re: Coreaudio-api Digest, Vol 13, Issue 115 (From: Jean-Charles Rousset <email@hidden>)
	>Re: Coreaudio-api Digest, Vol 13, Issue 115 (From: Gordon Rankin <email@hidden>)