PL611-01-N12 Alternative needed quite quickly

Technical questions regarding the xTIMEcomposer, xSOFTip Explorer and Programming with XMOS.
yes2i2s
Active Member
Posts: 60
Joined: Tue Nov 24, 2015 4:06 pm

PL611-01-N12 Alternative needed quite quickly

Postby yes2i2s » Wed Mar 01, 2017 10:17 pm

Hi XMOS community

We're just tidying up our first production prototype and have come to discover that the PL611-01-N12 used in the multi-channel development board is discontinued.

We cannot put our latest design to the assembler without a known-to-work replacement.

We need a part and basic schematic showing how we plug this into our design.

We need it pretty fast as we're doing the design amendments right now.

Please see the current schematic attached. The MCLK switch will be removed, this was for prototype purposes, we need to run off the PLL MCLK.

Thanks!

Tom
Image
User avatar
mon2
XCore Expert
Posts: 820
Joined: Thu Jun 10, 2010 11:43 am
Contact:

Postby mon2 » Thu Mar 02, 2017 6:46 am

Hi. We recall sourcing some of these parts for another developer last year through Arrow. Electrosonic is showing that they will have stock in April but quite confused on the part details. That is, if you source the PL611-01-N12 - will this part supply the same clock outputs as required by XMOS ? Best to confirm this with Microchip before sourcing.

From the XMOS documentation, this PLL generates 3 clocks,

24.00 Mhz on CLK0
24.576 Mhz on CLK1 if MCLK_FSEL is HIGH
22.5792 MHz on CLK1 if MCLK_FSEL is LOW

this means that the PL611-01-N12 is factory programmed to allow for the above values. Be sure to confirm this detail with any such supplier. MCLK_FSEL is selected by XMOS software.

To allow for backwards compatibility, you will need to insert a replacement circuit with the above features.

A suggestion is as follows:

1) use the Si5351A PLL which is available as a factory programmed part. You can request any value for the 3 output pins. However, there is a pre-programmed part available from Digikey for XMOS designs.

reference p/n Si5351A-B04486GT

viewtopic.php?f=7&t=4599&start=10

This part will generate 3 fixed clocks if you apply a local 25 Mhz crystal.

a) use the 24 Mhz clock as-is to replace the original PL611-01-N12

b) the other 2 clocks are 2 x the required frequency to replace the PL611-01-N12. For this reason, insert a D-flip flop to divide the clock by 2. You will need a single dual d-flip flop to support this new design.

Image

for example, use 74VHC74 - be sure to review the max frequency of such parts @ 3v3 voltage rail but ONSEMI version should be fine

https://www.onsemi.com/pub/Collateral/MC74VHC74-D.PDF

with the D-flip flop, you will now have the 2 required frequencies of 24.576 Mhz and 22.5792 Mhz.

c) finally, add a mux which will be selected by MCLK_FSEL (XMOS driven) to select the 24.576 Mhz or 22.5792 Mhz to your target device

The alternate idea is to contact SiLabs and request that they program for you a Si5351A to supply 24 Mhz, 24.2576 Mhz and 22.5792 Mhz using the OTP. This will then save you the cost of the 74HC74 flip flops.

The above idea should work as a valid replacement for the PL611-01-N12. Phaselink was purchased by Micrel and then Micrel was purchased by Microchip. The part appears to be available through Microchip but the version they show as a standard part is 10x the cost of the original PL611-01-N12 so cannot be the same. Electrosonic is showing ETA of April for the PL611-01-N12 and is under $1 USD.

If you are designing a new PCB, consider to have the footprint for the original PL611-01-N12 and the above design.

Disclaimer: We are not audio engineers but the above is a logical solution to the difficult to source PL611-01-N12. Perhaps someone from XMOS can confirm this thought process to move forward.
User avatar
mon2
XCore Expert
Posts: 820
Joined: Thu Jun 10, 2010 11:43 am
Contact:

Postby mon2 » Thu Mar 02, 2017 12:32 pm

New day, new idea. Could you not use 3 x fixed SMD oscillators and then select between the 2 different clocks using an external mux ? This would then replace the Si5351A as well.
yes2i2s
Active Member
Posts: 60
Joined: Tue Nov 24, 2015 4:06 pm

Postby yes2i2s » Thu Mar 02, 2017 3:11 pm

Brilliant feedback thanks. There are a couple of options there.

Re the oscillators, sounds like a nice simple idea, however I didn't think oscillators can be chucked in circuit like that, don't they require some some driving through a chip with an Xin / Xout? Any schematics for that? Both look doable just the first one takes a little more planning. I will try not to rely on the original PLL as sourcing issues could bite us in the ass at production time. I worked hard enough to source 5 from Canada last year through 3 hops. May have been us you helped.

Cheers!
yes2i2s
Active Member
Posts: 60
Joined: Tue Nov 24, 2015 4:06 pm

Postby yes2i2s » Thu Mar 02, 2017 3:26 pm

Also this diving the clock etc, is this going to dirty the jitter of the signal? being high end audio the MCLK quality is of extreme importance and makes a difference to the fine details in sound reproduction. The existing PLL does yield very acceptable audio performance.
User avatar
mon2
XCore Expert
Posts: 820
Joined: Thu Jun 10, 2010 11:43 am
Contact:

Postby mon2 » Thu Mar 02, 2017 3:40 pm

Hi. Yes we worked with you to source the few pieces from Arrow last year :) How are you ? Did they work out ok for the project ? If they did then the p/n confirms that the device outputs the 3 desired clocks. With this conclusion, if you can source through the pending stock from Electrosonic (under $1) then that is an option. Do contact Microchip to get their official status on this Micrel / Phaselink device. Perhaps the demand is not there so they are no longer listing this specific p/n ?

If you wish to proceed with the Si5351A idea, source the Adafruit PCB:

https://www.adafruit.com/product/2045

and the referenced XMOS device noted in this thread -> replace the Adafruit supplied PLL with the one from XMOS. Now, upon power up, the Adafruit board should output 3 clocks with 24 Mhz being spot on correct. Then add an external D-flip flop for each clock that needs to be divide by 2 and a mux to select either one via the XMOS select pin.

On the oscillator idea, you will use 3v3 fixed oscillators (ie. 3225 size or larger if you wish) and upon power up, the oscillator will output the exact 24 Mhz, etc. It appears that is what China is doing on their XMOS audio designs...consider to source one for confirmation:

zoom in on these pics and they are indeed using 3 fixed clocks and selecting between the 2 of the non-24 mhz source:

http://www.ebay.com/itm/XMOS-U8-Subcard ... Sw44BYMnOj

Each CMOS SMD oscillator should be around $ 0.30 USD (WTL Shenzhen, China) and is able to drive multiple loads, just like the original PLL would have done. One concern was about the clean and glitch-less switch over from one clock to another - will this lead to any issues ? audible plops / noise during the switch over ? it appears it is not an issue since the above ebay board is doing the same.

review the schematic from XMOS that features the original PL611-01-N12 device and note the mux in the design - apply the same into yours to select the 2 clocks. Just be sure to wire it up to match the logic of the XMOS selector.

That is,

if XMOS select pin is HIGH - then clock A is routed - review the XMOS docs for the clock value
if XMOS select pin is LOW - then clock B is routed
yes2i2s
Active Member
Posts: 60
Joined: Tue Nov 24, 2015 4:06 pm

Postby yes2i2s » Thu Mar 02, 2017 4:00 pm

Hi! We're doing okay, I lost my partner in a cancer battle 2 months ago, so we're only just picking this back up. Life throws canon balls but you must persist. Our prototype sounds excellent, went to the Bristol Hifi show just to spectate, our gear punches above it's weight so it's all looking good for the future :)

Audio popping is not an issue, the DAC we're using has elimination of any pops integrated.

Another solution, there are plenty of audio chips such as this: https://www.idt.com/document/dst/clock- ... le-systems

Perfect apart from a 48MHz processor clock. Can the XMOS x200 chip run from a 48MHz clock instead of 24? Either that or can it run direct from an oscillator?

Or, I could divide the 48 to a 24. Seems a better way around it and doesn't interfere with the audio clocks.
yes2i2s
Active Member
Posts: 60
Joined: Tue Nov 24, 2015 4:06 pm

Postby yes2i2s » Thu Mar 02, 2017 4:11 pm

Yes I even have an ebay DAC board at home with USB XMOS > oscillators through MUX > MCLK

May go for that, I wonder why the XMOS guys / rest of the industry don't do that though. I guess they like all the clocks coming from one part for simplicity.
yes2i2s
Active Member
Posts: 60
Joined: Tue Nov 24, 2015 4:06 pm

Postby yes2i2s » Thu Mar 02, 2017 4:15 pm

I think if the XMOS can run from 48MHz then I'll use a purpose built audio clock chip. If it can't I'll still do that but run the XMOS direct from the existing 24MHz osc on the current design.
yes2i2s
Active Member
Posts: 60
Joined: Tue Nov 24, 2015 4:06 pm

Postby yes2i2s » Fri Mar 03, 2017 12:19 pm

Does this complicate things:

I want to support 384KHz, I don't see it in the output options of this x200 development board even though it's enabled in the firmware (I changed max samplerates to 384). Does it require an additional clock frequency?

I want to support SPDIF as an input, what changes do I need to make to the x200 reference hardware / firmware to allow this and how does it operate?

Thanks again!!!

Return to “Development Tools and Programming”