time-nuts@lists.febo.com

Discussion of precise time and frequency measurement

View all threads

Re: pulling some crystals

HM
Hal Murray
Mon, Dec 18, 2023 3:54 AM

But - one cannot use DDS generated clocks for high fidelity ADC/DAC from the
FPGA outputs because the jitter on those generated clocks due to various
internal processes is too high. Needs a crystal clock (which si how I do it)

If you have an FPGA, I assume you are doing some sort of DSP.  Can you tweak
that software to compensate for the crystal frequency being a bit off?

--
These are my opinions.  I hate spam.

glenlist@cortexrf.com.au said: > But - one cannot use DDS generated clocks for high fidelity ADC/DAC from the > FPGA outputs because the jitter on those generated clocks due to various > internal processes is too high. Needs a crystal clock (which si how I do it) If you have an FPGA, I assume you are doing some sort of DSP. Can you tweak that software to compensate for the crystal frequency being a bit off? -- These are my opinions. I hate spam.
GE
glen english LIST
Mon, Dec 18, 2023 4:10 AM

"

On 18/12/2023 2:54 pm, Hal Murray wrote:

If you have an FPGA, I assume you are doing some sort of DSP.  Can you tweak
that software to compensate for the crystal frequency being a bit off?

"

not without high logic cost (sample rate converter (SRC) with 120dB
spurless performance over the whole 0 to fs spectrum.... running at 390
MHz working over 15% of anywhere inside 0 to  Fs). and high power
cost....heat.... .  that's non trivial !!!

it's easier having a good clock... Remember- this problem is currently
SOLVED -  Currently I use a 24.576 off the shelf high quality TCXO for
this job. I am merely bumping this up to 98.304 MHz to improve the phase
noise of the synth driving the ADC and DAC.

 I can easily deal with sample rate (SR) errors down at say 40 Msps,
but the SRC is complex because I need 0 to 0.45 Fs of workable
performance, so that's a big long on the fly computed polyphase filter
OH with LOTS of latency....  and down there, everything needs to run at
powers of 2 clock speed, so I cannot just use my 10 MHz TCXO/GPS  clock
driving the FPGA ...Cant run the FPGA from the 10 MHz clock- need to use
that for countre gating- I'd need to make 50 MHz from my 10 MHz  to feed
the FPGA . One of those SiLabs clock chips could do that I guess for $5
and 16 mm2...

anyway, I will let you guys know hopw I go once I am gone back to the
oscillator.

thanks for all the input.

-glen

" On 18/12/2023 2:54 pm, Hal Murray wrote: > If you have an FPGA, I assume you are doing some sort of DSP. Can you tweak > that software to compensate for the crystal frequency being a bit off? > " not without high logic cost (sample rate converter (SRC) with 120dB spurless performance over the whole 0 to fs spectrum.... running at 390 MHz working over 15% of anywhere inside 0 to  Fs). and high power cost....heat.... .  that's non trivial !!! it's easier having a good clock... Remember- this problem is currently SOLVED -  Currently I use a 24.576 off the shelf high quality TCXO for this job. I am merely bumping this up to 98.304 MHz to improve the phase noise of the synth driving the ADC and DAC.  I can easily deal with sample rate (SR) errors down at say 40 Msps, but the SRC is complex because I need 0 to 0.45 Fs of workable performance, so that's a big long on the fly computed polyphase filter OH with LOTS of latency....  and down there, everything needs to run at powers of 2 clock speed, so I cannot just use my 10 MHz TCXO/GPS  clock driving the FPGA ...Cant run the FPGA from the 10 MHz clock- need to use that for countre gating- I'd need to make 50 MHz from my 10 MHz  to feed the FPGA . One of those SiLabs clock chips could do that I guess for $5 and 16 mm2... anyway, I will let you guys know hopw I go once I am gone back to the oscillator. thanks for all the input. -glen