Home |
Search |
Today's Posts |
#1
|
|||
|
|||
DDS and phase noise.
I'm not an expert and need some help.
I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro |
#2
|
|||
|
|||
"Mauro" wrote in message
news So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Decreasing the frequency at which you clock the DDS will increase the phase noise. It is pretty hard to estimate as it will change quite a bit with small changes in divisor (i.e. output frequency). But as a general rule, you want to clock the DDS at some high multiple of the operating frequency. Obviously, if your clock has phase noise too, that is a bad thing. I would doubt that your frequency doubler would add phase noise, but I can't say for sure since I haven't got your circuit. But typically doublers introduce other forms of distortion which aren't as problematic for a DDS clock. So just shooting from the hip, I would expect your phase noise to be a lot worse from a 200 MHz clock than from the 400. It is still 10x the operating frequency, though, so I wouldn't expect it to be horrible. ... |
#3
|
|||
|
|||
In article , "Mauro" wrote:
I'm not an expert and need some help. I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro There are 2 main sources of phase noise in the output of the DDS: The residual noise of the DDS and the divided down noise of your clock source. The residual noise is inherent to the DDS and not dependent on the clock. It is a function of the technology used to build it. This should be specified by the vendor. It is usually measured by driving 2 identical DDS's from a common source, measuring the resulting phase noise and subtracting 3 dB to account for 2 devices. As for your clock noise: Your 200 MHz clock noise will be increased by 6 dB by doubling [plus any residual noise of the doubler -- use a Schottkey diode doubler for best results]. That 6 dB will be subtracted by the extra factor of 2 in the effective divide factor of the DDS. In short -- its a wash except for the doubler residual noise reduced by 26 dB [20*log10(20/400)] which should be insignificant. Dr. G. |
#4
|
|||
|
|||
"Mauro" writes:
I'm not an expert and need some help. I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro Assuming your source is not so good as to be below the noise floor of the AD9951, you will get slightly better performance without multiplying by two. The rule-of-thumb for perfect conversion is 6dB for each doubling. But it won't be perfect, so by doubling you'll be a little worse than 6dB and then dividing again you will not get as much back. If the doubling was perfect and the performance of the AD9951 was the same at 200MHz and 400MHz, you get the identical results. Of course, this might be bogus if there are other artifacts. Make sense? If you can build a doubler, then you can measure phase noise. Mix your 20MHz output with a known good 20MHz source in a double-balance mixer. Adjust the phase to get zero DC (quadrature). If you low-pass filter the output, the noise is an indication of phase noise. If you have a low frequency spectrum analyser available, you can get a nice plot. Otherwise you can use an AC voltmeter to get relative performance. Steve. -- Steven D. Swift, , http://www.novatech-instr.com NOVATECH INSTRUMENTS, INC. P.O. Box 55997 206.301.8986, fax 206.363.4367 Seattle, Washington 98155 USA |
#5
|
|||
|
|||
On Wed, 14 Sep 2005 20:29:09 GMT, "Mauro" wrote:
I'm not an expert and need some help. I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro I won't be to concerned with phase noise, with that dds I'd be more worried about spurs. Those AD dds chips are notorious for them. |
#6
|
|||
|
|||
From what I've seen in the data sheets, the closer you operate the RF output
to the Nyquist limit (Fref/2), the cleaner the spurs get. I've seen numbers of -90 dB or better on even some of the cheapo Analog Devices parts. Joe W3JDR "maxfoo" wrote in message ... On Wed, 14 Sep 2005 20:29:09 GMT, "Mauro" wrote: I'm not an expert and need some help. I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro I won't be to concerned with phase noise, with that dds I'd be more worried about spurs. Those AD dds chips are notorious for them. |
#7
|
|||
|
|||
Thanks to all who replied.
I think that the net is that the phase noise should remain about the same. The source is a low noise quartz oscillator based on J.Stephensen design, followed by a Schottkey doubler and a MAR5 amplifier. I was in doubt to use or not the Schottkey doubler. I think i'll keep the simplest solution: no doubler. DDS max output will be 39MHz. A reference clock of 200MHz is still about 5 times the max output freq, high enought above the Nyquist limit. Thanks again. 73 Mauro i2SUH "Mauro" wrote in message news I'm not an expert and need some help. I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro |
#8
|
|||
|
|||
On Wed, 14 Sep 2005 20:29:09 GMT, "Mauro" wrote:
I'm not an expert and need some help. I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro The AD9951 has a much better spur performance than earlier DDS chips, and I have been tempted to use one for an LO in a radio, it looks clean enough to work nicely. ============== Generally, I'd say your phase noise will be the same either way, and perhaps even a little quieter with the 200 MHz clock. The internal reference multipliers probably add a little phase noise, multipliers usually do add a little extra noise, not than the normal 20log(N) one might expect from theory. Reference multiplers DO allow running the DDS output to a higher frequency, which is often a major desing goal, but if all you want is 20 MHz output then you are fine with a 200 MHz reference. Jim Pennell N6BIU |
#9
|
|||
|
|||
I'm not going to use internal multiplier. I'm driving it at 200MHz directly.
I agree that internal multiplier is adding noise. You can see that from the datasheet. Checked with an AOR7030 receiver the output seems to be quite clear and sharp. Not a real measurement, but better than some other PLL. 73 Mauro "Jim" wrote in message ... On Wed, 14 Sep 2005 20:29:09 GMT, "Mauro" wrote: I'm not an expert and need some help. I've build a DDS around an AD9951. I drive it at 400MHz from a source at 200MHz multiplied by 2. I get an output freq of 20MHz. I now drive it directly from the same source at 200MHz. I changed the register inside DDS to still get 20MHz on output. I do not have any possibility to measure the phase noise. So the question: is it possible to estimate if the phase noise of the 20MHz output is getting better, worst or is it going to remain the same? Thanks for the support. 73 Mauro The AD9951 has a much better spur performance than earlier DDS chips, and I have been tempted to use one for an LO in a radio, it looks clean enough to work nicely. ============== Generally, I'd say your phase noise will be the same either way, and perhaps even a little quieter with the 200 MHz clock. The internal reference multipliers probably add a little phase noise, multipliers usually do add a little extra noise, not than the normal 20log(N) one might expect from theory. Reference multiplers DO allow running the DDS output to a higher frequency, which is often a major desing goal, but if all you want is 20 MHz output then you are fine with a 200 MHz reference. Jim Pennell N6BIU |
#10
|
|||
|
|||
"Mauro" wrote in message ... I'm not going to use internal multiplier. I'm driving it at 200MHz directly. I agree that internal multiplier is adding noise. You can see that from the datasheet. Checked with an AOR7030 receiver the output seems to be quite clear and sharp. Not a real measurement, but better than some other PLL. 73 Mauro The AOR 7030 uses a PLL. It has a dual tank VCO as G3SBI came up with. W4ZCB |