I can't find anyone other than Analog Devices that makes commodity DDS synthesizer chips. I wonder why.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement
Intersil
I don't know what you mean by "commodity", but at least Harris has made quite interesting chips for a while.
OK, that's interesting. $37 each, none available.
ADI seems to own the business.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement
Dunno. I'm tentatively standardizing on the AD9956, because it has quite a nice auxiliary PLL capability and really doesn't cost much more than the AD9952, otherwise similar but without the PLL.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
'cause it's really easy to whomp one up on the digital side, and most systems that would call out a DDS have an FPGA in there anyway?
I can certainly see an ongoing demand from the instrument segment of the market for DDS-ish things, and maybe for high-end radios, but I don't think the market is otherwise very large.
-- Tim Wescott Wescott Design Services
Harris Semiconductor became Intersil in 1999. They make the DDS chip that Digikey doesn't stock.
Commodity means to me that it's available at a sensible price, something that I would reasonably design with.
I want a low-jitter tunable clock from 0 to 25 MHz. A 12-25 MHz DDS and a binary divider chain would do it. ADI has some chips that would work, around $5 each, so that's OK. I could do it myself with an FPGA and a DAC, but that's hardly worth the effort. The lowpass/bandpass filter is the real nuisance, and nobody seems to want to integrate that.
I guess I should learn more about RF synthesizer chips, which everybody is making these days.
Just curious that TI and such aren't interested in the DDS biz.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement
Assuming the FPGA balls are available, an ADI DDS chip and a fast DAC are about a wash on price.
Probably not.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement
Maxim has the DS1077 series they inherited from Dallas Semi, but I believe I recall your feelings about Maxim.
-- ----Android NewsGroup Reader---- http://usenet.sinaapp.com/
That's a divide-by-N so it has chunky frequency resolution. And the internal oscillator is pretty awful.
So, I can continue to hate Maxim.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement
Maybe because the market is saturated? If the ADI chips do what you want, why not use them?
Given that your designs often have FPGAs in them, a DDS simplifies to a DAC of the required specifications and a filter that should be well inside your comfort zone.
I dug into the DDS thing some time back and found I could design the digital circuit for a very respectable DDS with spurs below -100 dB. Does this fit your definition of "low-jitter"? The only real limitation is the DAC - assuming an analog signal is needed.
BTW, most of, if not all the single chip DDS designs truncate the phase word which produces significant close in spurs. You may get the filter for free, but you pay for it in higher close in spurs.
I'm not sure how you get a 0-25 MHz clock from a 12-25 MHz DDS. Are you thinking in terms of a sine wave feeding a comparator followed by a selectable binary divider? Then the circuit would be FPGA => DAC => Filter => FPGA => clock.
-- Rick
I haven't looked at all the DDS chips available, but all the ones I have seen use a limited size table lookup with significant phase word truncation. This gives higher close in spurs than are desirable for many applications such as radios and low jitter clocks.
-- Rick
...the American Dental Association is limiting competition?
I guess you have read this about the AM noise problem with the newer ADI DDS chips that are missing the DACBP pin:
One easy way to characterize close-in phase noise is to use two of the DDSs under test, mix them, lowpass, and look at the spectrum of that difference. That doesn't need an extreme spectrum analyzer or a super-good reference. They can use the same clock, or different clocks, depending on what you care about.
People do that with really good OCXOs too, where the XO is better than any available spectrum analyzer. Just make sure they don't try to lock.
You can do the same in the time domain. We make digital delay generators and may want to see what the jitter is like, say, 1 second after a trigger. The easy way to do that is to trigger two independent units, one set to delay 1 second and the other to dalay, say,
1.000000010 seconds. Then any decent scope can see the jitter between the two output pulses. Sort of a super self zoom.-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers
Yeah, DDS searches turn up a zillion dentists.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers
They pretty much do. OTOH I don't believe that market is very large. Radios often use PLL and lower frequency stuff comes out of computer thingies in waveform files that are sent to a DAC. So that leaves a fairly small frequency range where DDS can play. Mostly function generators and such.
For example, for a new product design I need a phase-adjustable generator from zero to about 15kHz. It'll be a simple codec, no DDS. But the codec is also from Analog Devices so they don't lose that revenue.
-- Regards, Joerg http://www.analogconsultants.com/
Interesting, thanks. I always roll my own vregs for analogue stuff anyway, but not being able to bypass the internal reference is pretty stupid, I agree.
Mostly I picked one of the 400-MHz units because it's a pain to supply a really low noise clock faster than that, so I might as well put the PLL on the DDS output instead.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
You mean of course other reasons than the (at least) 188 patents hold by ADI on DDS technologies ? (number found with search tag "inassignee:"Analog Devices, Inc." direct digital synthesis" on GooglePatents
Friendly, Robert
I put a new power supply on my Efratom LPRO-101 rubidium standard ($100 from eBay some years ago), and wired it up to all the reference inputs on my synthesizers, counters, and analyzers (at least the ones that fit in the rack at the moment). One of my HP 3325As has the ovenized oscillator (a 10811A, I think). It still agrees with the Ru reference to way within 1 Hz.
That's how you usually measure stabilized lasers as well--beat two of them together, measure on an RF spectrum analyzer, and divide by sqrt(2). With three, you have enough measurements to characterize each one individually.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.