NDK low jitter clocks, for digital audio.

We are able to supply NDK NZ2520SD clocks, at 22.5792 MHz and 24.576 MHz.

All units are burned in, for 24 hours, and the phase noise is measured. The phase noise measurement is supplied for each unit.

See our eBay listings, or inquire, as to price and availability.

The NDK NZ2502SD clocks have gotten some attention as "low noise", "low phase noise", and/or "low jitter". In general, they are good parts. But, like any other part of this nature, they are not 100%.

Yes, on any given reel, there good ones. Even some very good ones. Some that are just ok. And some that belong in the trash. Sorry, folks, that is the plain truth. Which is why several manufacturers pay us to buy pre-screened, and sorted, parts from us. Since we have a sufficient quantity on hand, we can offer these to parties that only need one or two of them. (Of course, if you are a manufacturer, we would love to add you as a customer.)

So, buying one, from any supplier, only guarantees you will get one that oscillates. If you want one that really lives up to the hype, well, we are probably the only place to buy one. But, your money, your choice. You can take your chances with some other supplier, of for about the same price, you can buy one that comes with proof of performance.

All of the units we sell will out-perform the more expensive Crystek CCHD-957. Even at our lower priced units. For those of you who demand the best, and are willing to spend roughly the same amount, we can supply units that are in vicinity of 15 dB better.

Yes, you read that right: 15 dB better.

So, you can buy an expensive competitor's brand, or......
Buy this brand, from someone else, with no proof of performance, or.............
Buy one, with proof of performance, from a company with a background in designing, building, and measuring, low phase noise clocks.

The choice is yours. We hope you make the right choice.

Folks have asked us if we really have parts that are "15 dB better than Crystek". We do. Here is proof:

Since Crystek does not supply data at 1 Hz, we have to extrapolate what it would be, from the data they do provide. (Of course, we have measured lots of them, so we know how close to the truth we are.) In any case, the above plot shows the best performance we have found from the NDK NZ2520SD parts. So, what are you waiting for? Buy some, now! (OK, not really "now". Someone just bought those 4, and we will not have not any more, until we complete another round of sorting, for one of our commercial customers. Which means there will not be any more until July, 2014. If we are all lucky.)

Here it is..........................2017...............yeah, forget that date! Besides, we are probably switching to a different grading system, that is less confusing, as it does not rely on one phase noise number, at a single frequency. (No, we are not going to explain it. We get enough grief, as it is, from so-called "experts" who think phase noise below even 10 Hz is meaningless. Let's just say it is a series of parameters, that need to be met, in order to get the highest grade.) (The new method will be faster, and possibly less expensive! Mainly because the phase noise plot will be an option.)

"Yeah, but if you are going to do something different, how are we supposed to know how good they are? Especially if you think we are going to pay extra for a stinking plot, you fascists."

Fair points. One, it takes 2 hours to make one of those plots. Ok, it isn't like we sit there, for the whole 2 hours, staring at the screen. No, it just means we can not do anything else, while that machine is making that plot.

"Yeah, but we all know that you run those, overnight, and start the test as you are walking out the door, to go home, for the night. You even admit as much!"

Yeah, we do! But, all of that is fixin' to change, some time next year. (No, don't ask why.)

Anyway................uh..........oh, yeah, point #2.

Picking a clock, based on one phase noise number, at only one frequency really isn't the best way to do things. At least that is what we have concluded, after doing this for far too much time, for the last several years (because the manufacturing side of the company has been doing diddly-squat, we have to do something to keep us all afloat, to make up for those slackards!) But, since you want to know...............

Right now, my guess is the highest level will have a 10 Hz number somewhere near -113 dBc. The other grades might go as low as -111 dBc. So, there really isn't a lot of difference. But, since we will be able to test them faster, we can bring the cost down somewhat. Of course, a lot of what you are paying for is the X% that go in the trash. If you knew how high that percentage is, you would not complain. Not to mention how much all the test equipment costs. I could personally buy a really nice car, for the $$$ we have tied up in test gear, for this stuff. Lucky for you guys, we are all old and cranky, and have lost interest in cool cars. Some day, you will be our age, and understand why we are so miserable to deal with!

Crystek-style adapters

Some of you have asked for adapters, that you can use to drop one of these onto a PCB that was designed for Crystek parts. Rather that get into that business, we thought it would be better if we supply you a link, to where you can order one, designed by us. (Yes, if you don't like it, you can cuss us out. We are used to it, by now!), So, here is the link:

Order from OSH Park

Hope you like them, and they work out fine.

If there is a different footprint, that you desire, let us know, and we will look into providing a file for one. The guys at OSH Park (who we do not know) provide a good service, at a good price, for the DIYer. We use them for proto work, so we can safely say they are "OK".

Contact us at:


©2014 Analog Research