Hi ...
Any of you tried these (OCXOs):
IQD OCXOs (overall product page)
http://www.iqdfrequencyproducts.com/products/details/iqov-22-1-03.pdf (specific product with Vsupply 5 VDC).
Phase noise figures are:
Phase Noise @ 10MHz (typical):
–125dBc/Hz @ 10Hz
–145dBc/Hz @ 100Hz
–155dBc/Hz @ 1kHz
–160dBc/Hz @ 10kHz
–160dBc/Hz @ 100kHz
and overall 5 ppb frequency stability ...
Don't know the price, though.
Greetings,
Jesper
Any of you tried these (OCXOs):
IQD OCXOs (overall product page)
http://www.iqdfrequencyproducts.com/products/details/iqov-22-1-03.pdf (specific product with Vsupply 5 VDC).
Phase noise figures are:
Phase Noise @ 10MHz (typical):
–125dBc/Hz @ 10Hz
–145dBc/Hz @ 100Hz
–155dBc/Hz @ 1kHz
–160dBc/Hz @ 10kHz
–160dBc/Hz @ 100kHz
and overall 5 ppb frequency stability ...
Don't know the price, though.
Greetings,
Jesper
OC = Oven Controlled = $$$! Plenty of brands of awesome OC clocks, but you have to be willing to pay hundreds of dollars per piece...
Still trying to find a reasonable source for 50-100 pcs. each of the NDK NZ2520SD (not SB or SA) in 22.5792 and 24.576mHz. Waiting on a quote from Dove. Don't mind a several month lead time, but can't buy 1K pieces.
Anybody know of a distributor or reseller with stock of the above?
Thanks!
Since this thread has already been resurrected I thought I'd go a bit further.
I'm awaiting a Tent XO for my Shigaclone. I already have a dedicated Per Anders SSR01 super-regulator built to feed the clock (it will be isolated from the other 5V supply to the DSP). Since the SSR01 is a low noise, low output impedance supply I've been wondering about the size of the matching cap. Guido recommends a 100nF NPO, but I'm not sure if that's assumed to be with a "standard" supply. I've read some things that suggest if I'm using a dedicated low noise reg that I could go with a smaller cap, and maybe be able to use a (as I'm led to believe) better silver mica type. Can using a dedicated super-reg like this allow for a smaller value matching cap to be used?
In my case, Chip1Stop web shop
????????? - ??????????????
in Japan is the only source. They provide an over-seas handling.
When I requested a quote a few days ago, their offering were;
For 1 piece : 1,500 JPY/each (approximately 15 USD)
For 20 pieces: 990 JPY/each(approximately 10 USD)
For 51 pieces: 786 JPY/each(approximately 8 USD)
Expected delivery was 7 days after an order placement.
If you have any problems, please let me know. I may help you.
bunpei < at > ta2 < dot > so-net < dot > ne < dot > jp
By the way, NDK shows a phase noise measurement result of NZ2520SD, 22.5792 MHz ( In Japanese).
You can compare those values with those of Crystek CCHD-957.
http://www.crystek.com/crystal/spec-sheets/clock/CCHD-957.pdf
Hi Bunpei & Superdad,
I'm interested in buying some of these clocks (NZ2520SD) as well, however, am in the process of assembling a list of items to order from chip1stop. If possible, Bunpei, I'd really appreciate your help in this (and this time won't order from you until I'm ready to buy).
Additional to the 24.576 & (possibly) the 22.579 MHz, I'd also be needing a 45.158 MHz clock.
BTW I've been realizing that the NZ2520SD appears to be a quite special crystal on more levels - the phase noise - and then also because it is as small as it is. May potentially reduce track inductances significantly (which I assume is desired).
'Best for your weekend ;-)
Jesper
I'm interested in buying some of these clocks (NZ2520SD) as well, however, am in the process of assembling a list of items to order from chip1stop. If possible, Bunpei, I'd really appreciate your help in this (and this time won't order from you until I'm ready to buy).
Additional to the 24.576 & (possibly) the 22.579 MHz, I'd also be needing a 45.158 MHz clock.
BTW I've been realizing that the NZ2520SD appears to be a quite special crystal on more levels - the phase noise - and then also because it is as small as it is. May potentially reduce track inductances significantly (which I assume is desired).
'Best for your weekend ;-)
Jesper
@tonyptony:
My guess on this would be to know how the regulator performs (potential oscillation tendencies, frequency response etc.) in the practical circuit with the Tent XO. Part of this could be to measure/listen to what happens when you insert the different capacitors.
Also you may consider track lengths and widths as 1 nH of inductance means ~ 62 mOhm of impedance at 10 MHz - a value higher than many capacitor's intrinsic impedance.
(If this is interesting to you there's a microstrip calculator here:
Microstrip Impedance | Electrical Engineering Tools | EEWeb)
That is to say that at high speeds distances matters and depending on the actual wire/track design 1 nH may correspond to as little as a 1 mm length of wire/track. So being able to place the capacitor close to the XO may also have some say in how efficient the decoupling capacitor will be (including the length and distances between the capacitor's leads which may also have significant inductances).
Best regards,
Jesper
My guess on this would be to know how the regulator performs (potential oscillation tendencies, frequency response etc.) in the practical circuit with the Tent XO. Part of this could be to measure/listen to what happens when you insert the different capacitors.
Also you may consider track lengths and widths as 1 nH of inductance means ~ 62 mOhm of impedance at 10 MHz - a value higher than many capacitor's intrinsic impedance.
(If this is interesting to you there's a microstrip calculator here:
Microstrip Impedance | Electrical Engineering Tools | EEWeb)
That is to say that at high speeds distances matters and depending on the actual wire/track design 1 nH may correspond to as little as a 1 mm length of wire/track. So being able to place the capacitor close to the XO may also have some say in how efficient the decoupling capacitor will be (including the length and distances between the capacitor's leads which may also have significant inductances).
Best regards,
Jesper
This one is a good calculator...
Saturn PCB Design - PCB Via Current | PCB Trace Width | Differential Pair Calculator | PCB Impedance
Saturn PCB Design - PCB Via Current | PCB Trace Width | Differential Pair Calculator | PCB Impedance
Thank you so very much! That is extremely helpful, and the prices are very encouraging.
I sure hope that phase noise graph is based on real world measurements and not some wishful ambition of the NDK marketing department. First time I have seen a vendor publish a curve going down to 1Hz offset.
Best regards,
AJC
Hi Jens & Superdad,
... what are your time perspectives on getting these oscillators? I would be interested in ordering shortly - within a month's time. Is that convenient to either of you (I can see, Jens, that you are Danish so maybe we should consider ordering together)?
Greetings,
Jesper
... what are your time perspectives on getting these oscillators? I would be interested in ordering shortly - within a month's time. Is that convenient to either of you (I can see, Jens, that you are Danish so maybe we should consider ordering together)?
Greetings,
Jesper
if there is a GB to get the 45.158 and 49.152, I may be interested, but will see if thats possible. I dont see that the track inductance of the decoupling cap->clock at VHF is very important for a clock, its pretty much a constant load. the capacitor package inductance to ground on the other hand is more important for rejecting noise. this idea that clocks demand super low impedance regulators doesnt make much sense to me. I would be more likely to decoupling it well, add LCR filter or something and a very low noise supply, not necessarily low impedance
I agree with qusp that reg impedance isn't important (should be low noise). You won't find a reg that is fast enough (low impedance at HF) anyway. Clock needs current during the transitions (h/l, l/h) and rise time is 1-3ns for typical clocks (100MHz + harmonics -> up to GHz). Thus the cap (typically 10nF COG) close to the XO will provide this currents and layout is critical here. The small form factor of the NDK XOs helps but isn't that much fun to solder by hand
Also interested in 49.152MHz, so we might have some to order.
Also interested in 49.152MHz, so we might have some to order.
Some useful calculations:
Knee frequency, ie the harmonics content of the square wave you should care about, this is also the indicator that determines whether a signal is high speed or not:
fknee=0.5/tr20/80
Short electrical length of a line, a short line is one where the signal will travel from transmitter to receiver before any reflections hit the receiver. Generaly a line shorter than this should be pretty immune from reflections etc, though this depends on the line having no serious irregularities along its length.
lelectrical=tr20/20/6
What is important with clocks is matching the load impedance (traces and input buffer) to the drive capability of the oscillator, this becomes more critical when a buffer is used as the output of buffers can vary greatly. A series terminating resistor is often used, but without a scope or better still simulation software, its value can only be guessed, or calculated approximately, I believe this document has a lots of the maths involved:
http://www.develeast.ch/downloads/TN001_RiseTimeAndKneeFrequency_v01.pdf
This is one reason why some clock mods get to me, people just add a super duper clock module (though a lot of these I think are **** and are just sold to unknowing audiophiles as a solution, cheap ebay units) with a long length of wire (or even a short length) with no concerns about the signal traveling from a to b. The better ones at least do mention this and have options for different resistors etc, but a lot seem to be add a TCXO and everything will hunky dory.
And this is just interesting:
http://www.ecestudents.ul.ie/Course...es/Microsoft PowerPoint - EE6471 Lecture5.pdf
Knee frequency, ie the harmonics content of the square wave you should care about, this is also the indicator that determines whether a signal is high speed or not:
fknee=0.5/tr20/80
Short electrical length of a line, a short line is one where the signal will travel from transmitter to receiver before any reflections hit the receiver. Generaly a line shorter than this should be pretty immune from reflections etc, though this depends on the line having no serious irregularities along its length.
lelectrical=tr20/20/6
What is important with clocks is matching the load impedance (traces and input buffer) to the drive capability of the oscillator, this becomes more critical when a buffer is used as the output of buffers can vary greatly. A series terminating resistor is often used, but without a scope or better still simulation software, its value can only be guessed, or calculated approximately, I believe this document has a lots of the maths involved:
http://www.develeast.ch/downloads/TN001_RiseTimeAndKneeFrequency_v01.pdf
This is one reason why some clock mods get to me, people just add a super duper clock module (though a lot of these I think are **** and are just sold to unknowing audiophiles as a solution, cheap ebay units) with a long length of wire (or even a short length) with no concerns about the signal traveling from a to b. The better ones at least do mention this and have options for different resistors etc, but a lot seem to be add a TCXO and everything will hunky dory.
And this is just interesting:
http://www.ecestudents.ul.ie/Course...es/Microsoft PowerPoint - EE6471 Lecture5.pdf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.