EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Anson on July 16, 2013, 03:48:40 am
-
I am at the beginning stages of creating a circuit for a 10MHz OCXO I have. Thanks alank2! Its a fairly standard 5 pin OCXO a NDK 9100 series. It needs 12V in which is easy enough I have a 15v 1A Wall wort and a L7812cv for that. Was thinking of a small filter cap on the output maybe a 470µf electrolytic. It has a pin that I think is a vref for external adjustments alank2 I think just grounded it and adjusted the internal pot when he calibrated it against his rubidium. I was thinking of putting a pot on there for future adjustments. This is all fairly simple i guess but I just wanted to run it by you guys and see if there is room for improvement. The adjustment pot I was thinking of using a 10k.
Next is the hard part for me I wold like to set this up to be able to connect it to two counters so I need two outputs. Also the output of the oscillator is only 1.43V I would like to have 5v out. So I think I need to run the output into an amplifier but I am unsure which type or chip to use. I know many opamps need a split rail or dual tracking supply and I want to avoid that. I also want to avoid any frequency variation. Any suggestions?
-
Was thinking of a small filter cap on the output maybe a 470µf electrolytic.
You should probably AC-couple the output via a capacitor, but 470uF is a few orders of magnitude bigger than you want here. Use a 10-100nF ceramic.
It has a pin that I think is a vref for external adjustments alank2 I think just grounded it and adjusted the internal pot when he calibrated it against his rubidium.
I think you're normally supposed to float the pin to let the oscillator reach its natural frequency, if you don't want to adjust it vs. an external reference, in which case you need to apply an adjust voltage, typically 0 <= Vc <= Vref.
I was thinking of putting a pot on there for future adjustments. This is all fairly simple i guess but I just wanted to run it by you guys and see if there is room for improvement. The adjustment pot I was thinking of using a 10k.
That's how these pins are intended to be used. You place a pot across Vref and Gnd, with the wiper on Vc. Use a multi-turn trim pot, the value is not critical. I've heard tell that some of these have noisy Vref outputs and their performance can be improved by using an external reference, but that's something you should investigate with your OCXO.
Next is the hard part for me I wold like to set this up to be able to connect it to two counters so I need two outputs. Also the output of the oscillator is only 1.43V I would like to have 5v out. So I think I need to run the output into an amplifier but I am unsure which type or chip to use. I know many opamps need a split rail or dual tracking supply and I want to avoid that. I also want to avoid any frequency variation. Any suggestions?
If you're only interested in comparing frequency, especially if you are not using 50R counters, just hook them in parallel with a BNC T or something, it won't be all that critical, and any timer should work fine from the typically ~0.5-1Vrms these put out. If you want to square it up there are many ways, the easiest and cheapest is probably a 74HC04 with feedback, but precision timing types will poo-poo this. A fast comparator is another fairly simple option.
-
Thanks for the help. Leaving out an amplifier simplifies things. Should I put in any resistors between the output and the bnc? I wouldn't think it would do anything but I'm not an expert. Not yet anyway ;)
-
I'd use a lumped element wilkinson splitter to provide the two outputs, that way noise from one output will be much reduced on the other output.
http://www.microwaves101.com/encyclopedia/wilkinson_lumped.cfm (http://www.microwaves101.com/encyclopedia/wilkinson_lumped.cfm)
I'd also put an amplifier on the output, probably a discrete BFS17 or similar transistor design.
-
One thing to check, is whether the 7812 can actually regulate the voltage from the 'wall wart'. Many wall warts have a very rough and unstable output, and are only barely filtered at all in many cases. The result is that their output voltage may be quite noisy, with lots of mains ripple imposed on their nominal output voltage. So the dips may go a fair bit below the 15V specified.
On the other hand, then the minimum voltage difference a 7812 need between input and output when regulating, the dropout voltage, is frequently only specified as a 'typical' value of 2 V, not a guaranteed maximum. So I would check whether you actually get dips in the output voltage, particularly during start-up as the OCXO draws maximum current. Variable mains voltage may also cause periodic difficulties.
If you have problems, then either you need a better wall wart, or maybe using a better regulator with a guaranteed low dropout voltage can help. The LT1086 series comes to mind here, just as an example of a better mousetrap.
Your 1.43V output voltage is probably with no load attached to the oscillator I guess? RF signals are usually confined to systems, where all building blocks, like amplifiers and oscillators, have a constant input and output impedance. Most systems are 50 ohm, though 75 ohm is also fairly common.
Your OXCO is probably intended to be used in a 50 ohm system, so the output voltage should drop to half if a 50 ohm resistor is connected across the output to ground. Cutting this signal in twain and amplifying the results is not terribly hard, especially as you are only interested in one frequency, 10 MHz. Doing this on a single 12V supply rail is not a problem, as you can AC couple the signal path. You can build the required circuitry from standard SMD resistors and caps, plus common RF gain building blocks called MMICs (monolithic microwave ICs). This whole thing is called a distribution amplifier, and they can be bought ready for use.
The one catch is that a signal amplitude of 5Vpeak in a 50 ohm system isn't that easy to generate on a single 12V rail, unless you build a custom narrow band amplifier with tuned LC circuits. 5Vpeak into a 50 ohm load is 250 mW, so a class A amplifier would as a first estimate have to burn 1W to generate this level of power. A rough rule of thumb is that a class A amplifier can provide about 25% of its DC input power as useful signal output.
I can post more details later if you wish, if other people haven't filled in the gaps in the meantime. :D