You guys had a lot of great ideas to my original post, but I realized I probably didn't give enough information concerning the application. In a block diagram, the phase shifter has 2 inputs, one is for the 1MHz sine wave and the other is for a control voltage, probably a ramp generated by a 16 bit DAC. The output is the phase shifted sine wave. The ramp causes the sine wave to be continuously shifted up to +- 90 degrees. There will also be a feedback circuit which will stop the ramp at the appropriate voltage level. The sine is being generated by a very low noise ocxo, -110Dbc @ 1 hz from center, and the phase shifter needs to not mess this up. Thanks in advance for any ideas you may have.
So you could use - as Phil Allison said - a voltage-controlled oscillator t o produce an equally low noise 1MHz sine wave, frequency locked to the firs t, but phase shifted.
Voltage-controlled crystal oscillators do exist - they use a varactor to pu ll the crystal frequency. The varactor can't pull it very far, so you'd nee d to start off with two well-matched 1MHz crystals.
Some digital frequency synthesis chips can produce a reasonably clean pair of sine waves with any phase relationship you want to program in.
The noise on the synthesised output will be low level high order harmonics. If your master oscillator is clean and stable, the synthesised waveforms w on't have any more noise 1kHz away from the centre frequency than the divid ed down master oscillator frequency.
Analog devices have lots of application notes for their - not all that chea p - DDS chips.
formatting link
QodBjwATg
You'd have to digitise your phase-shifting voltage to get a digital number to control your phase shift, but that's no big deal.
I always thought serrodyne frequency translation took worked with microwave frequencies. I would think a frequency of only 1 Mhz, and phase translation measured in single Hz is a no-go for this method. If I'm wrong, I'd appreciate a link pointing me to relevant literature. Thanks. G-
Pretty accurate, unfortunately. It all comes down to working within fractions of a hertz at exactly 1 megacycle.The basic idea comes down to being able to make the output waveform match the frequency, waveform, and amplitude of an external non-connected target signal.
Given the latest revelation here, I don't think you need a phase shifter at all because this won't do what you want it to do. How will an ocxo ever match some random frequency? If there is no frequency lock, you will be phase shifting for ever, but you will run out of range eventually. Either that or I don't understand the problem. [Wouldn't be the first time.]
What is the frequency of this ramp? You could easily end up with an output waveform comprised of the sum and difference of the 1MHz sine wave and the even harmonics of the ramp waveform.
Phase shifted in reference to what? If you simply phase shift something XX number of msec, you end up with a delay circuit. If you phase shift it with a periodic waveform, you have a phase modulator. Which do you want?
Continuously shifted? That's a phase modulator as found in the common FM broadcast and land mobile radio transmitters. A simple reactance modulator won't do +/- 90 degrees, but it can be done with a PLL modulator or with a series of variable reactances (i.e. varactors). If you have direct access to the digital 16 bit lines driving your ramp generator, you could build a very linear switched capacitor delay circuit.
What do you mean by "appropriate"? Are you stopping the ramp by amplitude, timing, frequency, or phase? How do you determine when to start again? Why do you need this complication?
But the phase shifter will mess this up. When you phase or frequency modulate, or mix a signal that is low noise, you're going to get sidebands. Mother nature and Bessel Functions will make it happen and there's nothing you can do to stop it. Your 1MHz carrier signal will remain low noise, but over a bandwidth wider than a few Hz (depending on the spectra of the modulation), you're going to see noise and sidebands. The only way you can retain the low noise sidebands is by adding (combining) the signals, since they remain unchanged, but which doesn't do what you want.
Add a drop of wine to a glass of sewage and it remains sewage. Add a drop of sewage to a glass of wine, and you still get sewage.
can
I hereby declare 0.1Hz to be a suitable fraction of a Hz. That's 1 part in 10 million or about 2^24. You would need 24 bits of resolution in order to do that and you only have 16 bits on the modulation, which sets the bottom limit on phase accuracy. Try again please.
Well, the frequency is going to be 1Mhz, which isn't going to change. The waveform is going to be a sine wave, which will only change if you modulate it with your ramp. The amplitude can be set to whatever you want, or simply track the input from your stabilized 1MHz oscillator. In short, nothing changes between input an output except the addition of sidebands from the modulation and some even harmonic distortion due to modulation with a ramp. So, please answer the question. How much amplitude variation can you tolerate?
Mehinks your description is a muddle. I suggest you start over, and just specify what you have going in and what you expect coming out. Then we can figure out a black box that will do what you want.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
The problem is 1MHz is nominal. This requires a PLL using a vcxo. That design is everywhere. Colorburst lock for example. Actually locking to the colorburst is a harder problem.
Even very precise telco schemes use justification bits to account for frequency difference.
Yep. You have to eliminate the sidebands from the color burst frequency before you can lock effectively.
Which brings me back to my question... a phase shift in reference to what? If it's in reference to a fixed point on the 1MHz input signal, it's a delay circuit. If the phase shift ramp waveform is periodic, then it's a phase modulator.
Judging by the muddle, methinks that solutions are premature. There are many different interpretations of the original problem and little in the way of a consensus or a clarification on what he is trying to accomplish. For example, an "external non-connected target signal" means what? Wireless connected? Telepathy? Two tin cans and a string?
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
** A troll called "George" posted here in September last year asking about
*un-tuned* ferrite loop antennas for the AM band for a special app he refused to describe.
He was very concerned about the "phase response" of the antenna.
You guys had a lot of great ideas to my original post, but I realized I probably didn't give enough information concerning the application. In a block diagram, the phase shifter has 2 inputs, one is for the 1MHz sine wave and the other is for a control voltage, probably a ramp generated by a 16 bit DAC. The output is the phase shifted sine wave. The ramp causes the sine wave to be continuously shifted up to +- 90 degrees. There will also be a feedback circuit which will stop the ramp at the appropriate voltage level. The sine is being generated by a very low noise ocxo, -110Dbc @ 1 hz from center, and the phase shifter needs to not mess this up. Thanks in advance for any ideas you may have.
Again, I apologize for not describing the problem well enough. Even so, there are literally a ton of great ideas for me to look into, and I thank you all. Let me provide an overview of the problem and then maybe things'll be a bit clearer ( as mud, right...). This is an experiment in radio reception. The transmitter is an extremely low powered, unmodulated sine wave. Since the transmission frequency is known exactly, the receiver is built with fixed circuitry to specifically receive only that signal. Since there is no modulation (yet), the output of the mixer stage in the receiver is going to be the product of the phase difference between the received carrier and the local oscillator. I am looking for a way to manipulate the phase of the local oscillator to match the phase of the received carrier. When this happens, the phase output will be 0. I need a circuit to monitor the phase and change the L.O. such that it tracks the input signal and keep the output at 0. The lower noise, the better. Right now I am leaning towards a VCO scheme, but wanted to see if anyone has any better ideas. Once again, thanks.
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.