Modulate difference freq.

Best to do this as a simulation on a PC unless you use DDS that is.

2 very stable and maybe MILL spec VCO's that get biased from the same source using all the same component values that are closely selected for matches.. You could at least then discover the linearity of the components you're using ..something on the same lines as CMR (common mode rejection) I guess.

--
"I\'m never wrong, once i thought i was, but was mistaken"
Real Programmers Do things like this.
http://webpages.charter.net/jamie_5

"Harry Lang"

** A linear frequency modulator - silly.

....... Phil

Do you have a control input to the 3KHz signals? Are you using VCXOs or DDSes to make the signals?

If you have no control over the signal source, you can use phase modulation to do the frequency modulation. Phase and frequency modulation differ by one integration. If you are working over a modest frequency range, this may be the way to go.

How accurately do you need to keep the 100Hz difference?

Do you have to start with the 3KHz and 3.1KHz or can you change the circuit however you wish so as to make the desireed output?

Suppose there were a modulation ratio that could do that. Then, (3100 * Am * c) - (3000 * Am) = 100 where Am is instantaneous modulation amplitude, and c is some constant. Then (3100 * c) - (3000) = 100 / Am and 3100 *c = 100/Am + 3000 Suppose Am = 1; then c = 1 Suppose Am = 2; then c= 0.983...

You would either have to supply a non-linear modulation function to one of the modulators, or use something other than simple FM on the two signals. Maybe a PLL on one of the signals.

-- John

An FM modulated signal offset a constant frequency is still an FM modulated signal. Imagine one signal is the FM-modulated 3kHz. Mix that signal with a constant 6.1kHz and take the difference; be sure the mixer is very well balanced with respect to the input frequencies. Now you have an FM-modulated 3.1kHz signal. Mix the

3.1kHz and 3.0kHz signals and take the difference; it will always be 0.1kHz.

Right, that's something other than simple FM applied to two starting signals; and it sounds like a good way to do it.

-- John

Well, actually, that was a brain fart on my part! That SHOULD have been a circuit to simply add 100Hz to one, not to generate one that goes up while the other goes down. Sigh. Replace the 6.1kHz with

0.1kHz, but then be careful that you filter out the difference frequency, and do the filtering in a way that preserves the phase relationships: in other words, make sure that you delay both the 3.0kHz and 3.1kHz paths by the same amount, including the filtering. It would likely be easier to generate, say, 100kHz FM and mix that down to 3.0kHz and 3.1kHz, passing each through a low-order filter that has essentially the same delay for both of the outputs.

The problem if the delay is unequal: imagine a modulation frequency of, say, 200Hz, and a delay in the two paths that differs by 1/400 of a second. The the modulations will be out of phase: one will go up while the other goes down in frequency, and the difference is modulated. That's extreme, but to have the difference exactly constant, you need to have equal delays.

Cheers, Tom

That's interesting. It seems to be a pretty high precision requirement. Would it be possible to detect the 100 Hz difference, and trim the delay accordingly?

-- John

Well, yes, I suppose so, though _why_ anyone would actually want to do something like this I don't know!

Actually, things aren't all that bad if you start with a high frequency and mix it down with two mixers fed by local oscillators

100Hz apart to 3.0kHz and 3.1kHz. If you start with a high enough frequency, say several hundred kHz, a simple filter will kill the local oscillator and FM source feedthrough and the sum frequencies, and it can have high enough cutoff frequency that the delay difference between two copies of the filter can be well under a microsecond with no trimming of 5% tolerance parts (for the one I just simulated, anyway). But as you say, it's possible to get things like this to auto-calibrate if it's important.

Cheers, Tom