reverse power flow.

[gar]

100420-0639 EST

dbuckley:

Several comparators for zero crossing detection are: Analog Devices AD8561, AD8598, AD790, and AD8611. You many need slight filtering on the output.

NE555s or LM555 can be used for phase shift adjustment.

For SPDT switches Analog Devices ADG419 and Maxim and Vishay DG419. These have 30 t0 40 ohms on resistance and need to be followed by an appropriate high input impedance amplifier.

Some useful simple amplifiers are AMP-01 (50 k input), AMP03 (25 k input) for input to the switch.

For an output amplifier consider an ordinary OP7 using the positive high impedance input or an instrumentation amplifier such as AD620, AD625, and AMP02.

On the switch connect S1 input to common or ground, and S2 to your signal to be gated. D is the output to a high impedance amplifier.

These components allow a system with a response from DC to the upper limit of the amplifiers.

Although I created the tube type gate and I have reports showing the waveforms I do not remember the details of the gate and quickly I can not find any details of the circuit.

.

 

[chris kennedy]

Test, can't open page 3.

Got it, delete if necessary.

 

[dbuckley]

Yes, thanks, I can see where you're coming from, but its not quite my problem.

I generate a 50KHz-ish FSK squarewave using a PIC, so I can control that it always generates exact full cycles of each frequency, I use an RC network to make it into a sine wave, and a cap and a potential divider to make it symetrical about a midpoint, and this is injected onto an audio line. And it nearly works, except for an intitial bend in centre voltage, so I have a signal that audibly clicks (though quite quietly, but still enough to be annoying) , and I haven't figured out how to stop it.

I did consider flicking the PIC pin at a couple of MHz as an idle signal, which would mean only a frequency transition not a voltage transition, but I have several transmitters, and if they're all on they'll beat, and thight might be really ugly...

Any suggestions gratefully received.

 

[gar]

100420-2021 EST

dbuckley:

My initial suggestion is this:

Use an D to A converter as the signal generator. In the computer store the sine function in a table. It is sufficient to just store one quadrant. Maybe 20 samples per cycle might be enough. You would use a low pass or band pass filter to filter the stairstep waveform and smooth it out.

You can generate X full sine wave cycles starting and ending on a zero crossing. Then immediately switch to a different frequency starting at a zero crossing. You can also do phase flipping.

Usually FSK does not use two different frequencies very far apart so the filter won't have much effect on the desired waveform.

At 50 kHz how do you hear the signal? Is this a home or business project? Is the purpose a secret or not?

Would the signal be any good if you had two sine wave oscillators of the desired frequencies, my gating means to generate one frequency burst for X cycles, and a second gating means for Y cycles of the the second frequency, and the possibility of a gap up to one full cycle between bursts?

.

.

 

[gar]

100420-2044 EST

dbuckley:

In your circuit the time constant of the input low pass filter is about 3 microseconds, 3 1k resistors in parallel as a source resistance to an 0.01 mfd capacitor.

What is the purpose of the DC bias on C4? I assume the item identified as C2 is a 1 k resistor.

What does the waveform across C4 look like?

.

 

[gar]

100420-2247 EST

dbuckley:

Thinking more about your square wave drive it is not obvious why a multistage RC low pass filter should not be sufficient to filter your variable frequency square wave. If full cycles are used for each frequency there should be no discontinuities.

.