VCO, PLL and Search Oscillator.

[Station X]

I wonder if someone could explain to me what the function of a search oscillator is in a PLL/VCO circuit?

A bit of background, I'm trying to sort a problem with a UHF VCO designed to operate over the range 433 MHz to 434 MHz. It's controlled by a voltage derived from the usual phase comparator driving a varicap diode. That much I understand. It's all very familiar to me.

However the varicap diode also has applied to it a 30 Hz sawtooth waveform of sufficient amplitude to sweep the VCO frequency over a 5MHz range. As far as I can see this means that the phase comparator has to "fight" this voltage in order to maintain a reasonably constant voltage at the varicap when the loop is locked.

I have observed the control voltage on a 'scope, not an easy thing to do because of the low frequency involved. If I unlock the loop, by varying a fixed capacitor in parallel with the varicap, the sawtooth voltage becomes visible. It disappears when the loop is brought back into lock and the waveform reverts to a horizontal line. That part of the circuit appears to be behaving correctly.

So going back to my original question, What is the purpose of the search oscillator?

Thanks.

[Biggles]

Maybe some requirement to modulate the transmitter with a sub audible tone such as CTCSS? I have seen a few circuits where they directly inject the low frequency tone into the synth loop rather than later on in the audio circuits. Just a guess. Apart from that I have no other suggestions why the designer would do this.

[Radio Wrangler]

Early phase-locked loops used phase detectors which literally just did what they said on the can and nothing extra.

A simple phase detector is a close relative of a diode ring mixer or of an EXOR gate, and an EXOR gate is a close relative of what's known as a Gilbert cell mixer.

It the VCO is off-phase a little, it produces a DC voltage at its output which steers the VCO to reduce its phase error.

If the VCO is off frequency a little, that DC voltage will change as the instantaneous phase error changes. The phase detector output will be a sinusoid or triangular waveform at the same frequency as the difference in the two inputs to the phase detector.

If the VCO is substantially off frequency, the output of the phase detector is at a higher frequency and will be severely attenuated by the loop filter. The PLL can make no attempt to acquire lock.

Designers had to handle this. One method was to use DACs to pretune the VCO so it ought to be never further away than the loop bandwidth from where it should be, but this takes expensive precision. Another fix was to have a slow oscillator that would sweep the VCO across its full range if it was not in lock and eventually it would pass through the intended frequency slowly enough to acquire lock.

When this is done, either the designer relies on the output from a phase detector in lock being strong enough to fight the search oscillator and win. Alternatively the search oscillator is designed to stop when a locked condition is sensed. It was a common use for unijunction transistors operating as a relaxation oscillator. A locked loop would hold the tuning voltage away from the oscillator's switching points and the search function wouldn't cause sidebands.

Today, this is rarely used. Most often we use phase-frequency detectors which are made from a few flip flops and gates. If out of lock, these indicate whether the VCO is too high or too low, and give an output voltage end stopped in the right direction to steer the VCO towards lock. Phase/freq detectors allow the use of integrator type loop filters which have numerous advantages, but they can give a bit more phase noise than the best sampling/mixer type plain phase detectors. I've designed low phase noise loops that acquired with a phase/frequency detector and then switched onto a sampler for low noise.

David

[G6Tanuki]

It's basically a way of forcing the VCO to sweep the full extent of its range when out-of-lock; the idea being that the 'normal' narrow-and-accurate lock-loop will then grab the oscillator as it passes and thereafter hold it closely in-lock.

I've usually come across these when the frequency-range to be covered is wide and the needed lock-time is short; you have two competing requirements - the need to gain lock quickly (so rapid wide-sweeping range) and the need to thereafter stay closely/low-jitterly/low-noisedly in-lock (narrow range). You can't use one loop characteristic to achieve both.

[Martin G7MRV]

Some form of coarse tuning, a similar system is used in the Clansman PRC349, where a tap is taken off the divider chain to create a very slowly varying voltage onto the varicaps, such that the phase detectors output doesnt need to vary widely.

A much more complex version was used to lock the IF conversion in the Pye PF8, which although crystal controlled used a dual loop scheme to maintain frequency accuracy

[Station X]

Thanks for the explanations. It's now perfectly clear to me what the function of the search oscillator is. The rig is about 40 years old so that explains why a search oscillator was used. It's just a couple of NOR gates wired as an astable with the output integrated to convert the square wave to an approximate sawtooth.

Biggles. Thanks for your reply, but it's nothing to do with CTCSS.