Terminator3
Advanced Member level 3
Update: if i disconnect varactor from tuning voltage and measure current between varactor and ground, it is around 5mA. If dielectric resonator is removed, current is 0mA. Maybe there is too strong coupling between varactor half-wavelength line and DR? My oscillator consumes around 30mA. So 5mA coming from passive varactor+line near DR seems to be pretty huge value!
The problem:
I observe DC voltage around 1v (must be 0v) on tuning voltage line when dielectric resonator is put in place. This voltage pulls control voltage. For example, 200mv modulation meander becomes only 100mv in amplitude if DR is in place.
If I remove 1k resistor between DAC and tuning voltage input, then this voltage able to pull DAC output voltage to some percentage.
I am surely measure DC voltage, because there all proper decoupling is in place (thin lines, radial stubs, capacitors to ground). It is not like I am measuring RF with a voltmeter.
I think that I can pull this voltage to 0v if i put unity gain opAmp betwen DAC and varactor. But it seems to be totally wrong, because many VCOs use resistors of tenth KOhms in series with varactors without problem, because there is no current consumption.
At the same time all is working, VCO tuning range is around 10MHz. I think because of that unwanted voltage tuning range is reduced, and it must be around 20MHz if varactor is biased through entire voltage range 0..-2v. I am using FET as varactor, with variable capacitance Cgs.
My setup:
Half-wavelength microstrip line is coupled to dielectric resonator.
Microstrip line is connected to FET's gate, gate is biased at 0V
Tuning voltage is applied on FET's source 0...+2v, so Vgs varies 0...-2v, and negative voltage generator schematic is avoided.
Tuning voltage is applied from DAC output to FET's drain/source pin through 1k/680pF low-pass filter + thin line with radial stub.
Source and drain are connected together to prevent any unwanted Vds.
So it looks like traditional open-ended line with varactor, but FET is soldered on the place of a varactor, FET's drain and source are shorted, and biased with DAC. FET's gate is connected to half-wavelength line.
What could it be? Is it FET problem or the same thing happens when using traditional varactors?
The problem:
I observe DC voltage around 1v (must be 0v) on tuning voltage line when dielectric resonator is put in place. This voltage pulls control voltage. For example, 200mv modulation meander becomes only 100mv in amplitude if DR is in place.
If I remove 1k resistor between DAC and tuning voltage input, then this voltage able to pull DAC output voltage to some percentage.
I am surely measure DC voltage, because there all proper decoupling is in place (thin lines, radial stubs, capacitors to ground). It is not like I am measuring RF with a voltmeter.
I think that I can pull this voltage to 0v if i put unity gain opAmp betwen DAC and varactor. But it seems to be totally wrong, because many VCOs use resistors of tenth KOhms in series with varactors without problem, because there is no current consumption.
At the same time all is working, VCO tuning range is around 10MHz. I think because of that unwanted voltage tuning range is reduced, and it must be around 20MHz if varactor is biased through entire voltage range 0..-2v. I am using FET as varactor, with variable capacitance Cgs.
My setup:
Half-wavelength microstrip line is coupled to dielectric resonator.
Microstrip line is connected to FET's gate, gate is biased at 0V
Tuning voltage is applied on FET's source 0...+2v, so Vgs varies 0...-2v, and negative voltage generator schematic is avoided.
Tuning voltage is applied from DAC output to FET's drain/source pin through 1k/680pF low-pass filter + thin line with radial stub.
Source and drain are connected together to prevent any unwanted Vds.
So it looks like traditional open-ended line with varactor, but FET is soldered on the place of a varactor, FET's drain and source are shorted, and biased with DAC. FET's gate is connected to half-wavelength line.
What could it be? Is it FET problem or the same thing happens when using traditional varactors?
Last edited: