Use 1sv149 or bb112 in linear vco oscillator?

Hello I design a very simple spectrum analyzer and in the VCO section it is important that the frequency change linearly with variadions in voltage.
My sawtooth ramp generator has good linearity, so the question is which vricap to prefer for better voltage to capacitance linearity, the 1sv149 or the bb112?

The varicap is connected in series with a 100nF capacitor.

Varicap diodes don't provide a linear tuning range due to their natural Cap.vs.Vtune charateristics.Therrefore don't expect a linearity from a VCO builded with varicaps.If this VCO ( or any ) works at low frequencies ( few MHz ) you don't have to use varicap equipped oscillator, there are some very nice OpAmp based sinusoidal oscillators with very smooth linearity.

BigBoss said:

Varicap diodes don't provide a linear tuning range due to their natural Cap.vs.Vtune charateristics.Therrefore don't expect a linearity from a VCO builded with varicaps.If this VCO ( or any ) works at low frequencies ( few MHz ) you don't have to use varicap equipped oscillator, there are some very nice OpAmp based sinusoidal oscillators with very smooth linearity.

Click to expand...

Thank you, this info was useful.
However just for comparison, which of the two varicaps is more linear?

However just for comparison, which of the two varicaps is more linear?

Click to expand...

I think the characteristics are similar. What do you see in the datasheets?

However, how about a nice design exercise? Make an analog linearization circuit that generates the diode voltage for a linear input voltage to oscillator frequency characteristic.

The bb112 has sensitivity C1:C8 of 20:1 typ. but only good log-linearity of 5:1 for C2:C6 ( meaning flat from 2V to 6V )
The 1sv149 has a sensitivity C1:C8 of 15:1 min but good log-linearity of 8:1 from C1:C6

I hope you realize VCO phase noise must be minimized for a SA, and more Vc sensitivity often translates to more phase noise.

Also the VCO must operate at a much higher frequency to get good linearity in a small % range of fC
So if you are sweeping 0~100MHz you would operate more than 200~300MHz etc...for image rejection and smaller tuning range.

You can use two varactors in the pi filter using differential Vc to double the range.

FvM said:

I think the characteristics are similar. What do you see in the datasheets?

However, how about a nice design exercise? Make an analog linearization circuit that generates the diode voltage for a linear input voltage to oscillator frequency characteristic.

Click to expand...

You basically mean altering the inpit voltage in such a way that when fed to the vco, it makes it linear?
Intersting!

- - - Updated - - -

SunnySkyguy said:

The bb112 has sensitivity C1:C8 of 20:1 typ. but only good log-linearity of 5:1 for C2:C6 ( meaning flat from 2V to 6V )
The 1sv149 has a sensitivity C1:C8 of 15:1 min but good log-linearity of 8:1 from C1:C6

I hope you realize VCO phase noise must be minimized for a SA, and more Vc sensitivity often translates to more phase noise.

Also the VCO must operate at a much higher frequency to get good linearity in a small % range of fC
So if you are sweeping 0~100MHz you would operate more than 200~300MHz etc...for image rejection and smaller tuning range.

You can use two varactors in the pi filter using differential Vc to double the range.

Click to expand...

I have to say that I am not so familiar with these Cx ranges, what do they mean.
By looking at the graphs I see that the 1sv149 is more linear, i.e. when the ramp input volrage changes, the capacitance of the diode changes more linearly. But I am not sure about it.
However the linearization of a vco does not only depend I think on the linearization of the varicap. I believe amplitude stability throughout the range is more important, so ensure the mixer will be fed with a constant LO amplitude.
I know about the phase noise of the LO but I design a really bare bones SA here, using mainly discrete components. It has to be cheap and mainly for LF to 30MHz, the band of interest.

The resolution of an SA is mainly defined by the bandwidth of the filters. A single crystal has the lowest bandwidth, but fundamental mode crystals are produced to about 17MHz or so.
Having these in mind, I am thinking of a bad but practical scheme for the SA.
The single conversion SA will have an IF of 15MHz with a single 15MHz crystal providing high resolution (at low sweep rates). A 20KHz-15MHz vco will mix to receive frequencies below 15MHz (with a switched 15MHZ LPF) and above 15MHz (with a switched 15MHz HPF).

What do you think of this simple setup?

I know, it is not a good practice to have the IF inside the band of interest but although a LO/HI band switching is required (to switch in/out relevand input filters) it would be of quire high resolution and ultra cheap to implement.

By looking at the graphs I see that the 1sv149 is more linear

Click to expand...

Did you notice that the datasheets have different linear versus logarithmic axes? You would need to copy the data of both diodes to your own table/diagram for a visual comparison.

A 20KHz-15MHz vco

Click to expand...

Typo?

neazoi said:

What do you think of this simple setup?

Click to expand...

N.G.

If is possible to have an FM lab generator span a few decades, but this is not a trivial design and only works for relaxation oscillators with current control i.e. up to maybe 30MHz rather than resonant control i.e. varicap

DDS chips sweep up to 12MHz from AD.

I suggest your concept will hit a brick wall when you try and then fall apart with the IF in the RF range due to intermodulation lack of image rejection. Put the IF and LO well above the RF range, then have a 2nd and 3rd stage mixer for best performance.

FvM said:

Did you notice that the datasheets have different linear versus logarithmic axes? You would need to copy the data of both diodes to your own table/diagram for a visual comparison.

Typo?

Click to expand...

No not a typo. It can be done with excellent amplitude stability **broken link removed** however not in switching ranges, not at once.

- - - Updated - - -

SunnySkyguy said:

N.G.

If is possible to have an FM lab generator span a few decades, but this is not a trivial design and only works for relaxation oscillators with current control i.e. up to maybe 30MHz rather than resonant control i.e. varicap

DDS chips sweep up to 12MHz from AD.

I suggest your concept will hit a brick wall when you try and then fall apart with the IF in the RF range due to intermodulation lack of image rejection. Put the IF and LO well above the RF range, then have a 2nd and 3rd stage mixer for best performance.

Click to expand...

I do not expect performance near 15MHz, in fact there will be a dead zone there, but firtunatelly this is quite far from the 20m amateur band. Having a 2nd and 3rd IF adds complexity, I would like to build something really cheap, for hobby use. That is why I am thinking the single crystal filter at 15MHz, as it can lead to good resolution.
However the HPF/LPF 15MHz RF filters must be of high order and still some opposite mixer products will pass through.

you can search for a fractional N synth, PLL and VCO in 1 chip and sweep with digital tuner, according to your skills.

All are low cost. Then look at receiver chip with RSSI for Log to linear voltage response.