VCO: Negative VCO Linearity

2014/7/19

a good sweep signal source need a good VCO, good linearity. one years ago, i try some VCO and get some idea to make a linearity VCO. 

how to get a linearity VCO:

* use a series tank than a parallel tank,  and don't use clap because it's capacitor feedback parallel with the tank.

* well decoupled with the active device,  series with tank.

* use large C small inductor configuration

* select a strong feedback oscillator, deploy AGC for flat output

Franklin VCO Linearity

i try the Franklin oscillator, it's good, but still get worse linearity at high band, because active device still parallel with the tank.  EDN had release a article  Series LC tank VCO breaks tuning range records  which is match all of above criteria, then worth a try.

Result 

•  40Mhz to 127Mhz tune range, 3.175:1, wide band

•  from 0.7V to 16.5V whole tune range,  very close to 5Mhz/Volt, very linearity. (middle range might get 5.5Mhz/6Mhz per volte)

Varactor selection/problem and resolve

i try 2 different varactor, BB910 and BB152.   BB910 is very bad, from 10 Volt,  frequency/voltage reduce to few hundreds Khz/Volt.   can not achieve whole range linearity.  another problem is  the prototype is hard to start oscillation, but after oscillation begin, it's very stable and strong.

the improve schematic:

* Add Rs to reliable start oscillation

* 2.2uH RFC change to 1mH and parallel a 1k dumping resistor  , if no 1K resistor, frequency will dominated by 1mH RFC.

* for high frequency, R* need reduce to 5.6ohm

* small R* will influence the tank frequency, short R*  will raise  low end from 30Mhz to 40Mhz. 

* if deploy AGC like original design, at low frequency, R*  value is high,  get a lower low edge frequency, at high edge, AGC reduce R* to smaller one get high edge frequency, this will present a  more wide tune range.

R* = 5.6ohm  41Mhz -- 131Mhz 

R* = 33ohm,  30Mhz--- xxx  (lack of gain) 

*might need a capacitor bypass the R* for frequency >100Mhz

* Almost every AC couple will impact the tank frequency, AC couple the 1N1418 will raise frequency about 1Mhz.

the BB152,  tune voltage from 0.9V to 16V provide about 3.4:1 ( sqrt(62/5) )tune range.  my experiment oscillator given  3.2:1 tune range, almost reach the theoretical tune range.  the most impressive thing is , in whole range, slope of the Mhz/Volt is straight.   

AGC is not easy to deploy. original design might work, but that AGC transistor's Vce is too low. without AGC output level is significantly reduced when frequency raise.

Hartley  VCO

Actually, I suspect Hartley will have good linearity than clap or other capacitor coupled oscillator , and could get good frequency cover ratio. after a quick check,  it does so.

several notes:

* find a position to get the low band start oscillation

* extract the signal from very low end of the coil.

* use double or triple varactor to overcome the parasitic capacitance.

* short lead, or there might have spur ( frequency changed very sharply at one point, it's some parasitic oscillation)

* from 1Mhz to 100Mhz easy to get 2.8:1 tune ratio

* >100Mhz, hard to get so wide tune range, hard to find good feed back position

this VCO use Cascode configuration, is  a low frequency version oscillator which is one from VA3IUL present wind-band Oscillator.

linearity result:

actually, it's very bad. tune range is very narrow, because the negative component is serial with transistor's Cbe which is small and limit the turn ratio, if connect Cx, tune range will larger, but linearity almost useless.

Vackar VCO 

 result: 

44Mhz to 80Mhz

linearity:

bad

44M to 73M (1Volt to 12 volt),  2.6-2.9Mhz/volt

43Mhz to 80Mhz (>12  Volt),  reduce to 1 MHz/volt

let's check a Lambda negative resistance oscillator, the last one i'm interesting.

Lambda diode sch and construction ,  for high frequency the 2-33pF capacitor  make it easy to get oscillate.

Lambda oscillator is very very simple ,  connect a variable power supply to find a voltage it start oscillation. then this could be transform to a resistor at given supply voltage.

the linearity is amazing!  even through it's tune ratio only about 2.5:1. 

• • easy find a working point

  • from Khz to 150Mhz, Wide-band

  • good linearity

  •  tank energy is really high! this improve the noise performance.

  •  output is almost flat, whole band.

  • 100k 20k could replace with 6.8k/6.8k for high frequency, which need more gain and current.

  •  at VHF, like 100Mhz,  parallel a 100pF capacitor with 100k resistor. actually this capacitor is very important and most of web page don't mention this problem.

the negative resistance device is actually parallel with the tank at AC point of view,  and it's  impact the linearity much less.