Homebrew RF Siganl Generator Colpitts Fet Oscillator
Homebrew RF Siganl Generator Colpitts Fet Oscillator
Homebrew RF Siganl Generator Colpitts Fet Oscillator
V1 is 9.0 Volts dc
1
Resonate Frequency F 0 3
2 LC
o L = 1 uH
1
o Fmax =
2 1.0uH *16.5pF
o Fmax = 39.2 Mhz
1
o Fmin =
2 1.0uH * 34pF
o Fmin = 27.3 Mhz
Bias
The FET forms a classic Source follower
Because there is little or no current into the Gate
o There is no voltage drop across R1
o Using Kirkoffs voltage law we get Vgs = - Vs = -Id Rs
(Rs = R2)Vgs
Vgs 4
Id Idss (1
Vp
1.5
Now we can compute Id 11ma (1
4.0
o Id = 8.69 ma
Rs = Vs/Id = 1.5V/7.78 = 192 ohms
Vg = 0 by design
Vd = 9V power supply
NOTE: The R required to achieve Vgs = -1.5V in simulation was 3.3K
ohms
o So these calculations are off by an order of magnitude
o The model FET is not an exact match
o Looking at the Data sheet 5 we see a large variance for Vgs and Idss. This
and Reference 6 show that it is very difficult to calculate the biases for
and FET with any degree of certainty
Amplitude
I was not able to calculate the amplitude of the output.
Harmonic Noise
I was not able to calculate the Harmonic noise of the output.
Simulation
Simulations were over optimistic about magnitude and the ability to oscillate than
a real circuit until the C1-C3 were made non- ideal by adding 10ohm series
resistance, however this is then to pessimistic!
Frequency
C1, C2 = 33pf , L1 = 1uh => 36.8 Mhz
550.000m
Series1
Series2
500.000m
450.000m
400.000m
350.000m
300.000m
250.000m
200.000m
150.000m
100.000m
50.000m
0.0
-50.000m
0.0 10.000M 20.000M 30.000M 40.000M 50.000M 60.000M 70.000M 80.000M 90.000M 100.000M
Bias
Vd = 9.0
Vg = 0.0
Vs = 1.5
Amplitude
C1, C2, C1 = 33pf , L1 = 1uh => 2.0 Vp-p
C1, C2, C1 = 110pf , L1 = 1uh => 0.4 Vp-p
C1, C2, C1 = 33pf , L1 = 10uh => 11.0 Vp-p (Clipping)
C1, C2, C1 = 110pf , L1 = 10uh => 5.6 Vp-p
C1, C2, C1 = 33pf , L1 = 100uh => 11.0 Vp-p (Clipping)
Harmonic Noise
The values below are Vdb to the nearest peak in the frequency domain
(harmonic distortion)
C1, C2, C1 = 33pf , L1 = 1uh => -24.5 db
C1, C2, C1 = 110pf , L1 = 1uh => -29.5 db
C1, C2, C1 = 33pf , L1 = 10uh => -25.8 db
C1, C2, C1 = 110pf , L1 = 10uh => -22.6 db
C1, C2, C1 = 33pf , L1 = 100uh => -28.0 db
C1, C2, C1 = 110pf , L1 = 100uh => -28.0 db
Bias
I failed to measure these.
Amplitude
C1, C2, C1 = 33pf , L1 = 1uh => 3.0 Vp-p
C1, C2, C1 = 110pf , L1 = 1uh => 4.0 Vp-p
C1, C2, C1 = 33pf , L1 = 10uh => 8.0 Vp-p (Clipping)
C1, C2, C1 = 110pf , L1 = 10uh => 9.0 Vp-p
C1, C2, C1 = 33pf , L1 = 100uh => 9.0 Vp-p (Clipping)
C1, C2, C1 = 110pf , L1 = 100uh => 10.0 Vp-p (Clipping)
Harmonic Noise
The values below are Vdb to the nearest peak in the frequency domain
(harmonic distortion)
C1, C2, C1 = 33pf , L1 = 1uh => -14 db
C1, C2, C1 = 110pf , L1 = 1uh => -14 db
C1, C2, C1 = 33pf , L1 = 10uh => -15 db
C1, C2, C1 = 110pf , L1 = 10uh => -20 db
C1, C2, C1 = 33pf , L1 = 100uh => -12 db
C1, C2, C1 = 110pf , L1 = 100uh => -30 db
Amplitude Vp-p
C1, C2 = 33pf , L1 = 1uh 3.0 2.0 N/A
C1, C2 = 86pf , L1 = 1uh 4.0 0.4 N/A
C1, C2 = 33pf , L1 = 10uh 8.0 11.0 clipping N/A
C1, C2 = 86pf , L1 = 10uh 9.0 8.5 N/A
C1, C2 = 33pf , L1 = 100uh 9.0 11.0 clipping N/A
C1, C2 = 86pf , L1 = 100uh 10.0 11.0 clipping N/A
Harmonic Noise db
C1, C2 = 33pf , L1 = 1uh -14 -24.8 N/A
C1, C2 = 86pf , L1 = 1uh -14 -29.5 N/A
C1, C2 = 33pf , L1 = 10uh -15 -25.8 N/A
C1, C2 = 86pf , L1 = 10uh -20 -22.6 N/A
C1, C2 = 33pf , L1 = 100uh -12 -28 N/A
C1, C2 = 86pf , L1 = 100uh -30 -28 N/A
When I discovered that the frame of the gang capacitor was had to be
isolated from ground ans that it suffered frequency changes even when
touching an insulated knob I dropped the Colpitts in favor of the Hartley
Oscillator
References
1. UNKNOWN, The ARRL Handbook For Radio Communications, (ARRL 2007)
p10.13, (Figure 10.12 A).
2. Harris, Frank, Crystal Sets to Sideband, (2006 Rev10),
http://www.qsl.net/k3pd/chap10.pdf p3 online, accessed 2008.
3. Horowitz, Paul and Hill, Winfield, The Art Of Electronics Second Edition,
(Cambridge University Press 1989) Section 1.22 Resonate Circuits and Active
Filters p41.
4. Experimental Methods in RF Design Wes Hayward et al First Edition p 2.6 Fig
2.19
5. UNKNOWN, MPF102 N-Channel RF Amplifier, (FAIRCHILD 2004),
http://www.fairchildsemi.com/ds/MP%2FMPF102.pdf, online, accessed 2008.
6. Horowitz, Paul and Hill, Winfield, The Art Of Electronics Second Edition,
(Cambridge University Press 1989) Section 3.05 Manufacturing Spread of FET
Characteristics, p122-127.
THE BUFFER
SIMULATION
Coupling
o DC
Vout = .200mVp-p and Vosc has also dropped to just above this
range
o Coupling capacitor
150 pF (50 ohms at 20 Mhz) 350mVp-p
Imax = vb-.7/re
o Assume 1ma and vcc/2 => re = 4.5K
Assume gain = 3 => Rc = 13.5K
The calculate Rbias to get vcc + .7
However max Ic = vcc/(re + re)
Gain ~ rc/re
Icmax = vcc/rc+re
SIMULATION
Start with previous design
Add 50 ohm load dies so add emitter follower, can we loose fet drain follower
o FET
Coupling Capacitors also affect the upper bandwidth of the previous stage. I
usually model my coupling capacitors as the input to a stage only. As expected
the higher value of the coupling capacitor the lower the lower cutoff frequency is.
However as demonstrated below the coupling capacitor into the next stage also
affects the upper cutoff frequency.
o Unloaded amplifier and response
It appears that the most important transistor spice parameter for high
frequency is cjc. This is the BC depleation capacitance discussed above! And
found by replacing each spice parameter in a model one by one and noting the
difference in frequency response.
Gang capacitors use the frame as a common node. Therefore this common
node would have to be the node connected to R1 above. This means that the
frame would have to be insulated from ground, but even with an insulated knob
touching the knob would allow the capacitance of my body to change the output
frequency. Therefore I dropped the Colpitts design in favor of a Hartley
design.