PA: Exploring PA
@7/10 2012, updated @7/13
Several days ago, i roughly design and assemble a 27Mhz transmitter and receiver. to experiment the transmitter and simple PA design 27Mhz: Walkie Talkie Experiment. this is far from a good PA, provide me a chance to play around PA design.
the goal is to design a one LiPo supply walkie talkie given several hundreds milliwatt output power. But failed to get the desire power and transmission distance.
this piece of blog focus on the transmitter pre-amplifier and PA stage.
Fig 1 the original design
The obvious problem is:
1) T4 use 2sc8050, which Ft around 100 Mhz, given the AC current gain might around 100Mhz/27Mhz = 3.7, which is so poor. I don't have a medium RF power transistor for HF purpose, this is the most possibly usable one.
2) low Gain 8050 is definitely hard to driven and low power gain, and impact modulator stage even impact the oscillator stage.
3) lack of proper impedance match network. the 8050 data-sheet definitely won't give what i want.
4) output filter design is strange: The L3 resonant with C8, C5 at round 27Mhz make it work at UN-normal status. even-though it works, L3/L4/L5 is totally tie together, make it very hard to adjustment.
Adjust RFC
The first important thing is decouple the C5/L4/C6 from the class C transistor RFC, the L3. the method is change C8 to 6800pF [330pF is enough, i just pick one capacitor form junk box]
Test condition:
1. C8-> 6800pF
2. L3: 3uH tune to Max output
3. L5: tune to perfect sine wave
PreAmp Input T2 base
PA drive T4 base
PA out T4 collector
filter out C6 connect to a 50R resistor
ANT out disconnect Ant and connect a 50ohm resistor.
9V VCC, NO load (disconnect C8 from the T4 8050)
T4 base PAout
2.4Vpp +25Vpeak to -10Vpeak [10x probe AC couple mode]
4V VCC ( 4V VCC test had pitfall, does not tune to a best performance)
PreAmp Input PA drive: Parallel 50ohm PA out(50ohm) filter out(50R) ANT out(50R)
4Vpp 1.6Vpp: 900mVpp .6Vpp 500mVpp (suspect data) *400mVpp(suspect data)
9V VCC
PreAmp Input PA drive: Parallel 50ohm PA out(50ohm) filter out(50R) ANT out(50R)
5.2Vpp 6Vpp : distortion :2.2Vpp 8Vpp 9Vpp 8.4Vpp
result:
8050, class C :
1) current gain, 3-4 [Ltspice simulation result, close to 100Mhz/27Mhz ]
2) voltage gain: x3.5, 20*log(3.5)=10dB @9v, reduce to 4~5dB@4Vcc
3) low supply voltage severely damage the performance
4) around 160mW output@9V 6mW@4V supply
Improve: change R7 to 3.3R
R7 introduce a degradation for the gain, and can be reduced for better performance. based on Ltspice simulation, change to 3.3R is better for SS8050, more low value can't continuously improve the gain.
test condition:
1. C8-> 6800pF , R7->3.3R
2. L3: 3uH tune to max output
3. L5: tune to perfect sine wave
4. Test point:
CXO: T2 base
Modulator out, T4 base:
PA out: T4 collector, via 0.1uF to 50ohm load
Filter out: C6, L,4,L5 junction
Antenna out: the antenna.
5. 10X probe, system is running, for each test point, re do 2,3 tune to the best performance
6. All test with the transmitter working! input audio 200mV
4V VCC
PreAmp Input PA drive(T4 base):50R PA out:PA out 50R filter out: filter 50R ANT out:Ant out:50R
3.8Vpp 3.5Vpp : 1Vpp 8Vpp: 2Vpp 20Vpp:1.4Vpp 4Vpp: 3Vpp
9V VCC
PreAmp Input PA drive(T4 base):50R PA out:PA out 50R filter out: filter 50R ANT out:Ant out:50R
8Vpp 4vpp : 2.6vpp 20Vpp: 10Vpp 40Vpp : 4.4Vpp 10Vpp:10Vpp
Test method limitation
* the transmitter is in working all the time, receiver is 3 meters away monitoring signal. each time change test point, L3,L5 is re tuned to better performance. cause the 10X probe had several pF capacitor need to be tuned out.
* the result is not very accurate, especially without 50R load.
* When put 50R load at test point, cause the transmitter is in work, antenna (might) present a low impedance, the read out value should less than it actually be.
data analysis
-------------------
for SS8050, class C amplifier
1) Lt-spice simulation current gain, 3-4
2) voltage, power gain: x3.8 , around 10dB [ this not power gain just voltage gain]
3) low supply voltage severely damage the performance
4) around 250mW~300mW output @9V 10~20mW@4V VCC [driven power is too small for class C]
(9V VCC strong interference make my mouse stop work(left key stop response)!!!)
The Conclusion part
A typical class A PA, bias the base around 1/5 to 1/4 Vcc, the max swing from near VCC(should be 2*Vcc) to around 1/5Vcc, chose correct the BJT's collector impedance to given desired power output, efficiency < 45%. Class C is popular in CW or QRP rig.
What kind of carrier can use Class C PA
Class C is full of distortion, for a small signal it is not even amplified at all , so do not use it for AM/SSB application like this project.
Class C: minimal drive signal level
obviously Class C PA can not amplify the signal which voltage peak less than 0.7V, the transistor is not leave the cut-off region. by simulation use SS8050 or 2N2222 with 3.3R emitter degeneration resistor, the drive signal at least need peak voltage reach to about 1.4V peak, that is 13 dbm, 20mW deliver to a 50R system. if remove the degrade resistor, the drive voltage can reduce to 0.9Vpeak for gain about 10dB, that is 9dbm, 8mW.
but, a input impedance match network for class C is common, to provide a low impedance driven source. bifilar transformer given a 4:1 ratio impedance transformation, but also half the driven voltage.
update @ 2014/6/27
as a thumb of rule, signal present at BJT base (after impedance transformer), should get at least around 0.9Vpeak. [.@2014/7/10
jump to: PA: driven chain smoke Design check more experiment...]
What limit the Gain of Class C PA
------------------------------------------------
VCC voltage, the max swing is 2*VCC if no output match circuit this limit the max Pout. low supply might degrade the transistor's gain. The transistor current gain is another factor. for this simple PA, the SS8050 current gain lower to 4x at peak current.
For next step experimental
-------------------------------------
input and output impedance measurement is require for better design. the class B drive stage should be add before final class C with a impedance match transformer or match network provide higher current driven. the Class C is not suitable for AM, next time try the class AB linear PA.
Reference 1: use Ltspice explore the PA tune
A) trying replace strange output filter with a 50ohm - 50ohm low pass filter.
Green trace: input voltage
RED trace: output voltage
I(R1): the transistor current
I(R5): the driven current
use 2N2222, we get 18Vpp output to 50ohm [2N2222 might not capable use like this, just for simulation research ], driven current peak is 50mA.
use SS8050, we get 10Vpp output to 50 ohm [SS8050 might not capable use like this, just for simulation research ], driven current peak is 150mA!!!.