Also added a .01 uF cap from the + supply line to ensure adequate RF bypass. Currently the XTAL provides this function but I decided this is better.
Also changed the PI output network to 390pf-1uH-560 uF This added another 15 mW or so to my 100 mW. The transistor change was afterward.
Instead of directly grounding the crystal, you can ground it THROUGH a .001 capacitor. .001 uF is a low impedance at the RF frequency, and you still get a single freq., no-Rit pixie2.
RIT: Permanently ground the crystal through a smaller capacitor, say 50 to 60 pF. This raises the xtal frequency about 600 hertz. So a 7040 xtal == 7040.6 KHz. Connect a SPST switch such that allows you to parallel a .001 uF cap across the 60 pF, thereby totally providing an excellent ground to the xtal. The crystal now operates at it's proper (lowest) possible frequency.
A smaller, say 30 pF cap will provide 1Khz offset or so. Just remember, the well grounded crystal operates on the lowest frequency it was designed for.
AM BC BAND interference: Get Ferrite Chokes. They are the size of 1/4 watt Resistors. Do NOT use wirewound molded inductors, they are miniature antennas! The ferrite chokes I use are 5 millihenrys and have a green paint drop supposedly revealing their value in millihenrys. Green=5.
Looking at the schematic you see the KEYING LINE and Ground; SPEAKER LINE and Ground, POWER Positive Line and GROUND, a 1 Kohm resistor feeding reduced current/voltage to the LM386. Also, the Detected Audio from Q2 line goes to the audio input pin 2 or 3 of LM386.
OPlace a RF CHOKE: Audio Output; Key Jack; Detected audio to LM386 Pin 2; Power Supply Positive line. Next; add small Ferrite bead to Positive Voltage line going to each Transistor.
This next part is Tricky! I wanted a stable 5.62 volts to the LM386 and decouple it from the power supply (Car Battery). On the Pin 6 side of the 1 Kohm resistor (R5), put a 5.62 Volt Zener Diode. Change R5 to 500 ohms, or parallel another 1 Kohm as I did. Put in 250 uf to 470 uF electrolytic capacitor from pin 6 to ground of the LM386. YOU WILL GET IMMEDIATE IMPROVEMENT IN AUDIO AND BC INTERFERENCE REDUCTION!!!
Add 100 ohm resistor from D1 switching diode Cathode to Keying line. This suppresses the now SURGING current from 470 uF powersupply capacitor. Without the 100 ohm resistor, the keying device has to pass too much current to be reliable. My keying relay contacts stuck sometimes... Also, some current flow is possible thru D1 and the 10KOhm emitter resistor and the Q2 amp gets stuck in Self-Transmit mode!
Mine works FB and althoug I still get a small amount of BC interference from the AM station about 2 miles away, I find it disappears when I touch the enclosure. So it is just stray capacitance and when I put the lid on my PC BOard constructed/soldered enclosure, I expect all BC qrm to be gone.
My Pixie is a 40/80 dual band switchable and I beefed up the 2 transisters to a 2222 and a 2219 can type for the amp. I get about 1 watt out on a 12vdc battery. I'm very happy with the output and I do make some contacts but copy is very rough. I've checked into the Klites net on Sunday eves but that's a tough one and usually rely on my Icom for that.
I tried the 2219 and it is fine. Currently using a 2n5109. But the 2n2222A works well too. But I put in a bigger socket that fits the 2n2219 and 2n5109 so I'll keep those.
For rit I prefer the manual switch! I have the xtal permanently grounded with the 60 pF cap. Then switch in a .001 uF in parallel which puts the rig exactly at the xtal freq.
I don't like the automatic xmt offset of w1fb for fixed freq. rigs. Great if you have vfo's, but for a xtal rig, you'll get TWO sending freqs if you switch manually for RIT.
I used pin 3 for audio input to the LM386. It seemed more stable than pin2. I orginally used pin2. If you touch pin 2 (or 3) you provide HUGE BC signal and it gets detected.
If you Touch the PA transistor CASE, Same affect and it feeds directly into the LM386.
If you Touch you straight key hot-lead -- SAME Affect. If you use a speaker near another rig powered by AC line -- HUM and BC.
I used these broadband ferrites having 5 separate holes drilled in the body. These are commonly available RF chokes. Look like 3/8 inch section of a pencil. It comes pre-wound with solid wire strung in/out of the holes and finishes as an axially mountable part. I have lots of these. But I figure any RF choke in the millihenry range will work, but have more ohmic loss.
Even better and cheaper ferrites are probably available cheaply. I also used ferrite beads where the power is applied to the oscillator, and also where the bypass capacitor is attached to the collector load inductor of the power amp.
I found that the LM386 preferes NOT to run at 10 Volts. I put in a 5.62 Volt on the Pin 6 of LM386 just after the 1 Kohm resistor comes from the main battery supply. No more problem there. Now if I load it with a speaker, the LM386 doesn't squeal.
Also, I added a 100 ohm resistor 1/16 watt thing, to the pin 1/8 feedback loop of the 386 to tame the gain a bit.
NOTE: BC station is about 4 miles away, and very easy to detect. You can turn off the pixie2, touch the LM386 at audio input and hear on your headphones!
In order of importance.
I built a HI-Pass, 3 MHz filter the second day of the Pixie/49r contest and improvement was good. But have since put that circuit on the shelf. The ferrites have CURED the problem!
Well, I prefer manual rit. So get a switch and permanently ground the xtal thru a 60 pF cap. Switch in the .001 cap for RIT. The beauty of manual rit is you can use either switch position to transmit and receive at the other spot! Basically extends you frequency coverage. It is nice to note the xtal is on its base frequency when grounded thru the .001 cap, for FREQ spotting purposes.
I added 5 milliHenry Ferrite chokes to any jacks (Except the antenna :) Do NOT use regular wirewound inductors, they pick up the Broadcast band QRM by induction.
ADD a 250-470 uF bypass electrolytic for pin 6 of the LM386. Otherwise you'll get audio problems. OH, make sure the LM386 is getting sufficient current, a current starved LM386 will otherwise be a great detector for your local AM Radio station -- :)
I later decided to run at 12.7 or so volts from a car battery and the LM386 had trouble. I've sinced changed R5 to 500 ohms, put in a 5.62 Volt zener at pin 6 of the LM386. MUCH BETTER!
ONE PROBLEM with HAVING a 470 uF cap on pin 6 -- The audio muting diode passes the current thru your keying circuit to ground. So I added a 100 ohm resistor in series with the muting diode and that keeps my keying relay from arcing. I'm planning to drive a sidetone with the voltage developed across the 100 ohm resistor during key down.
Use a .1 uF capacitor at LM386 pin 6 too. I put a 100 ohm resistor in series with the 10 uF cap between pis 1/8 of the LM386 to tame any overdrive problems too.
I removed the 33 kOhm bleeder bias resistor in the power amp. Ferrite beads and .1 bypass caps in the supply lines to both transistors.
Remember: The current starved LM386 causes many BC interference problems. Be sure to put that 250 uF or so electrolytic at PIN 6. And if this doesnt help your audio, lower R5 a bit.
The stock pixie2 uses about 20 to 25 milliamps and is capable of running off of a 9 Volt transistor battery. I'm not sure, but another problem develops with the 741 in this circuit due to the additional current demands. Try the thing on a pair of 9 Volt batteries in parallel to check that.
Explanation of what I did and why I changed the 1 Kohm. This is a bit complicated, but when I hooked the stock pixie to the 9 volt battery I noticed a few things.
1) The LM386 Motorboated as the battery voltage got to about 7 volts. The spec sheet on the LM386-N1 recommends it run between 5 and 12 volts.
1A) 7 volts should have been fine, I deduced the LM386 was being starved for current.
2) The LM386 is a good detector for AM BC band. I decided to further decouple the LM386 from the power supply by putting a 5.62 volt zener to ground on the LM386 side of the 1 Kohm.
2A) I lowered the 1 KOhm to 500 ohms by parallelling another 1K. This allowed me to stabalize the LM386, decouple the Battery and Leads from RF, and supply more current as needed.
3) The LM386 had problem with strong signals and I observed problems with audio on CW peaks.
3A) I added a 470 uF electrolytic to pin 6 of the LM386 to serve as a current reserve for those peaks. This also improved audio response SIGNIFICANTLY, and eliminated more BC interference.
4) The mods ran too much current through the diode on key down. Also the idle current through the 10 Kohm resistor - Diode - to power supply was too much and this turned on the power transistor as if you were keying. (BAD PROBLEM!)
4A) I put a 100 Ohm resistor on the Kathode (Key) side of the diode. This reduced the current through the key contacts during key down, also kept the output transistor from turning on.
There you go Friend. I hope this helps. It was a lot of work, but very worth it! I have a very quiet pixie2 and works FB. The audio is strong enough to drive a speaker and the key clicks are virtually non-existant. The 470 uF bypass cap on the LM386 was totally amazing. Really helped the circuit in three ways: Audio quality, LM386 stability, Reduced BC interference to to RF detection of the current starved LM386.