@bucks bunny Here is a circuit idea for you using the Analog Devices SSM2212 Dual-Matched Low Noise NPN Transistor. I have configured it using Common Base and with dual +- 4V supplies (assuming an OpAmp will be there later). The Common Base boosts the Q so it is quite resonant. The Collector current is about 1mA, the value in the datasheet, hfe for this transistor is 600.
You don't like Common Base configuration lol.
But the SSM2212 is very low noise for a high quality audio preamp.
Ferrite Rods are used in Industry for high frequency welding type of applications.
This circuit seems to work best when the Collector resistor is a little bit higher than the Emitter resistor, It's voltage gain and Q increases as the Collector resistor increases until it reaches a "tipping point" where the Gain drops off and the Q drops quickly. I have seen old designs where they have applied AGC here I think to tune up to just below this tipping point? It would be interesting to make the Collector resistor 3K3 in series with a 1K pot to adjust Gain. Equations for Common Base are well known but I didn't work out why this "tipping point" happens. So the Common Base has low input impedance, high output impedance, near unity current gain, high voltage gain. They also say there is good isolation between input and output and I have noticed this.
By the way all the examples I found work with single supply, my circuit above is the only time I saw Common Base with dual supply - it simplifies the bias, less resistors simpler circuit.
Last edited:
I did compare common base to the non-inverting circuitry with LTSpice and well - you are right, I do not prefer this.
I have searched the internet and come to the conclusion that Common Base is universally badly explained. Nowhere have I seen split positive and negative supplies used and nowhere have I seen the "tipping point" explained, maybe you used different configuration/component values for LTSpice?
I would use Common Base if I only had one good transistor to work with.
I tried it with 2N2222 but the SSM2212 which is intended as a differential pair works better than the 2N2222.
It does make no real difference whether you apply single or split supply to your common base circuit.
Being a non inverting configuration, there is a possibility for pos feedback from output to input which could explain
your rising Q with more output resistance up to self oscillation.
If you are looking for a high Q pre-selection this may be fine for you.
At the time I pursue a broadband approach - where the common base circuit does not offer any advantage.
Btw it would help you a good deal understanding the circuits if you would simulate them.
Being a non inverting configuration, there is a possibility for pos feedback from output to input which could explain
your rising Q with more output resistance up to self oscillation.
If you are looking for a high Q pre-selection this may be fine for you.
At the time I pursue a broadband approach - where the common base circuit does not offer any advantage.
Btw it would help you a good deal understanding the circuits if you would simulate them.
Not exactly positive feedback.
I have redrawn the circuit to make it clearer and explain as follows. Consider the voltage ve, if a small current flows through C1 so as to make ve more positive then that will decrease vbe incrementally (because Base is tied to 0V) and the transistor will turn less on, so ic and ie currents will tend to drop together (ic approx = to ie) so that the voltage drop across R2 will tend to drop so that the transistor Q1A will tend to act to reduce the small increase in ve. So that the transistor will act so as to oppose a change of voltage ve. This creates a virtual low resistance at point ve (a bit like a virtual ground idea) and this makes the Q of the resonant circuit L1 C1 high.
Starting from the beginning again, if a small current flows through C1 so as to make ve more positive then that will decrease vbe incrementally and the transistor will turn less on, so ic and ie currents will tend to drop, if ic drops the voltage across R1 drops and voltage at point vc will increase, point vc goes up.
So the voltage gain from ve to vc is non-inverting (the gain is positive) but the feedback to point ve is negative.
So positive voltage gain but negative feedback built into this circuit. This is such a simple circuit but it has a high voltage gain. This circuit could be useful for low impedance moving coil transducers.
Appreciate your broadband approach, its just that this configuration interests me and I'm trying to leave notes for other contributors who may have an interest or wish to learn. Also it is clear that you have a good understanding of circuitry such as this.
Agree that a single supply version is likely equivalent but it is more complicated to understand (and to calculate bias).
Here is the Common Base Dual Supply circuit that I am referring to:-
Attachments
Last edited:
I also realise now what causes the "tipping point" I referred to before, its become obvious.
The current ie mostly depends on R2 and the vbe forward voltage drop so that ie = (vR2 -vbe)/R2 about 1mA. As long as vc is always higher than 0-vbe+vcesat that's 4+0.65-0.25 about 4.4 volts, then all is ok. If R1 is too large the voltage drop across R1 exceeds 4.4 volts, Q1A saturates and the current ic cannot reach the 1mA of ie. so the equality ic approx. equal to ie breaks ie stays at about 1mA but ic is below 1mA. The circuit stops working properly.
To allow a larger value for R1, the positive supply V+4 has to be increased so that ie approx. equal to ic can remain true.
The current ie mostly depends on R2 and the vbe forward voltage drop so that ie = (vR2 -vbe)/R2 about 1mA. As long as vc is always higher than 0-vbe+vcesat that's 4+0.65-0.25 about 4.4 volts, then all is ok. If R1 is too large the voltage drop across R1 exceeds 4.4 volts, Q1A saturates and the current ic cannot reach the 1mA of ie. so the equality ic approx. equal to ie breaks ie stays at about 1mA but ic is below 1mA. The circuit stops working properly.
To allow a larger value for R1, the positive supply V+4 has to be increased so that ie approx. equal to ic can remain true.
Last edited:
That should read "As long as vc is always higher than 0-vbe+vcesat that's 0-.65+0.25 about -0.4 volts, then all is ok. If R1 is too large the voltage drop across R1 tries to exceed 4.4 volts (-4-0.4) but it can't because Q1A has saturated"
Just one more observation, if R1 is increased such that ic < ie then ib has to supply the extra current. Therefore a good way to check operation is to measure the Base current ib.
Interestingly R1 at 3K9 is only good for small signals, for larger signals R1 needs to be lower (and voltage gain is then less) to avoid Q1A saturation. So again "It would be interesting to make the Collector resistor R1 3K3 in series with a 1K pot to adjust Gain" A better setup would be making R1 2K4 and the pot 2K.
There is no phase difference between input and output, an useful property.
Interestingly R1 at 3K9 is only good for small signals, for larger signals R1 needs to be lower (and voltage gain is then less) to avoid Q1A saturation. So again "It would be interesting to make the Collector resistor R1 3K3 in series with a 1K pot to adjust Gain" A better setup would be making R1 2K4 and the pot 2K.
There is no phase difference between input and output, an useful property.
Last edited:
Some more spice investigations.
This is a front end using op amps.
The ferrite coil feeds a hi-z input in the classical manner we found in old radios.
The only thing special is the second op-amp segment forming a varicap
providing a wide variable resonant frequency of the tank.
To be honest this is by far not my idea but a concept
I discovered during my school days in the sixties
when at first I tried to demystify the Vox Wah-Wah Pedal:
A potentiometer varies resonant peak of a pot core inductor (ca 1H).
At the end it boils down to a Miller capacitor in a stage with variable inverting gain.
A really clever design.
And it is just the same trick here.
This is a front end using op amps.
The ferrite coil feeds a hi-z input in the classical manner we found in old radios.
The only thing special is the second op-amp segment forming a varicap
providing a wide variable resonant frequency of the tank.
To be honest this is by far not my idea but a concept
I discovered during my school days in the sixties
when at first I tried to demystify the Vox Wah-Wah Pedal:
A potentiometer varies resonant peak of a pot core inductor (ca 1H).
At the end it boils down to a Miller capacitor in a stage with variable inverting gain.
A really clever design.
And it is just the same trick here.
Attachments
Last edited:
Below is my suggestion for implementing Q multiplication and a brief description. Please excuse my crude drafting.
I assume an antenna connected to a parallel L-C resonator and a low-noise opamp with voltage gain of +10. As a guess, I’m assuming the resonant tank circuit has unloaded Q of about 30 and have modeled the losses as a parallel resistor (Rp = 33.9k) in shunt with the inductor. Resistor Rf applies positive feedback to the the resonator to raise the Q. Threshold of oscillation is reached when loop gain is 1.0. That occurs when Rf = 9 * Rp --- that is, when the attenuation of the Rf and Rp voltage division is 1/10, and the forward gain of the opamp restores the gain to unity.
I’ve arranged Rf as a combination of a fixed R and a pot so that the multiplier Q can be driven from low Q into full oscillation, but of course these are designer options. I’m remembering a Heatkit Q multiplier from my teenage years.
I assume an antenna connected to a parallel L-C resonator and a low-noise opamp with voltage gain of +10. As a guess, I’m assuming the resonant tank circuit has unloaded Q of about 30 and have modeled the losses as a parallel resistor (Rp = 33.9k) in shunt with the inductor. Resistor Rf applies positive feedback to the the resonator to raise the Q. Threshold of oscillation is reached when loop gain is 1.0. That occurs when Rf = 9 * Rp --- that is, when the attenuation of the Rf and Rp voltage division is 1/10, and the forward gain of the opamp restores the gain to unity.
I’ve arranged Rf as a combination of a fixed R and a pot so that the multiplier Q can be driven from low Q into full oscillation, but of course these are designer options. I’m remembering a Heatkit Q multiplier from my teenage years.
Decoding of CW recordings
I was inspired to construct a decoder for SAQ and other CW signals. My design is a hybrid of analog and digital components on two PCBs. The analog has an RCA input jack for standard audio line. It has a 4562 with a gain of 15. It drives a IN914 and LM393 which outputs 3vdc asserted-low pulses which pull down the digital input. The singleton analog is powered by 15vdc.
The digital PCB has a NODEMCU 8266. I turned off the default WiFi so interrupt response would be reliable. There is an interrupt for the pulse from the analog PCB and other self-interrupt which triggers every 100 microseconds. ISR2 manages real-time state changes so that the timing of audio on-time and off-time can be monitored. I tested a file from ARRL. These files are not CW. They have sine waves which turn off completely from a high volume so I get reliable state changes. Timings of the dah-dash-dot changes vary but within a range so no problem. I attach a snippet which varifies that it is working.
Unfortunately, I tried CW processed files from here and weaksignals.com. I scope good pulses but timings are erratic. I believe the problem is that CW stations can not turn off completely and my shaping circuit does not cope with it so far. Perhaps I will think of something later.
I have a module for detecting WWVB in Colorado. That station drops power 17db to represent power off starting a time second. The receiver module has AGC to make a clean switch.
I was inspired to construct a decoder for SAQ and other CW signals. My design is a hybrid of analog and digital components on two PCBs. The analog has an RCA input jack for standard audio line. It has a 4562 with a gain of 15. It drives a IN914 and LM393 which outputs 3vdc asserted-low pulses which pull down the digital input. The singleton analog is powered by 15vdc.
The digital PCB has a NODEMCU 8266. I turned off the default WiFi so interrupt response would be reliable. There is an interrupt for the pulse from the analog PCB and other self-interrupt which triggers every 100 microseconds. ISR2 manages real-time state changes so that the timing of audio on-time and off-time can be monitored. I tested a file from ARRL. These files are not CW. They have sine waves which turn off completely from a high volume so I get reliable state changes. Timings of the dah-dash-dot changes vary but within a range so no problem. I attach a snippet which varifies that it is working.
Unfortunately, I tried CW processed files from here and weaksignals.com. I scope good pulses but timings are erratic. I believe the problem is that CW stations can not turn off completely and my shaping circuit does not cope with it so far. Perhaps I will think of something later.
I have a module for detecting WWVB in Colorado. That station drops power 17db to represent power off starting a time second. The receiver module has AGC to make a clean switch.
Attachments
Here's some technical and historical details of the transmitter, many not previously covered in this thread:
https://mastodon.social/@tubetime/110970146022678448
https://mastodon.social/@tubetime/110970146022678448
Hey @w5jag, thanks for the tip. Those ARRL practice files include extension characters but I am content to decode just the alphanumeric... for now. They also include some broken english and change weekly so they cannot be memorized.
"? 10 WPM ? TEXT IS FROM MAY 2022 QST PAGE 70 ? THE AUXILIARY COMMUNICATOR AUXC IS A RELATIVELY NEW POSITION .....(and so on)
lines processed = 30143"
So I can run a ARRL file through my contraption and, with a small C program, output text. However, no luck with these SAQ recordings. I call them "mixtapes" because they contain only tones. The transmitter signal is lost. SAQ being mechanical, it is probably not frequency stable as a modern unit. Then the mixing oscillator may have its own wiggles. And the telegrapher introduces some normal variance. All these factors combine with noise to muddy the timings.
Actually it is only necessary to capture the dit-dahs. I timestamp them with the ESP8266, print them over USB virtual serial to the PC where a little C program extracts the text. Finally, I believe the timing file can be used to gate an audio oscillator 750hz or whatever. So mixing is not needed.
I am looking for a not-mixed VLF CW transmission recording for more testing. The carrier freq can be upto 30khz. Nothing so far.
"? 10 WPM ? TEXT IS FROM MAY 2022 QST PAGE 70 ? THE AUXILIARY COMMUNICATOR AUXC IS A RELATIVELY NEW POSITION .....(and so on)
lines processed = 30143"
So I can run a ARRL file through my contraption and, with a small C program, output text. However, no luck with these SAQ recordings. I call them "mixtapes" because they contain only tones. The transmitter signal is lost. SAQ being mechanical, it is probably not frequency stable as a modern unit. Then the mixing oscillator may have its own wiggles. And the telegrapher introduces some normal variance. All these factors combine with noise to muddy the timings.
Actually it is only necessary to capture the dit-dahs. I timestamp them with the ESP8266, print them over USB virtual serial to the PC where a little C program extracts the text. Finally, I believe the timing file can be used to gate an audio oscillator 750hz or whatever. So mixing is not needed.
I am looking for a not-mixed VLF CW transmission recording for more testing. The carrier freq can be upto 30khz. Nothing so far.