bandwidth

[mstechca]

I was again tinkering with my superregen (a.k.a. Audioguru's "TOY" :lol: )
and I notice that nigel's information about the capacitor connected between the NPN's base and ground seems accurate.

I read that selectivity is supposed to suffer from incorrect <quench frequency>:<incoming frequency> ratio.

selectivity and bandwidth do relate to each other in opposite ways. wouldn't you agree?

I was wondering, could I calculate bandwidth based on the incoming frequency and the quench frequency?

if so, is there a simple equation for it?

Maybe this is the reason why I can pick up a TV signal when I adjust the quench oscillator capacitor to one value, and I can't pick up the same signal when I adjust the capacitor to another value.

 

[audioguru]

I was again tinkering with my superregen (a.k.a. Audioguru's "TOY" :lol: )

I was wondering, could I calculate bandwidth based on the incoming frequency and the quench frequency?

if so, is there a simple equation for it?

Maybe this is the reason why I can pick up a TV signal when I adjust the quench oscillator capacitor to one value, and I can't pick up the same signal when I adjust the capacitor to another value.

Your Super-regen has been changed so much that it might not have a quench oscillator anymore. Maybe changing a capacitor stops it from working.
Your super-regen doesn't have an FM detector to detect the FM sound of a TV station so bandwidth doesn't matter when it "slope-detects" the signal by tuning the receiver's peak to one side of the transmission.

 

[mstechca]

Your Super-regen has been changed so much that it might not have a quench oscillator anymore.

that is impossible :lol:
I don't know of any superregen without a quench oscillator.

so bandwidth doesn't matter when it "slope-detects" the signal by tuning the receiver's peak to one side of the transmission.

Why not?
A TV station requires about 6Mhz of bandwidth right?

My detector is very similar to the following:
**broken link removed**

my version is shown in the attachment.

I still think the capacitor on the far right and the pull-up resistor make up the quenching frequency.

 

[mstechca]

Since I got errors, I checked back on the thread and saw my post twice. so I changed the second post to this message.

 

[audioguru]

I don't know of any superregen without a quench oscillator.

Your circuit doesn't have a quench oscillator, instead it is a regen receiver without a manual sensitivity control. The Techlib article that it comes from calls it a regen-receiver, not a super-regen. A super-regen uses its quench oscillator to make its sensitivity adjustment automatic.

A TV station requires about 6Mhz of bandwidth right?

The FM sound of a TV station has a bandwidth of only 75kHz at max modulation in mono. The AM modulated picture has a bandwidth of nearly 6MHz.

I still think the capacitor on the far right and the pull-up resistor make up the quenching frequency.

No they don't, your circuit is just an RF oscillator with a fixed bias control.

 

[Roff]

I ran a sim on Mstechca's circuit, and it does squegg at about 35kHz.

 

[audioguru]

I ran a sim on Mstechca's circuit, and it does squegg at about 35kHz.

I have seen many super-regen circuits and they all have a resistor and capacitor from the emitter to ground to form a relaxation "squegging" oscillator. MStechca's circuit doesn't have any parts to make a relaxation oscillator.

 

[mstechca]

This is so interesting. They say that a superregen is better than a regen.
It must be the filters that help me.

I ran a sim

Ron, what software are you using to simulate my circuit?
I need a simulator, or at least an equation that can indicate the reason why you measured 35khz.

From what I know, the inductor and capacitor cannot produce 35Khz. They produce 200+ Mhz.

I tried the bias resistor and the capacitor between base and ground, and that gave me 3Khz according to 1/(120K * 3nF)

and why is a relaxation oscillator necessary?
If later, I wanted high bandwidth, doesn't the oscillator need to be high too?

I am using an inductor instead of a resistor, because I want more gain.

If that 35Khz signal isn't coming from a quench oscillator, then what is it? :lol:

 

[audioguru]

why is a relaxation oscillator necessary?

Look in Google for Super-regen and you will find articles about how they work.
Since the transistor has positive feedback by the capacitor between the collector and emitter, it builds-up a signal at the tuned circuit's frequency and begins to oscillate. The oscillation is peak rectified at the emitter and the rectified voltage built up on the emitter resistor and capacitor to ground causes the transistor to be cutoff, then it stops oscillating and the charge on the emitter capacitor drains and it begins amplifying again. The transistor is the most sensitive to outside signals when it is barely oscillating.

If later, I wanted high bandwidth, doesn't the oscillator need to be high too?

Yes, to avoid aliassing beat frequencies when the desired signal is more than half the squegging oscillator's frequency. You will have problems with the 23kHz to 53kHz FM stereo subcarrier and the FM SCA "storecasting" signals if the squegging oscillator's frequency is less than about 140kHz. Data transmission will also be upset if the squegging oscillator's frequency is too low.

 

[mstechca]

I still can't figure out how Ron got 35Khz for a quench.

Since the transistor has positive feedback by the capacitor between the collector and emitter

I already knew that part.

so if you are saying the resistor and capacitor between emitter and ground make a relaxation oscillator, my version would then qualify for an LC oscillator.

So I can just get rid of the grounded capacitor on the far right, and it would make no difference?, and the the components between emitter and ground make the quench frequency? (if Ron's sim is screwed)

and one more thing.. If I converted the collector inductor to a tuned circuit, what should I make the feedback capacitor value?, should I aim for high reactance (Xc) or low reactance?

 

[Roff]

Mstechca wrote:

Ron, what software are you using to simulate my circuit?

At home, I use LTSpice/SwitcherCADIII.
Mstechca, I don't have a clue about how to analyze or predict the self-quench frequency of this superregen receiver. That's why I sim'ed it. The quench (squegg) frequency is a complex situation. As the transistor starts to oscillate (RF), it draws more current. This brings down the voltage on the collector. The base current, being derived from the collector voltage,starts to decrease, but there is a lag due to the RC time constant of the base bias network. eventually, the base current is too low to sustain oscillation, so the collector voltage rises rapidly. However, there is a lag in the increase in base current due to the base bias network time constant, ... I think you get the idea.
Someone is going to say, "yeah, but one time constant in a feedback loop does not have enough phase shift to cause a single transistor to oscillate". I'm not sure how to answer that. The Q of the tank means that oscillations don't start instantaneously. The build-up of collector current is slow, due to the Q, so this is another delay.
Here is **broken link removed** I found.

Oh - I modified the schematic to something more conventional, and changed the oscillation frequency, which came out as 105.001MHz. The quench frequency is now around 150kHz.
Your original circuit oscillated at about 300MHz, and did self-quench at about 35kHz, as I said.
I used a 2N918, which was the closest model to 2N3563 that I cold find.

Below are some waveforms from the sim, along with the schematic. The lowpass filter is a 5 pole Butterworth. I set the corner frequency to 20kHz. The 1 ohm impedance was just for convenience. If you were going to do this in hardware, an active filter would be a better choice.

I'll have to add the other waveforms in the next post.

 

[Roff]

Here are the other waveforms.

 

[mstechca]

Aren't there two components somewhere in your circuit that make up 150Khz for the quench?

If I know the components for sure, then maybe I can just switch the values instantly, and everything will function better.

 

[Roff]

Aren't there two components somewhere in your circuit that make up 150Khz for the quench?

If I know the components for sure, then maybe I can just switch the values instantly, and everything will function better.

Yep. R1 and C1. Remember that C1 is also an RF bypass cap. I changed R1 from 10k to 50k, keeping the R1*C1 time constant at 33us, and the quench frequency changed from about 60kHz to 160 kHz. You don't have a scope, do you? As you can see from this experiment, it is difficult to predict the frequency.
I've been playing with another idea. I'll post it if anything comes of it.

 

[mstechca]

R1

that must be the base-bias resistor.

....keeping the time constant at 33us, and the quench frequency changed from about 60kHz to 160 kHz.

In terms of math, how does the time constant of 33us relate to the quench frequency. I don't understand how you got 60 Khz and 160 Khz.

You don't have a scope, do you?

No.

 

[Roff]

R1

that must be the base-bias resistor.

....keeping the time constant at 33us, and the quench frequency changed from about 60kHz to 160 kHz.

In terms of math, how does the time constant of 33us relate to the quench frequency. I don't understand how you got 60 Khz and 160 Khz.

I just explained that I don't know how to analyze it. I just chose 33us because that was the approximate R1C1 time constant in my first sim. My point was that you can't easily predict the quench frequency by knowing the time constant.
I got 60kHz and 160kHz by simulating the circuit with
1. R1=10k, C1=3.3nF (R1C1=33uS)
2. R1=50k, C1=660pF (R1C1=33uS)

The relationship isn't linear because:
1. The RF oscillation time is relatively constant for both cases.
2. The amplitude of the OFF voltage at the collector gets larger as R1 gets larger, due to the voltage divider formed by R2 and R1.

 

[mstechca]

I got 60kHz and 160kHz by simulating the circuit with
1. R1=10k, C1=3.3nF (R1C1=33uS)
2. R1=50k, C1=660pF (R1C1=33uS)

Is the 60Khz and the 160Khz the numbers the sim gave you, based on the R1 and C1 you provided, or is there a way I can figure these numbers out.

 

[Roff]

I got 60kHz and 160kHz by simulating the circuit with
1. R1=10k, C1=3.3nF (R1C1=33uS)
2. R1=50k, C1=660pF (R1C1=33uS)

Is the 60Khz and the 160Khz the numbers the sim gave you, based on the R1 and C1 you provided, or is there a way I can figure these numbers out.

Please read my previous post. I already answered this question.

 

[audioguru]

I didn't think that an "ordinary" oscillator from a simple FM transmitter would have squegging oscillation. I see that if it is overbiased so that its RF oscillations are mainly with the transistor saturated then the transistor would gradually conduct more and more until its operating point won't allow RF oscillations anymore. I think that the supply's internal resistance in series with R2 is important as well.

Most super-regen circuits use a bypassed emitter resistor as a peak detector so they set the squegging frequency, allowing for an adjustment of the transistor's base bias current to set its sensitivity.

 

[Roff]

I didn't think that an "ordinary" oscillator from a simple FM transmitter would have squegging oscillation. I see that if it is overbiased so that its RF oscillations are mainly with the transistor saturated then the transistor would gradually conduct more and more until its operating point won't allow RF oscillations anymore. I think that the supply's internal resistance in series with R2 is important as well.

Most super-regen circuits use a bypassed emitter resistor as a peak detector so they set the squegging frequency, allowing for an adjustment of the transistor's base bias current to set its sensitivity.

I'm not implying that this circuit is a good superregen receiver. I am just pointing out that, in simulations, it does squegg. I would be very surprised if a hardware version did not. I have seen plenty of sims that didn't oscillate when they should, but the opposite case is highly unlikely.