The example shown is just as an exercise in style, to demonstrate the feasibility.
But we all know that because something is doable doesn't mean it is worth doing. And that's especially true in engineering.
Applying this kind of positive feedback has exactly the opposite effect of negative FB: it increases non-linearities, sensitivity to temperature, decreases bandwidth, etc.
And to answer specifically your question:
That point is even before any positive feedback is applied: I consider a basic gain stage is already too nonlinear and unstable to my taste, and I systematically apply NFB, never PFB
There are some exceptions: some forms of bootstrapping are in effect a positive feedback, and some topologies, such as the cross-quad and the tringlotron can use limited and controlled amounts of positive feedback to linearize an amplifier.
But these are very special cases, and the PFB is not used to increase the gain (0.1dB is not a substantial increase).
But as soon as you reinject, G increases and if 1/G is derived from a resistive divider it dosnt change so again the gain increases to infinity (oscillation).
I would go with the opamp and gaurantee a constant gain, bandwith, stability, high input, low output impedances and audio quality. ( do 2 more pins really make it more complicated, you might actually simplify the rest of the circuit to where the complete opamp cct is actualy less complicated.)
My second choice would be to use a VAS with a current source load. (much more complicated )
Yes, I began to realise this after my post. This is why a grounded-base oscillator needs a transformer.
Provided you don't have too much feedback, an audio regen could remain stable and broadband. An audio superregen is different, as it is a quenched oscillation and an oscillation has a frequency so superregen is inherently narrowband. OK for receiving a particular audio tone, but not so good for speech.
Upon further thinking, this statement is probably wrong, but I think its not easy to keep the positive feedback low and consistant.
I am not considering speech for that particular application as it should work only on single tones, morse code reception. The local oscillator and the RF have a difference of about 1KHz so the directly converted output will be an audio tone at about 1KHz (maybe 0.5-2KHz or so) so high fidelity os not of concern.
minimal design means increased battery life in portable operation. An ELF receiver I have for example, operates using a series of 3 transistor amplifiers has a battery life of 3 months or so...
A darlington transistor could be a good approach but.. darlingtons are integrated circuits actually as they contain two transistors in a single package 🙂
I have not stick with positive feedback, if I find a high gain single transistor amplifier I would be happy. I was just thinking of positive feedback as a way to sqeeze even more gain. More audio gain in receivers could mean more easily to detect lower powered signals.
The audio should have enough volume to be heard easily on a silent room using a small speaker.
So it's a direct-conversion CW rx. You already have an LO using an active device and what? RF amp for pre-detector selectivity + heterodyne detector? This is hardly minimal. Plus you lose half your selectivity because of the image. Or LO, a 90 degree phase shift, 2 diode ring mixers, LPF, AF 90 degrees + combiner. Or you can have Weaver with more components.
Swings and roundabouts. If you really want minimal, then the super-regen is hard to beat.
w
Swings and roundabouts. If you really want minimal, then the super-regen is hard to beat.
w
I am sorry, but that is not the point of the discussion. But for your curiosity this is a part of a power efficient minimal TRX with DCR which uses the same transistor for LO of TX and RX too, followed by a very efficient class-E amplifier to produce 2W output on 50 ohm. Yet must use no more than 3 transistors in total to keep it minimal.
So the single transistor audio amplifier must deliver as much volume as possible, and that is what I am interested in.
Last edited:
Oh, and there is something else I forgot to mention about these regens, the need for special tapped coils spoils already simplicity. Better to have a two transistor RX using no coils (or at least simple to make coils) than having a single transistor multi-coil design like the macrohenrydyne.
Why do we so often have to wring information out of the OP? We have now just been told that it is for CW, so a narrowband audio stage is fine. We have just been told that it is part of a three transistor transceiver. Is this a club construction contest we have been helping with?
By the way, you may not get far driving a Class E amp directly from the oscillator, as Class E needs something approaching a square wave drive.
By the way, you may not get far driving a Class E amp directly from the oscillator, as Class E needs something approaching a square wave drive.
No contest clubs here, just me (Το εργαστήρι του 711). The positive feedback audio amp is of interest in other projects too, but I did not know of the limitations so I couldn't judge what is is important to mention or not, when I ask in this forum. From the time bandwidth limitation was issued I "revealed" that bandwidth is not a problem.
About the class-E thing, I would like to test the efficiency feeding the RF amplifier with much more signal than needed or having a pair of diodes to clip the output of the oscillator to produce a more square wave, but for the time being I consider only this audio amplifier section and the positive feedback as a small efficient amplifier for other such projects.
RF, audio efficiency today would employ IC technology - a "3 transistor" discrete design is going to waste power and require more total Si area since you simply can't buy discrete active devices with the same device size, processing technology - sub micron CMOS, vertical bjt structrues, lift off SOI...
Yeah, this isn't really true. Obviously if the audio is really quiet then there may be a problem hearing it, but detecting lower powered signals depends on the overall Noise Figure of the receiver.
It's probably at least as important not to throw away any volume by using an inefficient speaker.
So the design is actually driven by some kind of ideology? You don't want to use a single-package darlington because it's not 'one transistor'?
Volume = power. So it's not just a question of voltage gain which is what is commonly considered in audio. Does the detector produce its output in a 50 ohm environment? What is the power?
The power available from any given device depends on the efficiency and hence on the class of operation. Since there is no fidelity required, there is no necessity to operate in class A. The closer the transistor operation is to switching the less the dissipation in the device. A self-oscillating class D amplifier requires a differential amp (I think), but class B and C do not. Gain, power and device dissipation are more complicated to estimate in these classes however.
w
Thanks. You've got a nice collection of articles on the tunnel diodes, I'll take a look at it.
Also, I saw on the Internet some circuits where a 2N2222 is used to get a negative impedance device. Do you have articles on that too ?
I gave an explanation here:
http://www.diyaudio.com/forums/solid-state/173005-improved-current-source-sink-6.html#post2731765
This threads also contains various examples of negative resistance devices, simulated lambda or tunnel diodes, etc
Here is one link to 2N2222 using as a negative resistance:
A Negative Differential Resistance Oscillator with a Negistor
- Status
- Not open for further replies.