TheRadioBoard

Now that we know the antenna is connected to the hot end of the tank circuit, it could be loading the tank enough at some frequencies that it won't regenerate. The solution is to place a small value cap in series with the antenna, or else connect the antenna to a tap on the tank, or a completely separate antenna winding.

What you say is true, but I was looking at it the other way. By using an RF choke, you can use the same voltage (assuming you are not exceeding the max voltage of the FET) and greatly increase the gain of the stage. I checked the datasheet for the 2SK241 FET and the max drain-source voltage is 20v. Therefore, putting 9v on the drain should be no problem. That is less than 50% of the max allowable voltage for that FET.

Norman
KA1GUK

Given the phenomenal gain achieved by any regenerative amplifier, more gain may lead to more problems, like difficulty keeping the stage on either edge of oscillation. I'm running my detectors at 5.6-8 V and I think they'd be happy with less. Use the regen stage to get selectivity, and get the gain later.

Back in the 20s and 30s when people were poor and tubes and batteries were expensive, it made sense to beat as much gain as possible out of every stage. Not anymore. You can buy 40 dB of audio gain for a quarter.

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

Also correct, but when you are having trouble trying to get the stage to regenerate, increasing the gain is just one more thing to try to solve the problem. Checking to make sure the phase of the tickler coil is correct, increasing tickler turns, etc. are other problem solving steps. I'm just throwing more things to try out there.

Norman
KA1GUK

Hi folks,

Following KR1S's advice:



I added a 2K resistor in series with the pot and, exactly as predicted, the set now oscillates lower in the band as well.

I wanted to confirm what this is actually doing. KR1S said it "increases" the gain:



Whereas Dan mentioned it "decreases" the gain:



There seemed to be no disagreement with Dan's assessment, so is it right to say that this additional source reistance decreases the gain and drain current? What if I increased the source resistor even further? Then the set would always be oscillating, right?

This is my first time to play with (and build) a regen, so I'm still getting the hang of how it works and how it's supposed to work. After much fiddling with the tuning and regen control, I think I have the hang of it, but I have two serious problems:

1. Hand/stray capacitance. Currently the circuit is built on a tag strip using point-to-point construction, with no ground plane and no casing. When I move even slightly the earphone cable (about 0.5m long), the frequency varies widely and sometimes the set even pops out of oscillation. Also, when I move my hand toward the tuning capacitor or the regen control, the frequency varies slightly (though not as much as the frequency change caused by moving the the earphone cable). Grounding the set (using an alligator clip to connect ground to a cold water pipe or large metal object) prevents it from going into oscillation at all. (Note: the antenna is about 3m of wire strung inside the apartment.)

How can I fix this stray capacitance? Would mounting everything inside a thin aluminum case (dollar-store quality) help? Or do I need to rebuild the circuit using ground plane construction?

I find it rather surprising that the movement of the earphone cable can so greatly affect the tuning.

2. Hysteresis. When creeping up on the critical point, the set will suddenly "pop" into oscillation. This even happens if I creep up just before oscillation then don't touch any controls. After popping into oscillation, I need to back off on the regeneration pot (back further than the point at which it entered oscillation) to find the critical point again. Sometimes it will pop out of oscillation again, then I have to do a back-and-forth seesaw on the regeneration pot, while tuning across the band. Adjusting the regen pot also affects the tuning frequency slightly.

I gather that this isn't how a good regen is supposed to work... any tips here on increasing smoothness of regeneration?

Finally, I'm using a 200pf capacitor, which makes the 7mHz CW band frighteningly narrow. I plan to find a smaller cap or somehow bandspread my current arrangement (though reading KR1S's page on bandspreading makes it seem like more trouble than it's worth).

Even given all of these problems, I'm hooked on the idea of the regenerative receiver. It's amazing to hear a mush of jumbled signals suddenly emerge intelligibly when the critical regenerative point is reached.

Any advice on solving capacitance and hysteresis issues would be welcomed. Thanks!

It must increase the gain enough to overcome circuit losses, as it now oscillates at the low end.

Before too many people tell you too much, I suggest some simple experiments. Assuming you have a digital dc voltmeter (DMM), try measuring the source voltage as you adjust the pot. The source is the terminal the pot/resistor connects to. Then measure the total resistance of pot and fixed resistor (power off) from source to ground. Using Ohm's Law, dividing the voltage by the resistance will tell you the source current. Knowing the pot settings for oscillation at various places in the band you can determine the source voltage and current needed to make it oscillate.

Keep good notes. This experiment will pay off down the road!

Using separate Main and Bandspread tuning caps is indeed easier than trying to bandspread one cap that is too large!

73,

_________________

http://kr1s.kearman.com/
http://qrp.kearman.com/

qrp-gaijin:

I only added confusion to the discussion with my “gm” comment. I should have kept quiet.

Jim is right - the VOLTAGE gain is increasing as the source resistance is increasing. And it’s the voltage taken off the source that’s being fed back.

It is true that the transconductance, gm, is decreasing as the source resistance Rs is increasing. gm tells us how much the drain current (~ to the source current) changes for a small change in the gate-to-source voltage. It’s a CURRENT gain if you will.

But for a source follower - like I think you have - the VOLTAGE gain is given by:

Av = (gm*Rs)/(1 + gm*Rs).

So it’s a race -- gm is getting smaller as Rs is getting bigger. It works out that the voltage gain, Av, for the source follower, slowly increases toward “1” as Rs increases.

Sorry for adding confusion.

Sure hope I didn't screw-up the explanation and add even MORE confusion.

73, Dan

it can be further simplified to

Av=1/(1+1/(gm*Rs)).

a typical jfet has gm=5 - 10 mmhos, and Rs from a few hundred ohms to a few k ohmes, depending on the design. so typically 1/(gm*Rs)<<1, and that leads to

Av ~= 1.

very much like what you have for bjt followers.



gm is a given, at a given bias point. it would only change with Rs in the sense that higher Rs usually leads to different working points which may leads to different gm. But then not holding the bias constant pretty much invalidates any comparison.

Quick update and another question on this receiver. First, I increased source resistance to about 5K, which allowed the set to oscillate over the widest range of frequencies. Below or above this value reduced the usable amount of oscillation. I noticed that the set's ability to oscillate was also affected by the type of antenna coupling (directly connected to tickler or connected through capacitor) and the kind of output device I connected to the earphone terminals (e.g. it wouldn't oscillate when I tried to connect the earphone terminals to the audio in on my computer sound card, so I can't create any recordings of the audio output). Surprisingly to me, the experimentally-determined value of 5K source resistance also has the beneficial side effect of having almost eliminated hysteresis when moving in and out of oscillation. This makes the set much more pleasant to operate.

However, I have now discovered a new problem, which, after doing some reading, I think is called "fringe howl". The beat notes I get when trying to listen to CW aren't clean; they are raspy, screetchy. When tuning down to zero beat on a carrier, I get a "zipper-like" sound, whereas I expected a smoothly descending audio sine wave. I'm still very new to all of this, but I think a regen can do better than this, right? Are there any quick-fixes for this?

Anyway, maybe this circuit is inherently limited, so I think I'm going to try to build a better receiver and am tracking down parts for a Kitchin regen: http://www.electronics-tutorials.com/re ... ceiver.htm . I wanted to try his "high performance regen", but that uses some parts I can't get in Japan, so I'm going for the other design.

Once I get more knowledge and understanding of electronics and regens, I'd like to return to this circuit and do some of the analyses and modifications suggested (measuring source voltage/current required for oscillation, experimenting with amount and style of feedback, trying different coils), but for now my knowledge is too limited to really understand what I'm doing. So I think a Kitchin design would be a good next step to try.

I am considering building a simple regenerative receiver ... do you have any pictures of your version (from the Japanese site)?

I am not clear on the coil arrangement.

Also, do you have a "final" version of the schematic after your changes?

Let me dig it out over the weekend and I can post some pics. Be warned - as it was my first regen, it is not an example of good construction technique!

The main reason I built this circuit as my first regen is because it allows use of a pre-wound coil that is (or rather, used to be) available in Japan which made it simpler for me to build without worrying about coil winding and getting the phasing of the tickler correct. Also all the components are readily available in Japan. Outside of Japan you will probably need to substitute transistors. The 2SK241 is a single-gate depletion mode MOSFET. Probably a standard JFET (e.g. MPF102) would work here. The 2SC1815 is a cheap average-performance transistor (a 2N3904 would work here).

@Norm_N_Tam:



When I want to see the amregen.pdf in the link posted above, I get the following message:



Can anyone still provide this schematic, please?

http://web.archive.org/web/200805170021 ... mregen.pdf

Is it a Clapp using C-E capacitance as part of the capacitive voltage divider?

@Qrp-gaijin:


I think it's a Vackar oscillator.

Perhaps it could be modified to include Armstrong feedback, and we would have a first solid-state implementation of vladn's frequency-compensated dual-feedback regen. Any hints, vladn?