[KLQRP] Re: Terman on regen theory

[Gary Oneil]

Hi Sir John;

        Hey... I'm not ignoring nor avoiding you, but rather just found a 
time opening to respond... I'm going to try to link to your photo's up on 
the KnightLites site tonight... I think just a pointer to where you have 
them, unless you want me to mirror it there. Your choice, either works 
fine for me.

        I've been reviewing Terman's paper, and I have to say, it's not 
something I find easy to read when I'm heading off to bed. :-)  I have 
managed to make my way through it though. I also need to re-read Alan's 
comments, because he may have mis-typed... or I mis-read his comments as 
well... or I misunderstood him... Terman... or both. :-)  At this point, 
all options are equally viable. LOL!

        The first order of business though is with respect to your 
schematic... In your second note, you state the B-battery consists of a 
quantity of nine 9V transistor radio batteries... 81 Volts? or five 
batteries?

        A 9 volt battery drops to around 6 volts as I recall by the end of 
it's amp-hour lifetime, putting your plate voltage at around 30 volts. In 
the case of both batteries, their amp-hour rating is usually considerably 
lower at low current draws. This is due in part to the contribution of the 
rate of self-discharge. Not that this information is critical, but I think 
the estimates of 100 hours for the D-Cell and 1000 hours for the 9V 
transistor battery are a bit optimistic. It would be useful information to 
catalog however... Can you stand to listen to 40 meters that long? LOL!

        Finally... I'm curious about whether or not you would notice any 
difference in performance or distortion if you jumpered around the 2.5 mHy 
RF choke.  Is the omission of the RF sensitive parasitics otherwise 
present  with direct coupling to the audio transformer improve the 
stability of tuning for example? It would seem you could bypass this with 
a jumper and do a "quick and dirty" comparison. It would seem more robust 
with the choke in place, as it decouples any dynamic parasitic changes 
(from hand capacitance etc.) introduced in or near the audio section, so I 
wouldn't remove it by any means. It is an interesting curiosity though.

        I love that little audio bias network. Simple and elegant. The 
tube is biased to cut-off when no audio is present, and it appears to self 
bias itself into what I would guess to be Class B in the presence of audio 
above a few Hz. Furthermore, it would seem to dynamically adjust the bias 
just meet the Class-B  requirement and no more, thus minimizing plate 
current while allowing the tube to operate at it's maximum gain... i.e. 
grounded emitter. Simple, subtle, and very slick. I'm really mesmerized by 
this little trick. 

        I think I'll ponder Alan's note a bit more before I comment, and 
see if it's him or me that's tongue twisted. :-) His math skills are more 
polished than mine, and I think this helps him think a bit faster than I 
do. I have to ponder every little detail and work through it before I feel 
comfortable I have a good command of what's going on. He and I talked this 
through Thursday night, but what he wrote didn't seem quite as clear, and 
I think counter to my understanding. The concept is intact, but the 
details I think got flipped. That said, we both agree... regenerative 
detection appears to show some merit of being viable. LOL!

        Thank you again for sharing this wonderful experience with us at 
the PigOut. You are a true artisan, and a model representative of what I 
perceive as the foundation of the KnightLites mission.

        For those who missed the demo of Sir John's regenerative receiver 
Thursday evening, perhaps we can ask him to bring it along to next months 
Pig Pickin' and let us twiddle it's controls during daylight hours. It has 
a wonderful high quality real radio sound, and when strategically placed 
will allow you to join in on the chorus. :-) Stay tuned to 
www.knightlites.org/projects for a link to his schematic and photos.

72

Gary,  N3GO

Regards;

Gary E. O'Neil
Applications Engineering
PowerPC Enablement and  Applications
IBM Microelectronics 
3039 Cornwallis Road, 062/E112
Research Triangle Park, NC  27709
(919)-543-5750    T/L 441-5750


Resolution of disagreement ALWAYS leads to better understanding... 
assuming all parties survive the conflict. 



John Marshall <johnmars at mindspring.com> 
08/18/2006 12:01 PM

To
"Victor, Alan" <avictor at harris.com>
cc
Gary Oneil/Raleigh/IBM at IBMUS
Subject
Re: Terman on regen theory







On Aug 18, 2006, at 11:13 AM, Victor, Alan wrote:

> You beat meet, NO pun intended!

Hey, just because I stayed up late doesn't mean I was doing that! 
Besides, I'm a consenting adult, in the privacy of my own garage...

> Well I lost sleep on this last night and got my Terman book out as 
> well!! First, I WAS BRAIN DEAD in my thinking on how this works. A 
> short drive down the road and it hit me, of course. FIRST, and I have 
> not found the paper yet, T.S. GEORGE is the prof at U of F I was 
> trying to recall, who did a great paper on the regen in the Proc. of 
> the IRE. I will try to find. But getting back on the "how does this 
> work simple idea"... The open loop gain is the open loop product of 
> the amplifier and the resonator. The resonator (L/C..tickler whatever) 
> has some loss, while the tube has voltage gain, power gain, 
> whatever,....gmRL etc... The product of the open loop gain (AB)  needs 
> to be slightly larger than ONE. Since the feedback is positive we have 
> the CLOSED LOOP GAIN given by Gain(CL)= A/(1-AB) where A the the 
> Amplifier Gain and B is the resonator loss. As this product of open 
> loop gain (AB) goes to ONE, recall it is above one but the reduction 
> to more or less ONE is created by gm reduction, tube compression, 
> etc... or reduction in tube impedance!! In any case the CLOSED LOOP 
> GAIN will now go to INFINITY, VIOLA the signal input will experience a 
> HUGE GAIN, VIOLA great sensitivity. Now in some text books you will 
> see closed loop gain written as A/(1+AB) or A/ (1-AB). Obviously both 
> CAN GO TO infinity but one requires  both the products of the phase 
> shifts to be ZERO or 360 degrees, while the OTHER implies that one of 
> the blocks (usually A) already has 180 degrees of shift, so the other 
> block MUST provide 180 degrees of phase shift as well. Now the loss of 
> sensitivity as the REGEN CONTROL controls open loop gain, is clear. As 
> the open loop gain is INCREASED, in order to have the CLOSED LOOP GAIN 
> goto ONE, the level of COMPRESSION must increase. One way this is 
> accomplished is by reduction in the impedance level, and I think John 
> mentioned this. That makes sense, so to much OPEN LOOP GAIN, and in 
> essence that is what we are controlling in the REGEN CONTROL, can 
> REDUCE the sensitivity. Again, while the regen detector is 
> oscillating, the closed loop gain is HUGE. The input signal IS SUBJECT 
> to this HUGE closed loop gain, and I can appreciate Terman's words, in 
> effect the input signal has control on this closed loop gain value. 
> The math details of creating the beat, is important, because I do not 
> think the process is like a SELF OSCILLATING MIXER. That is the beat 
> is not created by mixing action (non linear) in the tube, but simply 
> by modification of the loop gain. The simple math question, can I 
> produce sum and difference terms by modulation of the GAIN? Latter! 
> Alan.

Now that makes perfect sense and it ties right in with Terman's 
description.  Thanks, Alan!

John, KU4AF


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.applegate.org/pipermail/klqrp/attachments/20060821/4423f415/attachment.html