TheRadioBoard

I said and I quote"You do have after all a regenerative detector"
I did not say it was a regen. One of the problems you run into expecially with old circuits is ant coupling or coupling to the gridcircuit/detector) overloads the tube. It does not take very much, The difference between good perfomance and a relatively dead set is tiny. I have explained a relaible way to test it and offered a sugggestion fromthe school of bin there myself You can do anything you want. Change tubes , rewind all your coils. Play with gridleaks but if that is the problem none of that will make any diffference. But if you dont try ? How do you think I learned this? it wasn't fun I asure you. And no one not once ever told me. MY father a former Collins design engineer would tell me if I was in your shoes this(and did a hundred times when i was a kid)" One thing is for sure Don, If you wont ry it it wont work."
Don VE3LYX

I truly appreciate all the suggestions that have been offered. I'd like to clarify a few things that have been raised and so I've tried to do so topic by topic. 1) Yes, the detector is regenerative. That's key to the design. 2) I've reversed the leads on L5 and, as expected, the detector goes into regeneration with it wired one way but not the other. 3) It is most definitely not a tracking problem. The antenna tank circuit and LO are separately tuned and as I reported earlier I can follow the injected signal from 5.6 to 7.5 mHz. 4) I don't have selective level meter or a spectrum analyzer, and am not trained to use sophisticated equipment like that anyway. 5) I have used the loosest possible coupling of the antenna but it is still deaf.

I neglected to report that I did tinker with the 'grid leak' in the LO with no discernible change in results. Sorry if I didn't report that last time.

So, that's where it stands for now. I need a break from this thing for a while. The garden needs weeding and anyway there's a regenerative receiver I've wanted to build for a while now. I'll come back to the Super Gainer before long, but I need to let my frustration level subside first. Thank you all!

_________________
Tom

A little food for thought at this site:

http://home.comcast.net/~btse1/vintrad/hb/dr.htm

The author does say 'it is a double regenerative, and it takes a lot of knob-twiddling to get a readable signal'. he also calls it a 'fun and worthwhile little project'. I suppose that could be true, after you get past all the knob-twiddling and other issues mentioned in this thread!

Norman
KA1GUK

The more I look at the Jones circuit the weirder it looks to me besides the suppressor grid DC level. Personally I'd do few things differently:
1. Why LO and RF are tuned separately and only fine tuning is ganged ? I'd personally use normal superhet tracked RF/LO tuning plus a fine tuning cap at the RF tank to compensate tracking error. This tracking error compensation is important here because Q-multipled RF can have very narrow bandwidth. But it also allows experiments with rising individual sidebands (using small intentional tracking offset towards either upper or lower sideband).
2. Regenerating coupled tanks (IF transformer) works, but may not be optimal - rising Q will create overcritical coupling with two closely spaced but very distinct spikes. It may be better/easier/cleaner to use plain single-tuned regen tank behind the converter.
3. It looks like the second detector is plate detector. The 25uF cap is not a grid leak, more like part of cathode bias circuit. Can't read the unit on a resistor parallel to it. I would rather use one of the tried and proven grid-leak or cathode-style detector stage.
4. There has to be fairly high level of LO injection into suppressor grid of the conventional pentode. I would go as far as try connecting suppressor directly to the hot end of the LO tank coil. No RC network needed, this point is grounded for DC. This may work well because in a conventional pentode suppressor grid is rather sparse, hence should have little parasitic capacitance and minimal LO stability impact.

I think the overall topology can work exceptionally well, just don't quite like the original circuitry.

Good catch on the detector. I never noticed that. I saw the capacitor resistor combination, and assumed a standard grid leak detector.

The later 1938 version did use a standard grid leak arrangement.

For reference, here is the 1938 version:

The way 6J8G is drawn makes my head spin . Hard to figure out electrode assignments. It is supposed to be a triode-heptode right ?

Unfortunately, the issue you just pointed out was fairly common in the 1930's. See the 2007 update in Dave's 'Pentaflex' write-up for another example:

http://www.makearadio.com/tube/pentaflex.php

Norman
KA1GUK

Probably worth reiterating again (thanks, Yogi) that there's no B+ on the Suppressor Grid on the 1938 version. Its tied to the cathode internally.

I'm beginning to wonder if the original Jones circuit actually worked or if it was a lucky fluke that made it to print? Does anyone have experience with a working one?

-Bill

Even better option for the original circuit is to connect 6C6 suppressor grid directly to the grid of 76 LO. That point has both negative bias (due to LO rectification) as well as high LO amplitude, so the conversion gain should be good.

Also, controlling RF Q-multipler via screen DC is not good in this arrangement, because it also changes conversion transconductance. A better way would be a potentiometer in the cathode circuit, shunted by a capacitor from the center tap to one of the sides.

Here is a sketch to illustrate possible rearrangement. It assumes fairly low B+, otherwise some current limiting may be needed for 76 plate and 6c6 screen. LO is Hartley (purely for aesthetics and symmetry ), but can also be Armstrong (to re-use existing magnetics).

Why using Q-multiplication in the Mixer/RF-stage? You should beter use it in the IF-stage. The tank circuit in the RF-stage is only intended for supressing image frequencies and other unwanted signals.

As the original design name suggests - it allows very high gain from two successive regenerative stages.

For me what is interesting about this arrangement is the possibility of selective USB/LSB reception without a dedicated BFO or true sync detector. It is frequently helpful even for AM stations when one of the sidebands has close interference.

The intended tuning range of the receiver discussed here lies I believe around 7 MHz. Your statement seems to imply that a regenerative RF front end, operating at 7 MHz, can be selective enough to slice off (or attenuate) one sideband. Am I understanding you correctly? I have been wondering for some time just how narrow the bandwidth of a 7 MHz LC regen really can be. I've been meaning to take some measurements but haven't gotten around to it yet.

On a different note, I understand that regeneration in the front end always degrades the S/N ratio and is not recommended for weak signal reception.

Assuming the band is "open" (40 M after dark), if the noise level drops when you disconnect the antenna, you have all the S/N ratio you need. Any regen that doesn't have any type of rf amplifier (with or without gain) ahead of the detector, has by definition a regenerative front end. On any HF band where I've used a regen, external noise always exceeds the internal noise floor.

73,

_________________

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

I think RF Q of few thousand should be achievable, hence some measurable/useful effect of sideband attenuation can be obtained even at 7MHz. Note two things in the circuit I posted (but not tested):

1. Front end is a pure Q-multipler, not intended to work beyond oscillation threshold. There is no noisy grid leak circuit. The device is biased at high current / high Gm point (lowest noise, maximum linearity). This is not the case for regenerative detectors.

2. The un-shunted part of cathode potentiometer has an effect of broadband NFB that should improve loop gain stability, therefore allow operation at higher Q without breaking into oscillation [there is a reference, I'll edit the post when I find it].

There is question how much antenna coupling would affect maximum achievable Q. I do not know. Perhaps antenna matching network with wide impedance ratio range will help. (I think in general regen front ends need quite large impedance transformation ratio from antenna to the tank, especially if optimized for max selectivity rather than syncrodyne operation.) A buffer in front of Q-multiplier would help also.

Very interesting. I had speculated about a similar idea in the solid state forum to get a single-signal regenerative receiver (regenerative Q multiplier front-end to slice off one sideband, followed by a regenerative detector). I imagine it would be very fiddly though with 2 regenerative stages operating at very close frequencies.

Back to your idea, some questions:

1. Does the "high current / high Gm for lowest noise and maximum linearity" rule apply for solid state devices also (FET, BJT)?

2. Regarding negative feedback to "allow operation at higher Q without breaking into oscillation": I've been very curious about this for a while. Charles Kitchin (N1TEV) has also mentioned in some of his articles the importance of avoiding premature oscillation to ensure highest Q; as I recall, he uses the throttle cap for this reason, and also I recall reading a reference that one of his designs uses a diode somewhere (in the emitter of a BJT?) to provide negative feedback and allow operation closer to the operation threshold. But in practice, how do you know if you are breaking into oscillation "prematurely"? In my experience almost any regen design can be tamed, by adjusting circuit parameters, to have "smooth" transition between non-oscillating and oscillating states. But this issue of negative feedback to allow operation at higher Q seems to imply that there is more to high-Q operation than just "smooth" regeneration control.

In practice, how would one ever even know if oscillation is occurring too soon (lower Q) or at a more delayed point due to negative feedback (higher Q)? The only way I can think of to measure if you're closer to the oscillation threshold is to measure the frequency response of the stage at threshold, for various values of negative feedback, to see if you are achieving narrower bandwidth or not. Are there any qualitative checks one can perform, when operating a Q multiplier below oscillation, to evaluate if the operation at threshold is higher Q or lower Q?

EDIT: I found the reference to the use of a diode to delay onset of oscillation improving below-threshold performance. It's in Bill Young WD5HOH's cascade regenerative receiver, Jan/Feb 2004 QEX. The author uses a diode from source to ground, bypassed by 0.1uF, in a 2N3819 FET detector to apparently allow higher gain below oscillation by delaying the onset of oscillation.