6ME8 1-tube superhet - A modest Proposal

[BobWeaver]

While discussing the 6ME8 direct conversion receiver here, Vlad had brought up the possibility of using a 6ME8 as a one tube superhet and emailed me his proposal:


Meanwhile, I'd also had some fleeting thoughts of this and had sketched a proposed circuit as well:


Fairly similar, but there are a couple of differences.

Vlad's idea of connecting the IF transformer #1 primary between the plates looks like a much better idea than mine. I'd been thinking along the lines of the audio detector transformer from my direct conversion circuit, but a center tapped IFT would have to be custom job.

On the other hand I see a bit of a problem with how the reflexed signals are combined in Vlad's circuit. The rotor of tuning capacitor should be grounded to avoid special mounting and hand capacitance problems. To handle this, I had used an extra winding on the RF coil to feed the signal through.

I think these two circuits could be combined, to produce something interesting, and possibly even functional.

[vladn]

I think IF1 primary connection in parallel to LO exitation is not optimal - they will shunt each other at both resonanse frequencies reducing the Q - I guess having weak inductive coupling in IF1 and weak capacitive coupling in LO may somewhat alleviate that but still. I think the correct way is to have LO and IF1 tanks connected in series. BTW am I correct that a cap in series with LO excitation coil is strictly to adjust the coupling ?

Input connection does not require separate RF coil if you change the order. RF tank can be grounded, followed by IF1 output followed by floating AF filter (especially if volume control can be replaced with AGC control as below). Not a big deal just trying to simplify things. Another option would be to move audio on a separate grid but I am not sure how much worse intermod would be.

BTW I really like Bob's 2-tube superhet idea for IF regen and AGC combo. The ability to continuously adjust IF bandwidth via IF regen control without volume change on such small superhet RX is impressive. I am not 100% sure BFO function from IF regen will work in this setup but it is definitely worth a try. AGC would be a must for maintaining very low level of IF BFO oscillations. Maybe volume control should be done via adjusting AGC level instead.

[vladn]

I've just realised that for AGC you need detector output referenced to ground so I guess it is ether extra RF winding like on your schematic or audio to grid 2.

[BobWeaver]

I think I'm going to sleep on this tonight.

[vladn]

I think the most important issue in this design is to make IF (and to some extent AF) amplification as linear as possible so prevent parasitic spurs from harmonics bleed between reflexing amplification bands. If the IF harmonic suppression in RF band is less than RF gain it may cause runaway oscillation when RX tuned to the IF harmonic spur.

Fortunately tube harmonic energy is mostly concentrated in the low order harmonics so it should be less of an issue for AF-IF separation (even though AF signal is of larger magnitude). However IF to RF harmonic bleed separation is a big deal in this design. I think it can be solved by:

- selecting IF frequency to be less than half of the bottom of RF band avoiding the strongest 2nd harmonic of IF bleeding into RF;
- using AGC to keep IF level in the small signal region that ensures better IF amplification linearity, this will also avoid runaway oscillation that could make it difficult to tune away from the parasitic spur;
- possibly linearizing the tube amplification by adding a cathode resistor trading off some gain for linearity, could also be used for manual gain control or volume control.

[BobWeaver]

Just a few comments on some things you mentioned.

The reason for the capacitor in series with the local osc. primary was simply a leftover from the direct conversion receiver to prevent the audio signal from being shunted through the osc, coil and disappearing. Since there shouldn't be any audio frequency diff mode signal in the superhet, the capacitor shouldn't be necessary.

You made a few comments about problem of harmonic generation due to tube nonlinearities. The one good thing about beam deflection tubes is that they are noted for excellent liinearity. I can vouch for that, because in the direct conversion set, there were ample opportunities for cross modulation of signals, yet I had virtually no problem in that regard.

Regarding AGC, I'm not sure what the cutoff characteristics of this tube are. I suspect it's sharp cutoff. However, AGC could still work to a certain extent, but it would likely be touchy. Maybe trying to get AGC in something like this is asking too much. Anyway it's just a matter of observing correct polarity of the detector diode, and carrying the DC through back to the grid.

I received this schematic and comments by email:

[BobWeaver]

I'm not sure how that cathode resistor and inductor would work out. Generally the DC should be forced through the resistor to put the cathode a some small positive voltage. Then the grid is biased slightly negative with respect to the cathode. This arrangement would keep the cathode and grid at the same potential.

I suspect that you don't need to be too concerned about things affecting the local oscillator. I found that in the direct conversion set, this oscillator is virtually impervious to anything that I threw at it. One would normally be concerned about the oscillator being pulled off frequency by a very strong RF signal, but this one was so stable that it needed a fine tuning control to get it close enough to the RF signal to lock in.

Regarding your final points, I think you are looking too far ahead, and risk overdesigning things. A prototype probably needs to be built now to see how it really behaves. Many of the concerns may be unwarranted, and other problems, not considered, may show up. A prototype would show which way to proceed.

[vladn]

Clarification on last cathode circuit - assuming that the tube is biased strictly by the grid leak RC circuit (also part of AF detector) with cathode DC at ground the inductor works similar to a current source in the long tail pair. When only inductor is present the common mode gain is virtually zero while differential mode gain (LO) is unaffected. When resistive shunting of the inductor is increased so is common mode gain. Assuming an output tube impedance of ~10k a pot of comparable value will vary common mode gain from 1 to max with no affect on LO.

There is an issue with the diode detector polarity on all diagrams so far - if the grid is DC coupled to the detector then grid leak current will reverse bias the diode. Yet in this orientation it forms a pseudo-AGC since the AM carrier increases negative offset reducing tube gain, while reversing it will do the opposite. So reversing the diode polarity in that sense is undesirable. Not sure yet how to solve it.

I've downloaded 6ME8 spec:
http://tubedata.itchurch.org/sheets/049/6/6ME8.pdf
Looks like it is speced for 14ma sum of plate currents with max being 30ma which corresponds to gr1 bias of approx -5.8V and -3.3V respectively. I am not sure that grid leak bias would be sufficient for setting proper operating point. Cathode resistor is probably needed.

Also I may be wrong but from I've read about BD tubes is they are extremely linear for differential mode amplification. Not so sure about common mode.

[vladn]

Q for the tube guys - does the tube diode have negative voltage cutoff ? If so I can understand how it can work in DC coupled AGC configuration as the negative cutoff compensates for negative bias from grid leak. Or maybe I am completely confused.

[BobWeaver]

In any circuits I've seen, there is no grid leak bias when AGC is used. The AGC provides all of the bias in order to control the gain. The grid may be connected to the AGC line through a large resistor to isolated it from other stages that are also controlled by AGC, but in this case the resistor is not acting as a grid leak.

[vladn]

I need to think some more on detector design, gain control and biasing. I really like the idea of negative feedback manual gain control. If it is done at the cathode the additional benefit is that extra winding in RF tank is not really needed since AF filter can be floating.

[vladn]

Is there any data on steady state grid leak current vs grid bias (in the negative bias voltage region) for 6ME8 or other tubes ?

[vladn]

I did a quick and dirty simulation of the mixer / IF reflexing topology in ltspice and it seems to work as expected. I do not have BT tube models so I just build equivalent circuit out of FETs.

BTW I think I found 1AD reflexing superhet circuit by Terry Judkins that Bob referred to :
http://home.comcast.net/~suptj ... erhet2.GIF
He uses IF regen, not true IF reflexing, only AF is reflexed. Great idea but different.

After thinking some more on AGC in BDT superhet I realized that the main difficulty is the impact of AGC on AF amplification. Strong stations involving AGC will limit the tube current and AF power headroom. The funny result is that you can only listen to weak/medium signals at high volume level .

Standard AGC design will actually overcompensate by dropping AF gain. While it is easy to compensate for that by introducing the AF-only negative feedback, the AF amp headroom will be greatly reduced at low currents and it will go into clipping.

One possible solution is to add another tube for power amplification (6BQ5, 6AQ5 or similar) and use BTD AF as preamp in small signal mode with negative feedback used to linearise the amp and negating AGC impact on AF gain. But this is a whole different project - big transformer, box for a nice large loudspeaker etc.

[BobWeaver]

[vladn]

I am actually thinking that off the shelf common mode toroidal chokes can be used for all transformers except audio and IF with just few turns added for L15 and L3. The L4/L5 one has to have large enough inductance not to shunt IF frequency. These chokes usually use powdered iron cores, not RF quality but may work OK.

[BobWeaver]

My past experience with non-RF ferrite has been quite bad. Anything designed for RFI suppression tends to use very lossy ferrite, so that the ferrite losses account for as much RFI suppression as the inductance does. I don't think I would chance it. I would go with a type 61 ferrite.

[Curt Reed]

I think my bad past experiences with toroids may have been because the wrong core material was used. Thirty years ago I did considerable experimenting with them and my experiments were all a flop. Most of the cores were salvaged from junk boxes found at hamfests and such and the core material was anyone's guess.

Someday I hope to get back into them and take up where I left off, but I will be better armed with more knowledge about them now than I had back then.
Curt
_________________
Curt, N7AH
(Connoisseur of the cold 807) CW forever!

[vladn]

Did some reading on RF core materials - you are right on EMI chokes, probably wont work. Also there is apparently an issue with temp stability with most core materials even RF ones, not good for LO.

However for L4/L5 an off the shelf pulse transformer should work just fine. DC current is compensated between two coils so there should be no core saturation issues. Needs inductance at least in the tens of mH though.

I've started ordering some parts for this thingy. Main question is where do you get 262 KHz IF cans ? My guess is these have to be tube-specific for the reasons of proper impedance matching.

[Norm Leal]

Hi

You can buy 262 KHz IF Transformers here:

http://www.oldradioparts.com/2a6fla.txt

There is an earlier post about making your own IF Transformers using two RF chokes mounted close together for coupling. Should be able to search for this information.
_________________
Norm

[BobWeaver]