Alternatives to YAEWSBM and EPSE01 module

The improvements below are mostly suited for new design (based on EW schematics) than for existing EW boards (cutting  traces and desoldering  may be required).

First version - using only 4 passive components (marked by red):


The principle is simply: we get the RF signals not from the LC circuits but from the collectors of Q2 and Q4. With  resistors R101 and R102, the transistors Q2 and Q4 are operated as some kind of output buffers. 

Output RF voltages depend on R101 and R102, and with values of 470 ohm we get the 4 Volt peak-to-peak magnitude. The caps C101 and C102 are not critical and can be 200pF ... 0.1uF.

Waveforms from oscillators and mixer are presented below:


The next version is a simplified equivalent of first:

Results.
The original circuit had the lowest beat frequency 90Hz (before the oscilators will be locking); the new schematics allow notes going down 15...20Hz.

As for me, I prefer an active mixer (let's call it "TBM01" = "transistor-based mixer"). It allows fully decouple the  oscillators and reach the ultra low beat frequencies:


Mixer's components are marked by red. R101 and R104 help decoupling. The Q101 is a general purpose p-n-p transistor (I  used MMBT3640 because it was in my collection).

Resistor R103 (330k) defines the amplitude of output audio; you may need another value because transistors have a dispersion of parameters (a current gain, in particular). 

The waveforms of output audio at different beat frequencies:

As you can see, the resulting signal is turned into positive area. Because of that the "Waveform" and "Brightness"  controls will operate in another manner.

It's remarkable that the waveforms have a distinctive point on the curve, the twisting piece, which helps tuning: when it located left of the peak then we are in the right (positive) beat zone. A mirrored curve signals that we have reached a wrong (negative) beat zone:

A second notable feature. The signal has a different manner of morphing when approaching to zero point, depending on whether we are in a  positive or negative beat zone. Different shapes -- different harmonic content. Still no ideas how it can be used in practice. 

Results.
The new synchronization threshold is so low that it's difficultly measured because of unstable environment (air flows,  vibrations, mechanic tension relaxation, and so on). On subjective sensations it lies below 4 Hz.

And finally, I present a well-repeatable (I hope) schematic, the "TBM02" version:


Now the amplitude of output signal is irrelevant to Q101 and depends only of R24/R103 (R23 is not necessary, but with  C101 it helps filtering).

Again, here are the waveforms of output audio (at different beat frequencies) in positive zone:

in negative zone:  


PS. All the results was obtained with prototype PCB 8x6 cm (L5 and L6 are mounted in hole, other components are SMD).  
The VPO and FPO have the areas 2x2 cm each, the mixer zone (0.5 cm width) distances the VPO and FPO. No shielding  between them. Antenna circuits are disconnected.

I only fear that it will not work correctly in an Etherwave without further modification around the following LM13700 wave shaping circuit because the latter expects the mixer signal to have a negative voltage offset, the peaks being between 0V max and -0.6V min. Thus the signal from your active mixer is completely out of range and will in the best case give an ugly waveform and in the worst case nothing at all at the LM13700 output.

Second, the additional resistors at the collectors of the oscillator transistors will increase the inner resistance of the oscillators and thus reduce the stability, similar to using inductors with a low Q factor.

You should better analyze, think and test things (including field tests on at least 10 Etherwaves with different circuit board versions) thoroughly from A to Z instead of publishing half baked stuff. That's highly unprofessional! 

"I only fear that it will not work correctly in an Etherwave without further modification around the following LM13700 wave shaping circuit because the latter expects the mixer signal to have a negative voltage offset, the peaks being between 0V max and -0.6V min. Thus the signal from your active mixer is completely out of range and will in the best case give an ugly waveform and in the worst case nothing at all at the LM13700 output.

Second, the additional resistors at the collectors of the oscillator transistors will increase the inner resistance of the oscillators and thus reduce the stability, similar to using inductors with a low Q factor.

You should better analyze, think and test things (including field tests on at least 10 Etherwaves with different circuit board versions) thoroughly from A to Z instead of publishing half baked stuff. That's highly unprofessional!" - Thierry

Mr.Pro dosn't know how to shift levels?
Mr.Pro dosn't know how to stably get aroused on naked/shielded Miss Inductance? 

Very interesting ILYA!  Mixing on the EWS is pretty easy to do in a lot of ways due to the large voltage swings at the tap points.  I played around with low voltage mixers for a while over on the analog thread but never hit on the single transistor topology you've found.  I'd be interested in hearing one or more MP3s of your results </RS>.

Just a native audio from mixer (without articulation) is possible at the moment. Waiting for shipping of LM13700s.  

The mp3 audio across the R24 (from negative beat side to positive): here.  here.

ILYA, is there anywhere else less obnoxious you might stick that MP3 file?  I downloaded the *.exe and ran it twice but couldn't figure out the correct combination of responses to just get your MP3 file and not all that other junk they're pushing.

ok, was reloaded to another hosting (the link is above).