To all readers, I added more comments about the Wien Bridge oscilator at
http://www.turneraudio.com.au/Wien-b...scillator.html
The text explains some minor changes.
I did find that above 500kHz, the square wave became slightly assymetrical so I've added R1a 150r plus C1a 100pF across R1 5k6 feed resistance to input base of Q9 at the Schmitt Trigger circuit.
The square wave at 1MHz now looks much better, and one whole lot better than in the original crummy low budget BWD 141.
But, I still have one problem. Such Wien bridge oscillators usually give their lowest THD and lowest Vo when the NFB ß factor is increased so that oscillations are on the verge of stopping, so that increasing NFB causes stop-start and more NFB makes it stop permanently.
Nobody wants Vo bounce, and everyone wants quick Vo amplitude settlement between F range switching, and in this case, F switching within each F range.
So in this case, I found the balance between low THD of 0.4% at 2.5Vrms output and 4% at 3.7V+ output. At 3.6Vo, I get 1.7% 2H.
It sounds a lot, but its hardly visible on the CRO and it does not alter measurements of amplifiers.
But when the Vo = 3.6Vrms, and THD = 1.7% 2H, -0.81Vdc is the gate bias of fet, abd its Rd measures 302 ohms and the Vd = 0.26Vrms, and the THD at d = 5% approx, mostly 2H. I'm not 100% clear about the properties of the j-fet. We have a voltage at the drain varying +/-0.36Vpk, so current flows positively and negatively, and the Rd in each direction is different.
So I can see that the j-fet has rather non linear drain resistance Rd, and when I get time I will try redesigning the whole NFB loop so there is DC flow in the fet from a second j-fet arranged as a µ-follower. Afaik, bias can be the same as I have it now. The 2k7 plus 1k0 VR1 can be cap coupled to the bottom j-fet drain and then I can see what happens and see if the Rd with Idc and Ed at say +8V is any more linear. The existing 2uF plus 1M0 for control amp Vo DC offset can be removed because of cap coupling of all NFB Rs to the fet drain via a cap.
Elsewhere on the Net you can see j-fets commonly recommended for dynamic ß control used in the FB network to stabilise Vo, but usually, there is **** all mention of the difficulties involved to get good usable operation.
Its also possible to maybe use a voltage clamp of 12 x 1N914 with 6 in series facing each way and with 1k0 trimpot all across my 2k7. The use of 6 such diodes in series means their stray C is minimised and their turn on character is not abrupt, but over a wider V & I range. Then THD should become mainly 3H, and the 3H is not conducted around the PFB loop as much as the 2H.
Its occured to me that anyone could have a single F Wien bridge oscillator with say 3 cascaded wien bridges, each with gain amp of 3x, with two amps having fixed NFB networks. But the last amp has its NFB enclosing the 3 networks with the 2 "buffer" gain amps between the 3 PFB networks. This will increase the Q of the PFB networks and thus any THD of the main amp with variable NFB will be less boosted by the PFB thus allowing NFB to reduce it by about -12dB more than otherwise. Its seems a lot of trouble to go to, but I think that if the PFB network had a higher Q you'd get a lot less THD without needing such a huge amount of open loop amp gain. One is almost tempted to use an LC circuit, or a gyrator used with a C for a parallel resonant circuit.
Keep soldrin', and if you can't stay healthy, try not to go insane.
Patrick Turner.