Don Pearce wrote:
On Mon, 30 Mar 2009 16:59:11 +0100, "TonyL"
wrote:

Since my last post I tried reducing the HF phase shift by simply
inserting a single 33 ohm resistor directly in series with the LM388
output. This amounts to the same solution that Don offered...that of
swamping the reactance with resistance. This *seems* to have fixed
the problem but I need to be sure before this load of kit gets
shipped half way round the world. Last thing I want is a midnight
phone call saying "This *&^% thing you sent doesn't work".

Just to clarify, mine doesn't swamp the reactance with resistance, it
actually removes the reactance totally - the load is purely resistive.


Assuming terminating resistor=cable characteristic impedance, yes.

But for various practical reasons I'm using twin-core twin-screen stereo
cable with an undisclosed characterisic impedance and a capacitance of
around 100pf/meter. My fault I didn't mention this....sorry.

If you are driving many loads, what about configuring it as a proper
distribution amp. Use a separate 50 ohm resistor to feed each cable,
and load the far end of each with 50 ohms. Everything is properly
loaded for flat response, nicely buffered from everything else, and
guaranteed stable. You get a net 6dB loss, which you can make up
simply in the feedback network.


If I were starting this project from scratch I well might do that.

I'm fairly happy with the stability now. I've tried adding large amounts of
capacitance at various places in the cable setup and the system remains
stable, the only effect of any added C being increased HF attenuation.