Motorboating in leak 20

[bronto64]

Hi, i built my first project, a mod of leak stereo 20, but i have motorboating when start to use it, at 40 - 50 cm of the loudspeaker. No signal input. Where start to find the problem? The 12ax7 valves are used, is this a good point to start.
Thanks, Federico

[SY]

This is almost always a symptom of coupling through the power supply. Make sure that each stage has its own RC decoupling and that it's wired correctly. Do you have the feedback connected with the proper phase?

Additionally, you want the coupling time constants to be well staggered, at least 5:1, better still 10:1.

[ray_moth]

LF instability is always a danger if you have two coupling capacitors inside the global feedback loop: too many phase shifts. This applies also to the Williamson design.

As said already, better filtering of the power supply would help and you should probably do that anyway.

A more complete solution (requiring some design skill on your part) would be to use direct coupling between the input stage and the phase splitter, as in the Mullard 5-10, thereby eliminating one of the phase shift components. You can still retain your 1/2 ECC83 input stage, instead of using the Mullard's EF86, but eliminate the coupling capacitor and change the biasing arrangement of the splitter (see the Mullard 5-10 schematic). This solution will give you an inherently more stable amplifier with a better voice. I must admit that I have never understood Leak's approach to amplifier design.

[bronto64]

Thanks a lot, i go to work whith this tips in mind.
Federico

[andyjevans]

A more complete solution (requiring some design skill on your part) would be to use direct coupling between the input stage and the phase splitter, as in the Mullard 5-10, thereby eliminating one of the phase shift components.>>

Yes, exactly. I built a few like that. Better still use valves like 6N1P, 6N30 (ECC88 pinout) or indeed ECC99 to keep the original pinout. All this needs circuit mods. A better sounding solution still is a DC concertina, e.g. as in Morgan Jones "Valve Amplifiers". I use a DC concertina with zero global NFB, which then allows me to choose lower mu input valves like the 12B4 (splitter) and something like ECC99, 6N30 etc in the first stage - maybe even E80CC or ECC40 for experimenters. I prefer the mono chassis like TL12+, since you can then use the octal preamp socket (which is rarely used) for an octal input valve like the whole 6J5G series (6C5G is rather nice, so is 6P5GT). Sound really starts to get good there. I also use at least one choke per channel (ST20) and preferably two (mono amps). The original power supply is pretty flimsy and does the amps no favours.

[ray_moth]

I agree that the power supply in the Stereo 20 is poor. So is the idea of using a single ECC83 as the input stage for both channels. It would appear that Harry Leak was rather cost-conscious.

The global feedback in Leak amps is quite high. In those days, it was a common belief that the more negative feedback the better. Reducing it should improve the sound and make it more "open". However, this is an ultralinear design so, in the interests of maintaining a reasonable damping factor, I would not be inclined to eliminate the globasl feedback altogether, unless you introduce local shunt feedback between the output tubes and the phase splitter. (Incidentally, doing this would avoid the need to change to direct coupling between the input stage and the splitter.)