karatestu wrote: ↑Sun Jun 27, 2021 7:23 am
I haven't bought a stepped attenuator yet, been too busy faffing about with speakers and power amps
As to introducing a buffer is concerned- I would only do it as a last resort . Adding components goes against my less is more philosophy (which I borrowed from RD).
If I ever did use one it would have to be as simple as possible and unity gain as you suggest Rex
Yeah, most of the time less is more, but occasionally more is more. Here is an abbreviated version of my voyage.
I 'discovered' high efficiency drivers. Ones that have obvious frequency response issues yet make most others sound quite sleepy.
The main issue with these drivers is that they make everything MUCH louder, including things we don't want, like noise, hum and switching noise. So i embarked on a journey to minimise this. Now NVA amplifiers are pretty quiet in most respects, but that damn volume control (direct coupled cermet) caused me no end of headaches. It was fine with normal speakers, drove me nuts with these ones. I'd turn off the amp and vigorously wind the knob up and down to make the things temporarily quiet, but within weeks it was noisy again and the simple act of altering the volume caused massive cone movement. WTF? It sounds amazing as long as i leave the volume alone?
Anyway, i noticed that there was a rather large amount of DC present on the outputs (I was using an AP70 at the time), so investigated. There isn't anything to adjust on these amplifiers, so to cut a long story short, it ended up being caused by unequal currents through the differential input pair. The base of these transistors needs have the same DC operating conditions at all times. Differences in the AC current are reflected in the output and that's the way it's meant to work, but differences in the DC current are also presented to the output. OK so what is causing this difference? A different resistance to ground it what finally dawned on me. The input bias current is flowing back out through the base resistor, into the volume control wiper and down to ground. THIS is what causes the volume control to become noisy and whenever you adjust the wiper position, you are adjust the DC operating conditions of HALF of the input pair. Result, DC offsets all over the map. Not enough to worry a normal speaker, but enough to cause craziness in mine.
I ended up putting a wima MPP cap after the pot's wiper, which immediately solved the issue and i couldn't hear any issues with that (which was a relief, as i hate the sound of electrolytic caps). I stayed like that for a while, then i replaced the cermet pot with a switched attenuator, which is much better, but i'm not all that taken with only having 24 or so different volume steps. I like to keep the maximum volume presented to me under strict control, too loud and my ears distort the sound, too soft and i feel like i'm missing out. Anyway...
I moved onto an active crossover (begone foul passive crossovers), and so ended up using a separate box for volume and switching, now driving 3 power amps, so 1/3 the AC resistance, 3 times the load. Now i could just adjust the input impedance of all these amplifiers, necessitating much phaffing about with the DC operating conditions of the input pair, or i could just buffer the output of the volume control, which would also have the effect of presenting a nice DC bias on the output capacitor. Much easier to buffer the output. I eventually hit on a buffer composed of 4 of the same transistors, 3 of which are linearizing the other one. It's zero feedback and only one stage long. It sounded better when i installed it and when i tried to measure it, distortion and noise when down compared to loopback conditions (which just means that what i was measuring it with could be better).
As always, YMMV.