Analog Attenuator?

acg · April 12, 2020

I feel that I am in a position to comment on much of what has been said in this thread because I have experience with most of the stuff that people have mentioned. I have one of @PeterStPhasure NOS1a dacs with its digital attenuation in computer software. I have a Placette Active Preamp that Peter has also had in his own hands for evaluation, but not the passive as used by @bbosler, although it uses the same volume control. I have tried and am still in the process of trying to find a pre-amp that will suit my system and my needs and have built and purchased and assessed a number of solid state preamps with analogue attenuation and I can only conclude that Peter is correct in insisting that the least impact attentuation is done in software through his NOS1a dac...no doubts about it.

But that is not the end of the story.

My system includes a very nice turntable with two arms and two phonostages, the NOS1a dac, a computer streaming input, FM radio...plus at least one spare input just in case another source is added down the track (probably Bluray/DVD input). That is six inputs. Because I do not graze only in the computer audio paddock the software based digital attenuation is only useful for one-sixth of my sources, effectively rendering it moot. Hence my need for a preamplifier.

My amps are a little bit unusual in that they require 4V+ input to clip, and they have the passive line level crossovers for six-ways at the input to the amplifiers, so it is an unusually complex load for whatever is driving them. Not to mention that those 6 channels parallel to a circa 5K load...which is low. So whatever is driving my amps needs more voltage than just about all modern sources output, which means my preamp needs gain, and very low output impedance. Surely SS gain stages will work? Sort of, not really, some much better than others, none have been ideal. Along the way I've been able to try various analogue attenuation options including the switched stepped attenuators (Goldpoint, Khozmo etc.), a couple of implementations of the Muses 72320 chip (which have been good, certainly better than the stepped attenuators in my circumstances) and shortly a simple circuit designed by Bruno Putzey that varies the feedback in a differential opamp gain stage to alter gain and hence volume. Also on my desk for implementation are some Intact Audio autoformers which based on my experience in other systems will probably be the ultimate solution, as @JohnSwensonand @Superdad have previously testified in this thread.

The beauty of the autoformers is that the good ones can handle more than 8V at 20Hz so are able to be used between the source/preamp and amplifier. They have this wonderful knack in this position of lowering the source impedance as seen by the amplifier as soon as any attenuation is used. So at -6dB attenuation the source impedance as seen by the amplifier is improved 16 fold, and it increases logarithmically from there reducing to pretty much zero by no later than -10dB attenuation. Compare that to a stepped attenuator (Goldpoint, Khozmo etc.) used in that position, the impedance seen by the amplifier is increased as you start attenuating and after a while starts to fall again. What does this all really mean? Well, it means that with an autoformer volume control, you do not have to limit yourself to gain stages with low output impedance (read: high levels of global negative feedback) if you do not wish to...you have more options in selecting the active linestage/preamplifier circuit to craft the sound you want. It also means that if you are not caught in the "race to the bottom" of THD and other specmanship, and cannot hear the difference between 0.001%THD and 0.01%, that you have options.

So, to summarise my experience, Peters software VC is sonically the best option if your only source is computer audio. Most preamps I have added to Peters NOS1a dac have diminished the dynamics a little, but there is one Class A circuit that uses the Muses 72320 chip that I could live with at this stage (not perfect, but satisfying). Yet to check in my system are the variable differential gain stage from BP and the Slagle Autoformer (though I do have prior experience with the autoformers in other systems). I love the simplicity of the option from Dave Slagle, and due to the nature of autoformers, the flexibility in selecting a gain stage before it.

acg · April 12, 2020

10 minutes ago, sandyk said:

Technical specifications such as Frequency response and Distortion etc. ?

Not specified as far as I am aware Alex. I did see someone that did publish some FFT's elsewhere and here is a sample...

2Vrms input, 0dB attenuation...

Frequency response -18dB...

Note the scale, 1.5dB down at 10Hz. Looks like generally less than +/- 0.1dB. I expect the line runs straight far past 20kHz.

acg · April 12, 2020

Actually, I should have looked further into that guys measurements. The graphs in my previous post were with a smaller lamination stack. With the full lam stack, thus giving the full inductance in the units as shipped, the graphs are as follows:

-18dB input (-3dB = 2Vrms), -6dB attenuation