Mike Keitz <PICLIST@MITVMA.MIT.EDU> wrote:

> It appears that two major complications crop up in practical
> application:
>
> Matching two attenuators for stereo use.

    Or FOUR attenuators, if you're building a balanced preamp.

> Preventing transisent improper attenuations in "binary tree of
> relays" designs.
>
> First on matching two attenuators.  For designs with lots of steps
> such as pots turned by motors or digital pots with extra range,
> automatic matching could be implemented.  Audiophiles hate the
> notions of feedback and automation.  But it could work really well
> here without degrading sound quality.
>
> The general principle is that after adjusting the volume, inject a
> test signal into the input of both attenuators and compare the
> levels of the test signal after the output.  Then adjust the
> attenuators until they are matched.  Once balance is achieved, the
> test source and analyzer would be disconnected by relays so that
> full fidelity is restored until the next volume adjustment.

    That's a good idea, but -- at least in our implementation
    -- it had a few problems:

        1.  It's hard to "adjust the attenuators until they are
            matched" if they have fixed one-dB (or so) steps.
            Adding a set of fractional-dB "adjustment" dividers might
            solve this problem in some circuits.

        2.  If the left-to-right balance is implemented by
            deliberately mismatching the left and right attenuators
            (as opposed to having a separate balance control further
            downstream in the circuit), your analyzer would have to
            compensate somehow.

> It seems like the simplest design would be to inject an in-phase
> signal to one channel and a 180 degree out of phase signal to the
> other.  At the attenuator output, a summing network would produce
> zero if the attenuators are balanced.
> ....
> The summing node could be deliberately mis-matched by the user to
> cause the autobalance network to put a deliberate unbalance in the
> attenuators (the "balance" control).

    That solves problem #2.

> Now for the other problem of transients in relay attenuators.
> Andrew said they sound great, other than when changing the volume.
> It seems like the simplest solution would be to place an inexpensive
> but smooth-changing parallel "bypass" attenuator in the signal path
> while the relays are changing.  The bypass attenuator would maintain
> smooth changes in level, though at lower fidelity, while the user is
> turning the volume control.  After the volume is adjusted, the relay
> attenuator would be set for the corresponding level and switched
> back in.
> ....
> The autocalibration technique described above could be used to match
> the bypass attenuators to the relay attenuators, giving nearly
> seamless changing from one to the other.

    We thought about switching-in a digital pot while the volume knob
    was being turned, but -- mostly because, as you mentioned,
    "audiophiles hate the notions of feedback and automation" -- we
    never considered automatically calibrating the two attenuators.

    Therefore, we rejected the idea on the grounds that there'd be
    an "unavoidable" volume change whenever we switched from one
    attenuator to the other.  Even though there'd still be SOME
    variation in volume levels with your solution (due to the fact
    that both attenuators move in discrete, non-continuous steps),
    that variation might not be unpleasant.

    It's a really good idea... I hope Jon Nicoll's paying attention.

    -Andy

=== Andrew Warren - fastfwd@ix.netcom.com
=== Fast Forward Engineering - Vista, California
=== http://www.geocities.com/SiliconValley/2499 (personal)
=== http://www.netcom.com/~fastfwd (business)