[sdiy] VSM201 Vocorder Question

[Tom Wiltshire]

Interesting idea. I've never seen it done.

Aside from the extra filter in-between the stages, there's nothing particularly onerous about it. Given that it means you can use higher PWM frequencies, you could perhaps reduce the strength/sharpness of the filters equivalently and get a similar effect. So although the filters multiply, there are some gains to be had. 

In fact, given that switches are cheaper and easier than really fast PWM, it makes me wonder if you wouldn't be better off combining a couple of stages of 7-bit PWM rather than trying to go straight to 14-bit PWM for your -80dB VCA. (Sorry to keep talking about PWM in terms of bit depths, but it's how I think of it. Digital counters are a simple way to generate PWM waveforms. It converts easily to percentages if that works better for you. )

Still, it's going to be several chips (let's say a 4066, and a TL074 for the filters) to build a simple VCA that you could do with half a LM13700. It's still not efficient, unfortunately. But it might make -80dB possible by this method, which is an achievement nonetheless.

Tom


On 2 Jan 2016, at 22:29, Mattias Rickardsson <mr at analogue.org> wrote:

> Why not use 2 analog switches in series and square the attenuation
> (double the dB numbers)?
> Some extra filtering needed in between them, but -40 dB PWM-VCAs could
> be doubled to reach -80 dB or tripled to get -120. :-)
> 
> /mr
> 
> On 2 January 2016 at 23:10, Tom Wiltshire <tom at electricdruid.net> wrote:
>> Ok, maybe -80dB is *really really* impossible! My judgement was "certainly tough, but probably not impossible", but I'm very willing to go with "extremely difficult bordering impossible" or "totally impossible" if you feel that's a more accurate evaluation! Certainly a 50KHz carrier isn't that high, so I'd probably want to cut your timings in half (e.g. 100KHz carrier), which makes it even more difficult. Certainly we need significantly sub-nanosecond switching times to get good results.
>> 
>> The point was just that -80dB isn't that good for a modern VCA, and it's pretty much out of reach for PWM VCAs. Lots of Blackmer-cell-based designs do much better than that, and even 13600's can probably pull -80dB out of the hat ( http://hem.bredband.net/bersyn/VCA/vca_shootout.htm )
>> 
>> Tom
>> 
>> 
>> On 2 Jan 2016, at 21:12, Richie Burnett <rburnett at richieburnett.co.uk> wrote:
>> 
>>> With carrier frequency of 50kHz, 0.01% duty ratio (for 80dB attenuation) represents a pulse width of just 2ns! That's getting near the sort of time mismatch you can get in turn-on and turn-off times for the switches. So the switch might not turn on at all, or might stay on twice as long!
>>> 
>>> -Richie,
>>> 
>>> Sent from my Xperia SP on O2
>>> 
>>> ---- Tom Wiltshire wrote ----
>>> 
>>>> +1 totally agree
>>>> 
>>>> If you can produce a 1% pulse width, you still only get -40dB. You need to get a 0.01% pulse wave to get -80dB. Tough to do. Not impossible, but awkward enough to make it stop seeming like such a great solution.
>>>> 
>>>> I've thought about this a bit because of using the PIC's PWM module so much. The best case output from that is either 8-bit or 10-bit, which means that -60dB is about as good as I'd get using it for a VCA, and that implies having a switching frequency which is much too low (31KHz) for many jobs.
>>>> 
>>>> Tom
>>>> 
>>>> 
>>>> On 2 Jan 2016, at 19:42, rburnett at richieburnett.co.uk wrote:
>>>> 
>>>>> <snip>
>>>>> Control range of PWM'd CMOS switches acting as VCAs isn't that great though.
>>>>> 
>>>>> -Richie,
>>>>> 
>>>> 
>> 
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at dropmix.xs4all.nl
>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy