[Fpga-synth] Envelope - linear attack & expo decay/release?

[Scott Nordlund]

I've checked a couple datasheets for analog envelope generator chips (CEM 3310, 3312 and SSM 2056).  For the attack stage they seem to use a sort of "truncated rising exponential decay", where an RC circuit charges to about 77% of the charging voltage before the circuit switches to the decay stage.  I'm not sure how this would compare sonically to a linear attack.  It might be best to play around with some software-generated envelope curves before putting much work into it.

Anyway, if you wanted to implement something more than an ADSR, this might be a nice way to do it, since no additional logic is required for rising vs. falling stages.

----------------------------------------
> Date: Thu, 2 Apr 2009 14:16:39 -0700
> From: ebrombaugh1 at cox.net
> To: fpga-synth at rubidium.dyndns.org
> Subject: Re: [Fpga-synth] Envelope - linear attack & expo decay/release?
>
> Scott Nordlund wrote:
>> I know I was always dissatisfied with the envelope curves on the DX7. I'm not sure but I would assume a linear envelope was run through a exponential look-up table to generate an acceptable curve. It sounded fine for decays, but long attacks (rising envelope segments in general) seemed to delay for a while and then suddenly go up to full amplitude, forcing me to use multiple envelope segments to get an acceptable attack.
>
> I've had zero exposure to the DX-series synths, but I could see that if
> they were using an expo LUT for the attack that worked the way the
> release does that it would have the effect you describe. Annoying.
>
>> If you look at analog envelope circuits, those with an exponential attack still use an exponential RC decay (I mean the curve here, not the envelope stage, with the rate of change slowing as it approaches the destination value), it's just that it goes up to a higher voltage rather than down to 0. I believe some also implement a linear attack with an integrator (I've never compared them to see which I prefer). The DX7 method presumably results in exponential growth.
>
> Yes - implementing an RC-like decay and release in digits shouldn't be
> too hard, and a linear attack would probably sound pretty good in
> combination. I could see a system where one would be able to select
> linear or RC for each segment separately.
>
>> So you may want to experiment with linear attack/exponential decay
>
> That's looking like the plan.
>
>> It's worth considering also (though it adds further confusion) that an analog VCA may have an exponential response...
>
> Seems like an expo VCA combined with a linear ADSR would sound a lot
> like the DX7 issue you mentioned.
>
> Eric
> _______________________________________________
> Fpga-synth mailing list
> Fpga-synth at rubidium.dyndns.org
> http://rubidium.dyndns.org/cgi-bin/mailman/listinfo/fpga-synth

_________________________________________________________________
Rediscover Hotmail®: Now available on your iPhone or BlackBerry
http://windowslive.com/RediscoverHotmail?ocid=TXT_TAGLM_WL_HM_Rediscover_Mobile1_042009