it pops out that my circuitry seems to be ok. i“ve been trying several different situations and powering methods to make sure it“s not about any outa influence. As none of my modules is doin any strange humm or noise thing i“m pretty sure the module works fine. But sure, i“m not an universal knowledge monster. After many years of mastering and a proper audio education i simply trust my ears.
Since beginning of my journey with Druid-Chips i“m in contact with Tom who“s been very helpful, quick and detailed in every aspect.
itĀ“s simply the 8 bit resolution i hear and some of you havenĀ“t recognized yet ā¦
for my definition this results in a very thin amplitude modulation on top of the frequency range while stepping through the bits, what sounds like a very light sample rate reduction what some of you should know from bitcrushers or low quality MP3s ā¦
iĀ“ve been comparing with another digital envelope from mutable instruments. āPeaksā with a resolution of 16 bits. With diving very deep into modulated sinusoidal waves i can hear the same behavior. Much softer, but noticable ā¦
Tom adviced me to do some more smoothing at the output what iĀ“d like to share with you (referring to my output circuit before), maybe someone else runs into the same conflict like me. IĀ“ve been trying this, but this is the moment when i realised (again), āDonĀ“t waste your time with Lo-Fi digital components when building your own dream synthā
. Maybe iĀ“m still totally wrong, but my ears donĀ“t lie ā¦
This module sucked a lot of time and revisions to reach the final stage. As it“s SMT now i must keep it in this state. In fact it“s Tom“s circuit with simpler CV ins.
I“m having some on stock and it will get its place in my racks, but this forced me to dive into analog envelopes.
I must state iĀ“m ābitā disappointed after all my effort, but itĀ“s not about ranting! Tom is doin a marvelous job and is sharing a wonderful knowledge base.
You can find the module on ecolabaudio.de with some fancy transparent LED output indication and Tom got one as christmas present aswell.
" I would guess that what youāre hearing is caused by the little tiny steps in the digital output waveform. Ordinarily, youād never notice them (youāre right - no-one else has ever mentioned it) but thereās usually some situation in which you can make the imperceptible perceptible. I think you might have found that situation.
Okā¦so assuming thatās right, or close, what can you do about it?
Hiss could probably be reduced by cutting the R32 value from 100K to 10K instead. That then affects the RC roll-off value, so that would need to go up to 470p instead to keep the cutoff the same.
However, if the steps are whatās causing the problem, perhaps we should do more smoothing. How much? Well, what do we get with 10K/470p?
electricdruid.net/rc-filter-calc/?f=&r=10k&r_series=3&r_errors=1&c=470p&c_series=1&c_error=10
This shows that the rolloff kicks in at 33KHz, allowing full audio bandwidth. But weāre not passing audio, and the sample rate is 31.25KHz, so weāre not getting any effective reduction of the sample rate noise by the filtering. The rise time numbers are perhaps more important for us, since we donāt want our shortest attacks to get slowed down by the filter. The transient response is usually regarded as five time constants, and the 5TC column shows that the cap will charge completely in 25usecs - far quicker than we need. Assume our shortest attack is 0.5msec=500usec. So we can increase the cap by a factor of about 20 (500/25=20).
That gives us 470p x 20 = 10n.
So my advice is to change the resistor to 10K, and increase the cap to 10n (or 4n7 if thatās too much) and see what happens. 10n gives a filter freq of 1.6KHz, and 4n7 gives 3.4KHz, so in both cases the digital steps at 31.25KHz should have a lot more attenuation".