SMR4mkII slow attacks with the 1M resistor

When you have the 1M resistor in place, is it normal for slow attacks to stay silent for a moment before kicking in? Also, if I set the sustain level at 1 or 2 the VCA completely closes. 3 is the minimum value it stays open for.

I’m guessing the offset is too much. Am I right in thinking a higher value resistor would decrease the offset and possibly diminish this effect? Obviously, if I’m correct then too much resistance would lead back to the bleed that the resistor was put in to mitigate, but just the right amount…

Is this something worth experimenting with or is the ‘latency’ effect just normal for the design? I can accept it if it’s just how things are, but it bothers me if mine is deficient.

Isn’t there some sort of calibration you can do?

Well the only trimmer on the SMR4mkII is the V/oct adjustment for the filter. What are you talking about? I guess maybe this should be in “Modding and analog circuit hacks”.

You can try without the 1M… Maybe your one doesn’t have a leak. If there still is a leak you could try with a 2M.

I don’t think it’s totally due to the 1M resistor.

It’s also because the VCA CV has an 8-bit resolution, and the amplitude is scaled by the velocity - so it cannot produce arbitrarily low levels - there’s always a point at which it’ll jump from nothing to something.

That does explain the point when it kicks in, but surely it should jump from nothing to something between 0 and 1 rather than between 2 and 3?

The VCA completely closes at sustain levels 1 and 2 as well as 0. The jump is between 2 and 3 rather than 0 and 1 so with a long attack time the VCA remains completely closed until the envelope reaches whatever voltage corresponds to sustain level 3.

Would increasing the resistance reduce the offset?

The VCA envelope is scaled by velocity. So a sustain value of 4 is truly a sustain value of 2 when the velocity is at half the maximum value.

> Would increasing the resistance reduce the offset?

Yes.

I thought it was only scaled by velocity if you had Vel -> VCA in the mod matrix? Or am I misunderstanding something?

But by default there’s such a routing, right?

Oh yeah, but that was disabled on the patch I was testing this with. I’m not that stupid.

I did not know that you had disabled velocity.

It just made sense for the kind of thing I was testing for… I disabled all modulation other than env2 -> VCA so I knew exactly what I was testing. I didn’t expect you to know I’d done that, I just meant that these things did occur to me. I certainly didn’t mean to be rude.
Is it possible for you make an educated guess as to what sort of range of increase in the resistance I’ll need? For example, would 1.1M overshoot, undershoot or be about right?

We’re talking about a parameter which is not specified by the manufacturer of the chip, so there’s no “right” value.

Now I remember something… A long time ago (2011?) I noticed that the PWM output produced erratic pulses patterns when set to a duty cycle of 1 (out of 255) - and this produced a very noisy sound. Only the VCA output pin was concerned. So there’s code to set this output to 0 whenever it should be 1 to avoid this condition. I’ll try to reproduce the problem, if not, I’ll remove the code. This defect could have been present in only one batch of chips, or maybe it was some misconfiguration of the PWM hardware registers in an earlier version of the code.

I had a look at this…

1/ I could not reproduce the problem with the PWM VCA output that caused me to shut down the envelope as soon as it reached a value of 1. So I will post soon a firmware update file which fixes this. I really can’t recall why this fix had been there in the first place.

2/ I checked on the three Shruthi I have here, and there’s no consistent “right” value for the resistor since it depends on a parameter (the voltage for the logic LOW level of the VCA PWM pin) which is not specified on the datasheet. 1M is a “catch all” value. To get the correct resistor value, do the following:

• Wait until the envelope has decayed to 0.
• Measure the voltage Vnull on the “VCA” pin of the control board.
• Measure the voltage Vee delivered by the -5V regulator (LM7905).
• Compute the resistor value using the following formula:

18000 x Vee / Vnull.

For example, I measure 48mV as Vnull and 4.98V as Vee. The right resistor value is 1.87 M.

I have two Shruthis for which the value is in the 50mV range, and another for which the value is in the 10mV range (requiring a resistor 5 times bigger).

Ah ok, I was just hoping that you might be able to guess based on however you found the value of 1M originally.

I only started this thread to find out if something was wrong with my Shruthi, if this is how they all are then it’s no problem.

EDIT: Thanks for the formula! That’s exactly what I was hoping to find out.