External VCAs

[jjonas]

Hi,

 

now that I've gotten external filters and AOUT_LC working, I was thinking of trying my luck with external VCAs "to overcome the infamous ADSR bug of the SID" :-)  I was wondering whether anyone else uses external VCAs, and if so, which one? I'd be looking for relatively simple DIY. Web search produces several options, but I'm not sure which ones would be suitable as there's often a lot of tech speak I don't understand for sure.

 

If someone is using a particular DIY VCA for which there is a schematic on the net, I'd be interested. Otherwise, what do you think of e.g. the following options:

 

http://yusynth.net/Modular/EN/VCA/index.html ("Simple VCA")

 

http://hem.bredband.net/bersyn/VCA/vca_shootout.htm (several)

 

http://www.sdiy.org/philgallo/mgbvca.html (SSM2164 Based Voltage Controlled Amplifier)

 

 

[jrp]

Using a vca is a very good idea!

You want a log responce VCA in combination with the MB_SIDs envelopes.

All of the VCAs i tried required a small cap on the cv line to tame the initial click.

 

I tried this LM13700 design http://www.musicfromouterspace.com/analogsynth_new/DUALVCA/DLLVCA001.html 

Note that both sides should be log! Several versions of these on the net.

It sounds good and gives a nice decay!

 

SSM2164 is also an very good candidate. I settled on using these as there are 4 vcas in a chip, easy to use. If you invert the aout-channel in the ensamble menu just a trimmer is needed to interface to this vca. Also good sounding.

 

The VCA included in my CEM3379 filters was not working that nice due to it´s uncommon cv responce.

From 0-200mv it´s log - crowding the whole range of -100db to -20db into this little cv.

the ramaining 20db get controlled linear with cv from 200mv to 5v.

The effect of this: it sounds less dynamic. Decay and release is much longer and somehow squashed. Still it´s useful for some other tasks around my sid(...)...

Edited May 12, 2014 by jrp

[jjonas]

Thanks for your reply!

 

The MFOS is good for having a schematic and PCB layout available (I'll draw my own PCB in Eagle), though the SSM2164 would be more footprint-effective with four VCAs in one chip. Do you have a schematic for the solution you are using?

 

At the moment this schematic (pdf) for a SSM2164 VCA looks like one I can read, any opinions? For each channel's CV it mixes two CV inputs (1A+1B = CV1), I guess the other branch (say B) can be left out, plus the four TL082s can be replaced with two TL084s..?

 

A few more questions. I found out while googling that seppoman had designed an SSM2164 VCA as well, though he never got around to posting the schematic or PCB here. Anyway, that with an external VCA "you just set the oscillators to A=0 D=0 S=15 R=0 and control volume through the SID engine's envelope" to which TK adds that one must activate the GSA flag in the OSC menu as well. Does this mean that with external VCAs one has to manually change ADR=0 and S=15 with every patch, separately? I.e. the SID envelope is not overridden by ensemble setting V2A, the envelope information just gets sent to AOUT in addition?

[jrp]

in fact the osc-envelopes are independant from the main volume that is forwardet to aout.

So yes and no. If you activate GSA the oscs will always be "on", playing at their sustain value. So in fact if A=0 D=0 you can set the osc-volume with S=1 to 15.

It is also possible to use without GSA, then you should just adjust the release so that the osc wont be turned off before your vca modulation has fadet out the note.

 

I built my ssm vca on veroboard and never bothered to draw a detailed schematic.

I would recoment having a look into the datasheet, in the meentime i try a little drawing... I am working on a complete analog channel with some effects right now, but that´ll take some time...

 

The schematic you linked will work, but it´s more than you need.

Since the cv(aout) can be inverted in the ensamble you only need a trimmer to scale it down when driving the vca directly.

 

Note that complete Aout module is not needed. I drive mine directly with the DAC connected to the core.

 

 

i attached a little drawing.

Edited May 14, 2014 by jrp

[jjonas]

I did take a look at the datasheet, but the shematic there was not explicit enough for me: it puts the text "power supply and biasing circuitry" where I'd need the circuitry to be :-)  So that's why I tried looking for a schematic that had everything in it, without as much assumed knowledge as in the datasheet schematic.

 

I'll try to draw a schematic this week based on the one I posted, the picture you drew and the SSM2164 datasheet, and post it here for you and others to evaluate, is that ok..?

[jjonas]

Actually the text "power supply and biasing circuitry" appears to refer circuitry inside the chip and not outside – I was too hasty with the datasheet, in fact it seems to offer a good application schematic (page 8).

 

I drew the following schematic (below) for one channel, do you think it's ok? Mostly it's the datasheet schematic, but I combined it with a few caps from the other schematic (pdf) that I referred to, and didn't include a trimmer for adjusting CV as the setup I'm planning includes an AOUT_LC board, and the CV is set with the trimmers there.

 

I left MODE pin open because it was open on "the other schematic (pdf)"; according to the datasheet leaving it open means a choice in favour of "Class AB" and against "Class A", which in turn means

 

"Class AB operation refers to running a VCA with less current in the gain core, resulting in lower noise but higher distortion. More current in the core corresponds to Class A performance with its lower THD [total harmonic distorition] but higher noise.Class AB operation refers to running a VCA with less current in the gain core, resulting in lower noise but higher distortion. More current in the core corresponds to Class A performance with its lower THD but higher noise."

 

I don't really know what this means in practice, do you think there's an obvious choice? One could always build an option into the circuit, so that MODE pin can be connected through a 7,5k resistor to +12VDC with a jumper. Or I just try it out when breadboarding the circuit. But I'm still interested in any opinions on this!

Edited May 15, 2014 by jjonas

[jrp]

Your schematic looks fine, although you have to consider that the ssm2164 will be at unity gain with 0v and attenuate with a cv going positive. around 3v should be enough to completly shut off the signal. So your cv needs to be inverted (easiest way is by inverting it in the ensamble settings) and scaled down. I don´t know if the trimmer on the Aout module is sufficient in it´s range for this (i am driving my vcas with the dac directly, TLV5630 outputs around 5v and can drive 2k).

If you cannot scale the cv down to the needed range with the trimmer you need to add an aditional voltage divider in your cv line. Like the trimmer in my drawing. Impedance of the divider can be anything from 10k to 100k, the VCAs are truly voltage controlled so impedance doesn´t matter. You could try this :

 

Aout --- 10k --- SSM2164 cv in

                      |

                   10k

                      |

                  ground

 

this will reduce the cv from the aout to one half of it´s original range. Then adjust the trimmer with cv at full level (volume=0 when aout channel is inverted) until the sound is completly vanished.

If your cv range is too high the notes will be delayed as the envelope will attack through too much silence in the beginning.

If the cv range is too small you will still hear sound even though the note should be turned off.

 

I left the mode pin open. For a synth neither distortion nor noise in the order to be expected will be of any concern. I doubt that you´ll hear any difference.

Edited May 16, 2014 by jrp

[sneakthief]

A couple of related thoughts:

 

- The SSM2164 is discontinued, however Coolaudio now makes an identical version called the V2164.

 

- The SSM2164 is very easy to destroy with improper voltage input if you don't take the proper precautions:

 

The SSM2164 exhibits a catastrophic failure mode when the V+ pin is powered and the V- pin disconnected

 

http://electronic-sea.net/SSM2164.html

[jjonas]

Hi,

 

thanks for replies! I haven't tried yet whether AOUT_LCs trimmers are enough to do the trick, but I'll do that before I finalise my VCA PCB, and include a voltage divider if necessary.

 

Based on sneakthief's reply I made a second draft of the schematic, with an additional diode in case of a "catastrophic mode", also and for reverse voltage protection. I'm not sure but I think 1N4148s will do for reverse voltage protection, their V_R is max 75V and continuous forward current 200mA, aren't these the relevant numbers..? Plus I hope I put the polarity protection diodes the right way in the schematic :-)

 

-12VDC line name (at the bottom) looks confusingly like it's positive, but in fact it's just smashed (previously it wasn't).

[jrp]

in fact the 2164 is really one of the easiest destroiable chips i know. Next in line from my experiance is LM13700 and SSM 2018...

You can wreck it by simply touching it if you are carrying static charge. And most likley you are... It is good advice to ground yourself (eg by touching a grounded metal case of some equipment) before handling the chip. Don´t forget that this also counts for components that are connected on the pcb. I killed more than one of these chips simply by sticking a resistor into the breadboard while experimenting.

Since i do the grounding trick before touching a part like this i never had any issues.

 

There are a couple of sellers on ebay who still offer the ssm2164 or the coolaudio replacement.

[jjonas]

Hi,

 

I'm back on this topic since now I'm ready to start practical VCA tests with the MBSIDv2. However there are a few things I'd like to know before I start experimenting, just to make sure, and would like to ask people's opinion. (Here's the SSM2164 datasheet for reference.)

 

1. Am I looking only for attenuation?

 

2. According to the datasheet, maximum supply voltage is +-18VDC (p. 3); in my setup it's +-12VDC. Control voltage range is given as V+ and V- (p. 3), am I right to think this means that in a setup that's only for attenuation, the CV range is from 0 to 12VDC?

 

3. According to the datasheet,

 

"Notice that a capacitor is connected from the control input to ground.  Because the control port is connected directly to the gain core transistors, any noise on the VC pin will increase the output noise of the VCA.  Filtering the control voltage ensures that a minimal amount of noise is introduced into the VCA, allowing its full performance to be realized.  In general, the largest possible capacitor value should be used to set the filter at a low cutoff frequency.  The main exception to this is in dynamic processing applications, where faster attack or decay times may be needed." (p. 8, emphasis added.)

 

I have used 1uF caps, like in the example configuration on p. 8. It sounds like the MBSID VCA case would be one where relatively "faster attack or decay times may be needed", but I don't really know whether 1uF is "too big" for this purpose or not. Maybe it can only be found out with experimenting, but if anyone has opinions on this, I'd be interested.

[jrp]

1 and 2. This is depending on your design targets. As far as i remember the SSM2164 can give you 20db of gain with negative cv.

The gain constant is -33mv/dB.

It is the easiest to use a cv in the range of 0v to about 3,3v for a -100db to 0db range.

That´s the voltasge on the controll pin. If your CV is ranging to 12v (not really possible if your supply is +-12v) it needs to be divided by 4 to give a usefull range.

It is common practice in modular synth, but also when designing a sid channel, to settle for a given cv range and use trimmers on circuit blocks to adapt for those cv ranges.

If you look at my other post about my analog channel you can see a schematic that works with an inverted cv (ensemble setting). A trimmer is used to set the maximum attenuation by ear. -80db is propably sufficient.

 

3. The sice of the capacitor is depending on the impedance of the cv signal. Note that the trimmer will already present a series resistance, forming a lowpass together with the cap. This is exactly what we want here.

If you imagine a fast raising spike on your cv, it will be smoothed out by the lowpass.

I adjusted the trimmer for max attenuation first, then selected a cap that gives enough smoothing to eliminate any clicks, but still enables very fast attack times.

Make the cap too big and your notes will fade in slowly although attack is set to minimum.

 

The pf caps on the signal input and the buffering amp are necessary to prevent oscillation. Very important i had to notice!

[jjonas]

Hi,

 

thanks for your reply!

 

I have three chained AOUT_LC boards, two first for cutoff & resonance (channels 1-4, they're working), and third for VCA (5-6). (Channels are set in setup_*.asm.) So far I haven't stuffed the VCA board with chips, I've just checked the CVs first. So far I haven't been able to change the CVs on the third board from the control surface. (I have enabled V2A in the ensemble menu for the core in question). But at this point my question is about this section in the setup_*.asm file:

 

    ;; only relevant if one or more AOUT_LC modules are used:
    ;; define the resolution configuration here
    ;;   0: first channel 12bit, second channel 4bit
    ;;   1: first channel 8bit, second channel 8bit
    ;;   2: combines M1,M2 and/or M3/M4: first channel 12bit, second channel 12bit, third channel 8bit, fourth channel ignored!
    ;; all other values invalid!
#define AOUT_LC_RESOLUTION_OPTION_M1 0
#define AOUT_LC_RESOLUTION_OPTION_M2 0
#define AOUT_LC_RESOLUTION_OPTION_M3 1
#define AOUT_LC_RESOLUTION_OPTION_M4 0

 

I'm not sure what 'Mx' refers to here, but I take it M1-M4 refer to the four AOUT_LC boards that can be chained. The first two boards (for cutoff/reso) are set with jumpers to 12bit/4bit, so the setting is here the same. However for VCA it seems better to make both channels the same, so I've set the jumpers for 8bit/8bit, and figured that I should set the setup_*.asm accordingly. Is this correct?

 

BTW the AOUT_LC module info page says (bottom of page) that

 

With two chained MBHP_AOUT_LC modules, you can control CutOff and Resonance of the right channel SID the same way. With V2A, volume can be handled a similar way (4 chained AOUT_LC modules required for this case).

 

Is this correct, i.e. that you need two boards for VCA? Or have I misunderstood something?

[TK.]

Your configuration is correct, in this case only a single AOUT_LC is required to control two VCAs

 

Best Regards, Thorsten.