[sdiy] Moog 904A clone CV inputs gain confusion...

[Jean-Pierre Desrochers]

Very detailed answer for my questions Terry !

After tweaking some part values and made some tests on my Dotcom and 
Moog clone modules
I had to make compromises and I decided the final and best solution for 
my system.
Since everything in the Dotcom system goes  0-5v (ADSR for instance)  
and +/- 5v (like for LFO, S/H modulation)
and the fact Moog 904A,B,C modules are the only different 'guests',
for them I'll still use Moog's 100k resistors for the 1v/oct CV inputs
(CV input summer gain not touched, original for keyboard CV resonance 
tracking)
and 43k resistors (instead of 100k) for their 'modulation' CV inputs (a 
swing of 0-5v from my Dotcom module will spread all the filter 
spectrum).
+/-5v modulation swings CV's will be attenuated by my  CV summer modules 
for best modulations before they get
To the 904's ..
So far I can now really hear these VCF's at their best!

Thanks !

J-Pierre

**************************************************************************

> -----Message d'origine-----
> De : Terry [mailto:daytona at verizon.net]
> Envoyé : 7 août 2014 16:36
> À : Jean-Pierre Desrochers; synth-diy at dropmix.xs4all.nl
> Objet : Re: [sdiy] Moog 904A clone CV inputs gain confusion... test 2..
> 
> Easiest solution would be to perhaps change a resistor on the Dotcom 
> envelopes to match the output on the original Moog 911 envelope 
> generators. That Moog spec for the 911 output was >taken from a sheet 
> printed in October 1965. (I have the original) They did not print the 
> part that reads "Specifications subject to change". All the 911's here 
> are circa 1969, and all >three output at least 6.3 volts. Moog made 
> many revisions since 1965. 0 to 6 volts is the magic Moog number 
> controlling VCO, VCF. VCA.
>      The Dotcom Q109 envelops here, two of them, put out a maximum 4.9 
> volts.
> Thats 1.4 volts less than the 911's and in filter response almost an 
> octave and a half less. When I patch in the Q109 to the Moog 904A they 
> do not open the filter as much as the 911's >do. Same with the 902 VCA, 
> the
> Q109 envelopes will not open the VCA as much as the Moog 911. Easy to 
> figure out why when measuring the actual voltage outputs.
>      Looking at the 911 schematics, resistor R27 is "Selected" for Esus 
> level.Nowhere on the schematic is that voltage spec listed. Mine are 
> all set to roughly 6.3 volts. Two are >original with the system and a 
> third circa 1969 was added. Can ask Roger at Dotcom if there is a 
> similar resistor on the
> Q109 to make them Moog compatable.
> 
> The "Fixed Control Voltage" on the 904A is just that, and front panel 
> calibrations are accurate. Regardless of the actual pot sending ground 
> to
> -6 volts through a 51K resistor, the effect is the same as sending -6 
> volts CCW to +6 volts CW through the 100K CV inputs. This can be easily 
> be verified by observing the response by >patching in the same 
> voltages. The cutoff frequency is the same when the dial is set to zero 
> as it would be sending 3 volts and offsetting this by turining the 
> fixed control voltage to -3 >volts. This is a voltage adder that adds 
> the three front panel inputs the input node and the fixed control 
> voltage. So yes, adding in -6 volt offset from the fixed voltage, one 
> would now >need 12 volts at the CV input to open the filter. There are 
> many reasons why in an open patching system these negative offsets 
> would be needed. Depending on if using a keyboard to control >the 
> filter tracking the CV at middle "C" could be 3 volts. This 3 volts 
> added to the 6.3 volt 911 output may set the filter higher than needed. 
> So a negative offset would be dialed in via >the fixed control voltage 
> to compensate. In a typical patch one might have both an envelope and 
> keyboard tracking CV all added together and a dial with a negative 
> offset is handy to bring >all the voltage inputs in range. Also not 
> uncommon in a modular systen to send two 911 inputs for a more complex 
> envelope or other modulation sources that all add together where a 
> >negative offset is needed.
> 
>      The Moog 901 oscillators are the same with the fixed control 
> voltage used as an offset to operate the 901's in the specified voltage 
> range. The
> 901
> operate best in a -0.5 to +6.5 volt range spaning 7 octaves. The 901's 
> crap out if you send more than 6.5 volts to them added together. Easy 
> to do if you do not add together all the >voltages and calculate 
> correctly.
> After the 6.5 volt limit is reached the 901 will stop responding. Have 
> seen this on a demo where the operator claimed the 901 only had a 3 
> octave range, and it was clear it hit the 6.5 >volt limit. If the 
> offset was dialed in to -2 volts, the 901's would have had the full 5 
> volt range of the keyboard. Easy concept to grasp if you just stop and 
> think about how all these >voltages are added together. In a fixed 
> system like a Minimoog this is all patched in for you. It's up to the 
> operator in a modular system to get everything in range. The Moog 
> system >mostly has all the voltages you need for positive and negative 
> offsets right on the front panel. The +6 volt offset on the 901A might 
> seem unnecessary when they do not respond above 6.5 >volts. Untill you 
> try to patch in a 0-6 volt ribbon controller that you flip over to play 
> like a guitar. Since the pitch is backward this way, you invert the 
> voltage so it outputs 0 to -6 > volts. Now the +6 volt offset gets the 
> 901's in perfect range.
> The Moog system was very well thought out, but different from today's 
> systems in so many ways.
> 
>> 
>> Hi 'Daytona' and all members,
>> 
>> Replying to your last email about my Moog 904A clone freq range
>> questions on the DIY list (installed in my Dotcom system), I need some
>> clarifications about the original 904A CV input gains..
>> 
>> You stated:
>>> The 0-5v envelopes from
>>> Dotcom are less than the 0-6.3v swing the R.A.Moog 911 envelopes have
>>> and was designed to use. Could always use multiple inputs from the EG
>>> or use an amplifier to get up to R.A.Moog specs.
>>> 
>> Well I'm a bit confused here..
>> Looking at Moog's archives ADSR module 911
>> http://www.moogarchives.com/m911.htm
>> 
>> They say the peak DC output on Sustain level Esus is 5.5volts +/-10%
>> (Not that far from Dotcom 0-5vdc max) Starting from there I heard a
>> lot of 904A VCF being modulated by their Moog 911 ADSR with quite
>> satisfactory freq headroom.
>> So I should be able enough to modulate the VCF cutoff freq when
>> applying my Dotcom 0-5vdc ADSR voltage to one of the VCFs CV input
>> right ?
>> Well that's not what is happening.
>> 
>> Looking at the original 904A schematic:
>> http://www.freeinfosociety.com/electronics/schemview.php?id=943
>> The overall VCF spectrum can be spreaded using the front panel pot
>> called FIXED CONTOL VOLTAGE from 0v (filter wide open) to -6vdc
>> (filter shut off). OK.
>> The pot's wiper goes to a 51k resistor (R12).
>> But all the other CV input resistors (R1,2,3) are 100k values...
>> To me, that means that if I shut the front panel pot off
>> (-6vdc) and I try to 'full open' the VCF from one of the CV inputs
>> I'll need to feed 2 X +6vdc (+12vdc) at one of the 100k input resistor
>> to get the same CV gain right ?
>> Well from my knowledge Moog's CV voltages are not +12vdc.
>> How did the Moog modulars use their 5.5v CV voltages in this case ?
>> 
>> So far I built my 904A clone with all the same parts including the
>> 100k CV resistors.
>> The results: The front panel works very well has it should, but there
>> is a lack of gain in my modulations.
>> That's understandable from what I stated above.
>> 
>> So... to get the CV levels working, 3 solutions:
>> - Use external CV summer modules with gain.
>> All my external CV summer should have more than 'one' has a gain they
>> have now (maybe a gain of 3 instead) so the 904A VCF CV inputs will be
>> driven with more current for a better modulation.
>> 
>> - VCF's CV input resistors R1,2,3 could be lowered in value (40k..51k)
>> so a 0-5vdc swing will more 'drive' the 904A CV inputs.
>> But doing so I'll loose the 1v/oct range adjusted with trimmer R8.
>> So maybe I could only change resistors R2,3 for lower values and keep
>> R1 at 100k for 1v/oct purpose..
>> 
>> - Could always use multiple CV inputs at the same time for VCF
>> modulation.
>> 
>> These solutions are not pleasant ones...
>> 
>> In fact what is it going on these levels in the real Moog modules
>> world ?
>> 
>> J-Pierre
>> 
>> **********************************************************************
>> ******************
>> 
>>> Those 904A responses sound about right to me. The 0-5v envelopes from
>>> Dotcom are less that the 0-6.3v swing the R.A.Moog 911 envelopes have
>>> and was designed to use. Could always use multiple inputs from the EG
>>> or use an amplifier to get up to R.A.Moog specs.
>>>   The frequency range is just that and not the oscillation range. My
>>> stock
>>> 1969 904A did not even self oscillate by design. It was a few months
>>> latter that R.A.Moog made the revision to decrease the feedback
>>> resistance to allow self oscillation in the 904A. This required a mod
>>> to the stock 1.8K feedback resistor by adding a parallel 2.2K
>>> resistor piggy-backed.
>>> This gave a summed resistance of about 990 ohms (1K) and would begin
>>> to oscillate around the 7.5 mark on the regeneration dial.
>>> Pic of the mod before soldering here:
>>> http://mysite.verizon.net/vze6s1hi/sitebuildercontent/sitebuilderpict
>>> ures/904afeed  The lowest sine oscillation on this now modified 904A
>>> is around 140Hz.
>>> The
>>> response is not that linear but I can get about two octaves in tune.
>>> Requires a patch with adjustable intonation to vary the
>>> v/oct.response input. It really was not initally designed to be a 
>>> sound source, so...
>>> The
>>> 904B is the same way and when forced to oscillate is not very linear.
>>> But
>>> an octave and maybe two can be coaxed out of it.
>>> The lowest range setting on the 904A is the interesting one. The
>>> filter in this setting will not go much above a certain point, no
>>> matter how much voltage is applied to the CV inputs. Can hear it
>>> using a white noise source. There is a soft wall the 904A hits and
>>> nothing above that frequency passes.
>>>  The R.A.Moog I have here will soon turn 45 years old with a build
>>> date of Aug.11,1969. Still plays fantastic, and I play it almost
>>> every day! Just completed a few DIY modules for the Moog.
>>> Pics here:
>>> http://i58.tinypic.com/20a6ziv.jpg
>>> http://mysite.verizon.net/vze6s1hi/sitebuildercontent/sitebuilderpict
>>> ures/setup201
>>> 
>>>> Hi list,
>>>> I just finished the assembly of 2 new modules for my dotcom synth.
>>>> They are both Moog 904A low pass filter clones.
>>>> I built them with the same circuitry as the original.
>>>> I'm in the testing phase now.
>>>>  Reading the Moog archive original specs:
>>>>  http://www.moogarchives.com/m904a.htm
>>>>  It says that the 3 freq ranges are as follow:
>>>>  Cutoffs:
>>>> Position 1: 1hz to 5khz
>>>> Position 2: 4hz to 20khz
>>>> Position 3: 16hz to 80khz
>>>>  Now when I put the filter to max resonance, no waves at the input
>>>> and I move the control voltage knob I read on the scope:
>>>>  Sine oscillation from:
>>>> Position 1: 200hz to 5khz (32mv p.p.) Position 2: 250hz to 18khz
>>>> (32mv p.p.) Position 3: 350hz to 75khz (40mv p.p.)  It shows the
>>>> real min oscillation freqs for the 3 ranges are higher than the Moog
>>>> archive notes specs cutoff..
>>>>  Also the CV inputs seem to be calculated for 1v/oct right ?
>>>> Well 2 things appears:
>>>> Applying ADSR 0-5v envelope to this filter seems to move its cutoff
>>>> freq not as much I would have expected..
>>>> (50k 1% instead of 100k 1% would be better in the CV summing part
>>>> ???) and the linearity of 1v/oct is questionable too..
>>>> At full resonance, I can tune the 1v/oct CV range trimpot for some
>>>> of the filter span regions but others are not linear and out of
>>>> tune..
>>>>  The LTSpice tests I made with this original schematics showed these
>>>> behaviours too at low resonance freq.
>>>> The 2 modules I assembled behave exactly the same.
>>>>  Is there anybody who worked with that filter that could point out
>>>> any 'normal' caracteristics for that specific filter ?
>>>>  Finaly, is it behaving like it should with some 'normal
>>>> irregularities' ?
>>>>  Thanks
>>>>  JP
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy