back to JH. homepage

JH.'s Farfisa VIP 500 page

20.11.2006 - Finally, a Farfisa!

Normally I'm into synthesizers and electronic FX, but I always wanted a Farfisa organ again, since I'd sold this VIP345 many years ago.
I'm not so much into straight combo organ playing - it's stuff like Rick Wright's performance on Set the Controls for the Heart of the Sun on the Pompeji video that gives me inspiration.
I want to set up a corner in my little studio, just with an organ and FX devices. As Compact Duo's are rare and expensive, and as I quite liked that VIP 345's sound, and as I wanted to have a second manual, I thought hunting down a VIP 500 would be a good idea. So I bought one on ebay - picked it up yesterday.
The upper manual looks and sounds familiar - the lower manual is ... how shall I put it ... crying for modifications.

The infamous Lower Manual

Even with all three drawbars activated (Flute, Oboe and reed), it still sounds "flutish".
I think of summing these 3 internally and feed it to one drawbar, and develop two new filters for the other two drawbars.
Problem is, it's all derived from a 8' *Square* wave, which doesn't leave much room to sounds that won't sound somewhat hollow.
Well, might run a mix of unfiltered square signal to one of the LM drawbars, to give the maximum of harmonic contents, bring it out on a separate jack, and run it thru external filters.
I've started to set up a single-oscillator, 50% pulse, VCF wide open, sound on my OB-8, and to run it thru various filters of a modular synth. I hope to find some interesting treatment for a bland 8' square wave of the VIP that way.
If everything else fails, I can disconnect the LM key contacts, bring them out on a 50-Pin connecter.
Imagine that, VIP sound on the upper manual, and playing my PolyKorg Clone on the lower manual.
But I only consider that as a last resort - I really want to just make the organ's LM more useful by small modifications.

22.11.06 - Looking Inside

OK, I did open it.
Impressive stuff - built like a tank.
Made a lot of photos.

No way to replace these divider circuits with something more reminiscent of a Farfisa Compact's sync'ed oscillator dividers, though. The divider board is too integrated with the key switches, mechanically. I guess I'll concentrate on the Lower Manual with my modification ideas now.

Here are the pictures. Click on thumbnails to open the large photos. Some of them are huge (> 1MB), you can even read component values. (All images copyright (c) 2006 J. Haible)






































27.11.06 - Lower manual analysis and modification

The VIP 500's lower manual has 49 keys (4 Octaves) which are only fed with 8' tones (square waves from the frequency dividers). That means the lowest C always has a fundamental frequency of (roughly) 139 Hz, and the highest C has a fundamental of 2218 Hz. The harmonics of this raw material from the key contacts goes up from the respective fundamental to way above the hearing range. Even numbered harmonics are missing, due to the 50% duty cycle of the square waves.
These harmonically rich (though lacking in even nubered harmonics) tones are then heavily filtered into much more mellow sounding voices, which are available at the 3 white drawbars labelled "flute", "carinet" and "reed".

This filtering is most severe for the darkest of these three options, the flute. It's supposed to be close to a sine wave. This picture shows the filter courve from 139Hz (lowest note) to 16kHz (standard value considered as upper end of hearing range). Note the heavy filtering of higher frequencies to reduce the harmonic contents of a square wave to a "flute" sound. Also note how the same filtering attenuates the fundamental (and thus the volume of the whole voice!) in the upper keyboard range. And it's not just theory: If you only register the Flute on the lower manual, the sound almost fades out towards the upper end of the keayboard.

The filter function for the "clarinet" allows a little more harmonics to come thru. It's still pretty severe filtering, though. See picture.

The "reed" is the brightest of the three lower manual voices. It's filter function is shown here. It's still not a very bright filter, though. If you're setting a combination of "flute" and "reed", you can get a decent lead (melody) voice, that - in my book - still resembles a kind of flute sound. A searing solo voice, as possible from a synthesizer or from the VIP 500's upper manual, is totally out of reach on the lower manual.

The following picture shows the frequency response of  a LM flute drawbar (at maximum) with the LM reed drawbar gradually mixed in. It's evident how the sound becomes gradually brighter, and yet stays much on the mellow side even with maximum setting of the reed addition.

Now, with a little modification, I made the reed filter response look like this. If you gradually mix this modified reed drawbar to the flute drawbar, you're getting this, much brighter response. In this new version, maximum flute plus lowest reed setting (1 of 0 ... 4) sounds very similar as maximum flute plus maximum reed in the unmodified version. And starting from that, you have 3 more levels of brightness to go! Now you can play rich chords that rival the upper manual's richness of tone, and searing "square leads". Don't expect miracles'though: We're still dealing with square wave input material, and missing even order harmonics.

The modification is very simple. Solder a 1nF capacitor (polyester, polystyrene, ceramic, whatever you please) between two points that are accessible from the top of the topmost circuit board, when th eorgan lid is opened. Nevertheless, I take no responibility for any harm and danger to you, or your organ, or whatever, if you do this modification. Unplug the organ from the mains, and follow the usual precautions, before you start. (If you wonder if you're qualified to do a mod like this, you probably aren't.) Don't blame me when your organ (or you) go up in smoke.
Here's a high resolution picture where to solder the capacitor. First, carefully heat the leads to the two resistors in question (just a millimeter or so, that shows on the upper side of the pcb), and make them "wet" with solder. Cut and bend the capacitor's leads to fit to these two connection points, then wet them with solder as well. This makes it much easier to solder the final connection.
In case you don't like the sound of this, you have removed this modification in half a minute, by just unsoldering the 1nF cap.

06.12.06 - "un-squaring" the Lower Manual

Forget the previous entry. That little modification does make the LM brighter (when using the Reed Drawbar), but it still sounds hollow. That's because square waves are used as "raw material" everywhere in the organ (unlike the Farfisa Compact organs, which used an assymmetrical pulse wave). And square waves have no 2nd harmonic - in fact, no even numbered harmonics at all. This is quite convenient if you want to filter a (more or less) sine-shaped waveform: Your filter needs to be close to unity gain at the fundamental, and considerably attenuating at 3 times the fundamental frequency. At twice the fundamental frequency (where a 2nd harmonic would be located), the filter has nothing to do at all, because the 2nd harmonic is simply missing. So it's easy to see that the filter doesn't need to be as steep as it would have to be when starting from a waveform that contains the 2nd harmonic. Or, looking at the situation from a different side, with a filter of a certain "steepness" you can run a wider range of  keys thru the same filter, without getting the fundamental attenuated.
I won't go into the increadibly cheap implementation of the VIP 500's Flute filter on the Lower Manual here, which in fact causes a huge amount of attenuation for the fundamental on the higher two octaves on the keyboard, here. Maybe later.

For the moment we just note that squares are a good starting point for sine waves.
And for sounds that emulate stopped flue pipes. Or a clarinet. For anything that's missing equal numbered harmonics.
But if you try to add a lot of treble, letting the square go unfiltered or barely filtered (like I suggested above), your sound will always have this "sqare-ish", hollow sound. Good for synthesizer "square leads" - and no good for much of anything else.

18.12.06 - The final stategy, or 3 Steps to a good-sounding Lower Manual


Three steps:
1. Adding even numbered harmonics to the raw generator signal
2. Modify the three Lower Manual drawbars to allow control of a wide harmonic content
3. Add an insert point for the LM sound, to allow phasing, chorus, distortion, you name it, to be applied to the LM individually.

1. Adding even numbered harmonics

The following picture shows the waveform and harmonic spectrum of the raw generator signal, that is fed via 68k resistors to the Lower Manual key contacts:



Note the missing even numbered harmonics - the absence of these is responsible for a "hollow" sound.
The next picture shows the modification. Where the original 68k resistors feeds to 8' square wave, a new 130k resistor feeds a 4' square wave at half the 8's amplitude:



Now, where to get the 4' signals? The cable bus that runs from the divider board to the lower manual only carries the 8' footage.
But, of course, looking for a 4' signal for a specific note (say: C3), we just need the 8' signal from the note one octave above (say: C4).
So all we have to do is adding a 130k resistor for each key contact, and a wire that taps the signal one octave higher.
Here's some photos how this is done:





Easy! No rocket science, just a lot of soldering work.
Remaining question: what about the highest octave?
Well ... I was prepared to run 12 extra wires to the divider board, which would certainly have been possible.
But then I found in that range, there isn't much "hollow-ness" in the sound, so I just left the highest octave unmodified.
It's ok., really. In th eold days, they called it "keyboard foldback".

And finally, here's what the new waveform and spectrum looks like:



We now have a staircase waveform - a crude approximation of a saw wave.
We now have a 2nd harmonic, and it's even the strongest of all harmonics!
The 4th is still missing, but this is ok.
These are simulation results - the real signals look very similar. The glitch after the second stairstep down is a simulation artefact - it's not there in real life.

2. Modify the three Lower Manual drawbars to allow control of a wide harmonic content

The original choice of sounds on the VIP 500's LM is a joke:
Very dark and hollow ("Flute" Drawbar)
Almost as dark and hollow ("Clarinet" Drawbar)
A tad of brightness and still hollow ("Reed" Drawbar)

After the modification, the drawbars have the following functions:
"Flute" (unaltered) - Low Frequency range, dark, fundamental (but not hollow anymore!)
"Clarinet" (modified) - a 1kHz Band Pass filtered signal. Remotely related to the Farfisa Compact's "Oboe" register.
"Reed" (modified) - Bright, Buzzy, Brilliant
Basically, the whole audio range is now divided into the sections < 1kHz, around 1kHz, and > 1kHz.

The modifications are shown in red in the following schematics excerpt:
(For the unusual value of 36nF, I used a 39nF capacitor that measured 36.5nF. I could have used 33nF and 3.3nF in parallel.)



Here's a picture that shows the various modifications on the LM printed circuit board:



Note the 560mH inductor, wound on an RM core, and located where the mains transformer's stray field won't induce hum. This Inductor is used for a passive 1kHz band pass filter that filters the "Clarinet" (now rather: Oboe) drawbar.
(The picture also shows the buffer stage for th einsert point, as described below.)

And here's a drawing that shows what I've changed on the board, in detail:


3. Add an insert point for the LM sound

It's obvious that even with a rich waveform to start with, a set of only 3 drawbars doesn't allow the sophisticated filtering we'd like to have. Instead of adding a lot of additional fixed filters inside the organ (and then - use what to select and control them ??), I decided to add an Insert point for the lower manual, to allow for external filtering. Or external FX processing. Or watever.
So the combined signal of the three LM drawbars (which provide a rough pre-shaping of the sound) is buffered with an emitter follower, routed to the tip connection of a 1/4" TRS jack, normalised to the ring connection of the same jack, and then routed back to the VIP 500's "output preamp".
Now I have three options to use the Lower Manual:
Play it just as it is (it's a lot more useful after the modification, even without any external processing), or
Use an Insert Cable (one TRS plug to two 1/4" mono plugs) for running the LM signal thru outboard FX, guitar pedals, etc., or
Use an ordinary 1/4" cable to run the LM signal to an amplifier / mixing console separately from the UM signal. In that case, the main output carries the UM signal, without the LM signal being added inside the VIP 500.

I sacrificed the Headphones output for this. (The HP output was pretty useless anyway, with it's unusual pinout.) Thus, I didn't even have to add an extra jack to my organ.

Here's a picture of the emitter follower buffer stage for the insert point:





19.12.06 - Reducing the noise of Upper Manual drawbars

I noticed my VIP 500 had some background noise. I wouldn't call it very noisy, but with the additional gain of a distortion pedal (you do run your organ thru Fuzz boxes and tube amps, dont you?), and just the 16', 8' and 5 1/3' active on the Upper Manual, it's definitely there.
I searched the circuits for noise sources. And I found the noise is louder when the Upper Manual Flutes volume is increased on the output section slider. So the main suspect was the "Flute Preamp" - that is the summing amp for all UM flute drawbars, whose output is then fed into the output mixer ("Output Preamp" in Farfisa speak).
It could have been a noisy transistor, but my guess was an unsufficient supply voltage filtering. The Flute Preamp is a single stage transistor amp with degenerate emitter feedback, resulting in a high output impedance. The reference voltage (AC ground) for the output signal is the positive rail, not 0V. So any noise on the rail will appear in the output signal. The rail is filtered locally with a 1kOhm, 470uF low pass. A quick simulation showed that the 1k could be considerably  increased without loosing much headroom for the amplifier, because it's current consumption is low.  I replaced the 1k with a 8.2kOhm resistor, which should reduce the noise about 18dB. And Bingo - it works. I haven't made any measurements to confirm the 18dB, but it's a lot more quiet now.
See all the details about this little modification in the following picture (click on image to enlarge):



to be continued ...

A very rough sample of Upper Manual playing Church Organ, and Lower Manual playing Synthesizer (UM and LM processed separately with JH. Synthi Clone, then going thru Tape Echo and Quantec.) Sorry for the nasty distortion.

back to JH. homepage

check out my music:
My latest Album "The Secret Life of Trees"