Doc Sketchy's new analog FREQUENCY SHIFTER!

[dmr-songquito]

Here are some thoughts about the modulation inputs. One may think that the separate SIN and COS inputs are not required, because we have MOD, but both modulation input types provide their own unique advantages in my opinion.

a) MOD is very convenient to be used with any modulation source that does not provide separate SIN and COS signals.

b) The separate SIN and COS inputs may be connected to a Thru Zero Quadrature VCO. I assume the following. If the VCO signal passes the zero point and starts to oscillate "backwards", SIN and COS should swap their roles. The SUM and DIFF outputs of the frequency shifter should also swap their roles then. SUM then starts to output the upshifted signal and DIFF the downshifted one (as opposite to their "normal" roles). I did not try it yet, because I don't have a Thru Zero Quadrature VCO yet. Maybe I am wrong? The Doepfer A-110-4 would be a candidate.

What do you think?

You're absoutely right. My freak shift arrived a few days ago and I placed right next to the DOEPFER A-110-4 to allow exactly for those seamless through zero frequency shifts. And it works like a charm.

This is the inharmonicity corner of my rack, where I also have the Harmonic Shift Oscillator which internally uses shifting to create inharmonic oscillations. Next, there's the DOEPFER A-126-2 Frequency Shifter and the Dreadbox Antiphon which I use as a flexible sine oscillator bank.

Since someone asked for a comparison between the freak shift and the A-126-2, here are some early impressions of mine:

Freak Shift pros:

• Virtually no bleed through of the shifting oscillator
• Second dome filter lets you use any signal for shifting
• Integrated feedback path with breakout option
• Separate outputs for up- and downshifted signals
• High quality, hand-soldered module with wide front panel

Freak Shift cons:

• No internal shifting oscillator, but mod input is actually more useful
• No black front panels available anymore

A-126-2 pros:

• Available with black vintage front panel
• Less deep module (rougly 5.5 cm)
• DOEPFER quality

A-126-2 cons:

• Audible bleed through of the shifting oscillator, getting more problematic when adding an external feedback path
• Internal shifting oscillator does not track V/octave which makes it less useful for me
• No internal feedback path
• Separate outputs for up- and downshift need extra breakout module
• Sine and cosine inputs seem to be swapped, but could also be a problem with the A-110-4, need to check with a two channel oscilloscope

I'm thinking of making a more in-depth youtube video comparing both modules, but this may take some time.

[ronv]

dmr-songquito, thank you for confirming! Based on your response, I just ordered a Doepfer A-110-4 Thru Zero Quad VCO one minute ago :-)

Thank you as well for the comparison of the two frequency shifters. I am also very happy that there is virtually no bleed through of the shifting oscillator in the FREAK SHIFT. This is great! I have a quite small bleed through of the input/aux signal in my FREAK SHIFT, but this is absolutely no problem at all. Do you also observe this (just to be sure that everything is fine)?

[Dcramer]

Hey Dmr, in regard to tracking with the A-126’s internal oscillator, is there a way to set its CV control to get it close to tracking?
I have an original A-126 and if I set it around 1-2 o’clock the internal OSC will track 1/Oct, but of course it’s quite tricky to get it perfect and the OSC doesn’t go low enough for phasey stuff.

All of this is simple on the Freq

[mcetilia]

PMED

[rigus]

I should also mention that building 50 of these Freak Shift modules would have been absolutely impossible had I not figured out how to solve the general Dome filter problem, which I learned from Orchard's 1950 paper, by way of Douglas's 1961 thesis.

Really vaIuable information! I am trying to build an 18-stage Dome filter. Do you have a reference for the Douglas thesis at hand, or the formula for calculating time constants and capacitor/resistor values?

- Paul

[dmr-songquito]

I have a quite small bleed through of the input/aux signal in my FREAK SHIFT, but this is absolutely no problem at all. Do you also observe this (just to be sure that everything is fine)?

To answer your question, I dived a little deeper in my comparison test of Freak Shift vs. A-126-2. I generated a test tone scenario, where I have a sine oscillator at roughly 4000 Hz and shift it with a 1000 Hz carrier. For upshifting that yields 5000 Hz and for downshifting 3000 Hz. No fun to listen to but useful to visualize the frequency shifters' output behavior in a spectrogram. Note that I also added a slight vibrato to the 4000 Hz oscillator only to make it visually more pleasing and better distinguishable from the carrier.

Spectrogram of some test tones on the Freak Shift.

In the top row, I indicate how I changed the knobs over time. I'm quickly fading in and out the different oscillations at the output. What can be seen is that there is a tiny bit of bleed through of both the input signal as well as the opposite shifted signals in the extreme settings of the knobs. No trace of the carrier oscillator though, which is nice.

Spectrogram of some test tones on the A-126-2.

Different situation on the A-126-2, as there is no dedicated knob to dial in wet/dry ratio. According to DOEPFER's specs, it's internally wired to have completely wet output. One can see that both the input signal and the opposite shift signal (lesser amount) bleed are bleeding through. Even worse, the carrier and it's first harmonic are always present in the output. It's the two straight lines at 1000 Hz and 2000 Hz.

Sorry for being so picky about the signal leakage I had only used DSP-based frequency shifters so far, even implemented my own frequency shifter in VCV rack. Of course, in digital implementations leakage is no problem at all, but aliasing is the big deal there.

[ronv]

I have a quite small bleed through of the input/aux signal in my FREAK SHIFT, but this is absolutely no problem at all. Do you also observe this (just to be sure that everything is fine)?

To answer your question, I dived a little deeper in my comparison test of Freak Shift vs. A-126-2. I generated a test tone scenario, where I have a sine oscillator at roughly 4000 Hz and shift it with a 1000 Hz carrier. For upshifting that yields 5000 Hz and for downshifting 3000 Hz. No fun to listen to but useful to visualize the frequency shifters' output behavior in a spectrogram. Note that I also added a slight vibrato to the 4000 Hz oscillator only to make it visually more pleasing and better distinguishable from the carrier.
freak_shift_test_annotated.png
In the top row, I indicate how I changed the knobs over time. I'm quickly fading in and out the different oscillations at the output. What can be seen is that there is a tiny bit of bleed through of both the input signal as well as the opposite shifted signals in the extreme settings of the knobs. No trace of the carrier oscillator though, which is nice.
a-126-2_test_annotated.png
Different situation on the A-126-2, as there is no dedicated knob to dial in wet/dry ratio. According to DOEPFER's specs, it's internally wired to have completely wet output. One can see that both the input signal and the opposite shift signal (lesser amount) bleed are bleeding through. Even worse, the carrier and it's first harmonic are always present in the output. It's the two straight lines at 1000 Hz and 2000 Hz.

Sorry for being so picky about the signal leakage I had only used DSP-based frequency shifters so far, even implemented my own frequency shifter in VCV rack. Of course, in digital implementations leakage is no problem at all, but aliasing is the big deal there.

Many thanks for this thorough comparison and the very nice visualization! This is very helpful and shows clearly the quality of the FREAK SHIFT. According to the spectrogram, one can also clearly see the small bleed of the main input signal in the 5th section (just when "fully wet" had been dialed in), but this is probably the lowest bleed of the existing analog frequency shifters and totally fine. It is really nice that the modulation signal does virtually not bleed into the output signal.

BTW: My Doepfer A-110-4 Thru Zero Quad VCO arrived In the meantime, so I can try it out as the modulator in the next days.

[Dr. Sketch-n-Etch]

Just a thought: Bear in mind that some of the time constants in the Dome filters are pretty large, so some signals may take some time to completely dissipate out of the unit -- not so much bleed, but analog delay.

[daphnid]

Here are some thoughts about the modulation inputs. One may think that the separate SIN and COS inputs are not required, because we have MOD, but both modulation input types provide their own unique advantages in my opinion.

a) MOD is very convenient to be used with any modulation source that does not provide separate SIN and COS signals.

b) The separate SIN and COS inputs may be connected to a Thru Zero Quadrature VCO. I assume the following. If the VCO signal passes the zero point and starts to oscillate "backwards", SIN and COS should swap their roles. The SUM and DIFF outputs of the frequency shifter should also swap their roles then. SUM then starts to output the upshifted signal and DIFF the downshifted one (as opposite to their "normal" roles). I did not try it yet, because I don't have a Thru Zero Quadrature VCO yet. Maybe I am wrong? The Doepfer A-110-4 would be a candidate.

What do you think?

Does this not work the same as when using a thru-0 VCO (like Rubicon) into the mod input?

I'm considering getting one of the thru-0 quadrature Doepfer VCOs for this purpose, but am also very intrigued with the new Orbit 3 module from Joranalogue as a modulation source, which has some phase related outs (in intervals of 60 degrees). I already have an Erica Octasource for quadrature lfo rates and would like to spice things up a bit.

It would be really nice to have thru-0, just hoping that the mod input can take advantage of it in the same way as the sin/cos ins.

The A-110-6 does seem like the ultimate modulator for this thing though. Eagerly awaiting delivery, I want to thank Doc Sketchy for sharing this with us all

[luchooo]

Hi all,
So this module is 80mm deep right? Anybody knows if it will fit in a rackbrute 6U?
I measured mine and the lower part of the case seems to be almost 80mm

[pieter]

Hi all,
So this module is 80mm deep right? Anybody knows if it will fit in a rackbrute 6U?
I measured mine and the lower part of the case seems to be almost 80mm

I just measured mine with a caliper from the back of the panel* to the top of the power cable header, and it is no more than 77 mm. I will update modular grid. Some modules have the power header as the only tall part on the back, but not this one. There are caps all over the PCB that are nearly as tall as the power header. You will need all 77 mm everywhere behind the module.

* I do not know how others measure the depth of modules, but the above method makes the most sense to me. So you should measure the distance from the top of the rails to the bottom of your case, both for the upper rails and the lower rails, in case there is an angle to the front relative to the bottom. If you are really paranoid, you could cut out a cardboard panel and glue a cardboard piece to the back that sticks out 77 mm, and see if it fits. Or build something with Lego if that's easier.

[luchooo]

If you are really paranoid, you could cut out a cardboard panel and glue a cardboard piece to the back that sticks out 77 mm, and see if it fits. Or build something with Lego if that's easier.

Thank you so much for the help! I will get an exact measurement as soon as possible and see if it will fit.

[a100user]

Just a thought: Bear in mind that some of the time constants in the Dome filters are pretty large, so some signals may take some time to completely dissipate out of the unit -- not so much bleed, but analog delay.

Just tried it and alas the Freak Shift does not fit in a 6U BruteRack, it’s about 3-4 mm too deep

[ZenPunkHippy]

?

[luchooo]

Just a thought: Bear in mind that some of the time constants in the Dome filters are pretty large, so some signals may take some time to completely dissipate out of the unit -- not so much bleed, but analog delay.

Just tried it and alas the Freak Shift does not fit in a 6U BruteRack, it’s about 3-4 mm too deep
ABF66B01-7EF6-4AA9-BD63-22AE478800F4.jpeg

That's truly heartbreaking thanks for sharing the photo and info

[mandarinpile]

I think that Freak Shift is worth buying a new case for

(patiently waiting for my unit)

I googled today some other frequency shifters. Seeing the price on Synth-Werk SW 1630 was my favorite moment today. Because I got to remember what a great thing we have here in this thread

[ronv]

Here are some thoughts about the modulation inputs. One may think that the separate SIN and COS inputs are not required, because we have MOD, but both modulation input types provide their own unique advantages in my opinion.

a) MOD is very convenient to be used with any modulation source that does not provide separate SIN and COS signals.

b) The separate SIN and COS inputs may be connected to a Thru Zero Quadrature VCO. I assume the following. If the VCO signal passes the zero point and starts to oscillate "backwards", SIN and COS should swap their roles. The SUM and DIFF outputs of the frequency shifter should also swap their roles then. SUM then starts to output the upshifted signal and DIFF the downshifted one (as opposite to their "normal" roles). I did not try it yet, because I don't have a Thru Zero Quadrature VCO yet. Maybe I am wrong? The Doepfer A-110-4 would be a candidate.

What do you think?

Does this not work the same as when using a thru-0 VCO (like Rubicon) into the mod input?

I'm considering getting one of the thru-0 quadrature Doepfer VCOs for this purpose, but am also very intrigued with the new Orbit 3 module from Joranalogue as a modulation source, which has some phase related outs (in intervals of 60 degrees). I already have an Erica Octasource for quadrature lfo rates and would like to spice things up a bit.

It would be really nice to have thru-0, just hoping that the mod input can take advantage of it in the same way as the sin/cos ins.

The A-110-6 does seem like the ultimate modulator for this thing though. Eagerly awaiting delivery, I want to thank Doc Sketchy for sharing this with us all

I am afraid that the MOD input will not be sufficient if you would like to use a through-zero-oscillator for modulation with the goal to switch the up-shifted and down-shifted outputs when the zero point is crossed. The frequency shifter does not know when the zero point of the modulation oscillator has been passed and that the modulation oscillator runs "backwards" now.

The reason is that there is no "backwards" in a single, symmetric sine wave signal. It just swings the voltage as before from zero to positive back to negative, and back upwards, and so on. The "through-zero" of a through-zero-oscillator is that the signal does not stop completely when the control voltage is as low that the frequency passes the zero point, but that it starts to swing again. This enables modulation in both directions around the zero frequency.

We need a through-zero quadrature oscillator that has separate SIN and COS outputs. This means that when the frequency control voltage gets lower the oscillator almost stops at a certain point and when the control voltage gets even lower, the oscillator does not only start oscillating again, but the phase shift between the quadrature outputs SIN and COS is reversed as well. Instead of the COS signal being 90° behind the SIN signal, now the SIN follows COS with 90° difference.

If these two signals (SIN and COS) are connected to the frequency shifter's SIN and COS inputs and the manual switch selects these inputs as modulation source, the following happens. If the through-zero quadrature oscillator passes the zero point, so that SIN and COS change their roles, the frequency shifter outputs "SUM" and "DIFF" change there roles as well. This means that the "SUM" output now starts to output the up-shifted signal and "DIFF" starts to output the down-shifted signal in opposite to the "normal" state. This enables you to use an LFO to modulate the through-zero-oscillator around the zero point and get a nice spatial kind of symmetrical frequency modulation that is great in stereo.

[johannes]

thanks. Great explanation!

Would be curious how that would sound like.
And Would also be interesting how a symmetrical tzfm waveform effects the timbre of frequency shifting.

@doc
i was thinking about getting a rubicon2 for its nice linfm sound and as a mod for the freakshift but i guess there is no way to get a sin/cos outs from it, right?

[daphnid]

If these two signals (SIN and COS) are connected to the frequency shifter's SIN and COS inputs and the manual switch selects these inputs as modulation source, the following happens. If the through-zero quadrature oscillator passes the zero point, so that SIN and COS change their roles, the frequency shifter outputs "SUM" and "DIFF" change there roles as well. This means that the "SUM" output now starts to output the up-shifted signal and "DIFF" starts to output the down-shifted signal in opposite to the "normal" state. This enables you to use an LFO to modulate the through-zero-oscillator around the zero point and get a nice spatial kind of symmetrical frequency modulation that is great in stereo.

I see, thanks for the explanation. I'm very interested in exploiting the stereo effects that can happen by flipping the outputs with through zero.

I'm still considering the Joranalogue Orbit 3 as modulator, but one thing I'm unclear about, and maybe someone can clear this up: Would the Sum and Diff outs flip roles in a similar way when the Orbit 3 switches which well/attractor it's orbiting? The Orbit 3 doesn't have a 90 degree shifted out but it does have 60 degree shifted outs. So I guess it's a "sexture" VCO?

Considering that I already have an octature LFO and quadrature function generator it seems like the Orbit 3 would add a lot more to my system as a whole. However this thru-0 sum/diff stereo flip-flop behavior seems like it could create some nice psychoacoustic effects, which is exactly what I've been building the stereo functionality of my rack around.

[Dr. Sketch-n-Etch]

thanks. Great explanation!

Would be curious how that would sound like.
And Would also be interesting how a symmetrical tzfm waveform effects the timbre of frequency shifting.

@doc
i was thinking about getting a rubicon2 for its nice linfm sound and as a mod for the freakshift but i guess there is no way to get a sin/cos outs from it, right?

You don't need to. Just send the Rubicon's output to the MOD input and Dome-filter it. You get the advantage of through-zero without the need for separate SIN and COS waveforms.

[johannes]

Thanks, just to make sure: i dont need individual sin and cos signals of a tzfm modulator to switch the up-shifted and down-shifted outputs when the zero point is crossed?

[dmr-songquito]

If these two signals (SIN and COS) are connected to the frequency shifter's SIN and COS inputs and the manual switch selects these inputs as modulation source, the following happens. If the through-zero quadrature oscillator passes the zero point, so that SIN and COS change their roles, the frequency shifter outputs "SUM" and "DIFF" change there roles as well. This means that the "SUM" output now starts to output the up-shifted signal and "DIFF" starts to output the down-shifted signal in opposite to the "normal" state. This enables you to use an LFO to modulate the through-zero-oscillator around the zero point and get a nice spatial kind of symmetrical frequency modulation that is great in stereo.

To make this more confusing, my A-110-4 module has the "SIN" and "COS" outputs swapped. That means I have to wire it's "Sine" output to the "COS" input of the Freak Shift and vice versa. Only this way, I get the upshifted signal at the "DIFF" output and can properly modulate the frequency shift through zero. To visualize this, I used the two-channel oscilloscope of VCV rack on the A-110-4 outputs, where I wired the "Sine" to left channel and "Cosine" to right channel.

Two channel oscilloscope on the sin (cyan) and cos (magenta) output signals of the DOEPFER A-110-4.

In their specs, DOEPFER claim that

Cosine is nothing but a sine with 90 degrees phase shift according to the formula: cos(x) = sin(x + pi/2)

But it should the other way round: sin(x) = cos(x + pi/2). Which is equivalent to: cos(x) = sin(x - pi/2). I did the same test setup with other modules that produce coupled sine and cosine signals and here the outputs are correct. As an example, the Harmonic Shift Oscillator from New Systems Instruments does it correctly, with very clean sinusoidal outputs. The downside is that it does not go through zero.

Two channel oscilloscope on the sin (cyan) and cos (magenta) output signals of the New Systems Instruments Harmonic Shift Oscillator.

[Dr. Sketch-n-Etch]

UPDATE

Hey everyone, sorry I've been a bit slow getting Freak Shifts out the door. The remaining Dotcom modules will be all finished over the next few days (I've actually got a week off from teaching, so I will be building like a fiend next week). Then it's back to eurorack modules. Orders coming in now (including 3 new orders just today) won't ship until April. I'm now up to number 119 on my customer database, and 6 people never paid, so that means that I will have built 113 of these things when I finish these orders -- all by hand. That's about 700 hours of my life.

[Dcramer]

Gee…and here I’m hoping you’ll have time to build some other cool stuff…

[johannes]

Yes, please a rubicon successor