Alembic - SF-2 [schematic]

[PaulBass]

I looked all over the web and I can't find a schematic for this preamp. Anyody know where I can find one? I emailed Alemic and never got a reply

[phatt]

Hey man , it's just 2 Para EQ circuits with a couple of extras thrown in.
So don't don't get to excited about brand names.
Any half decent para Eq circuit can be done and reap same or similar results.

Floppin hang chaps,, these brand names are just so overratted
but hey brandnames $$$$$ SELL $$$$$
Phil.

[Jarno]

Floppin hang chaps,, these brand names are just so overratted
but hey brandnames $$$$$ SELL $$$$$
Phil.

+1, I really really like their basses, but they are milking their brand to the limit. Look at the price of their onboard electronics, insane. 1600USD for two twopole filters and two mixing stages.

It's all high quality stuff, no doubt about it, but 1375USD for this filter seems a bit rich. To their defense, they do tell all you need to know on their website, so their's no mojo involved.

[phatt]

Hey man , it's just 2 Para EQ circuits with a couple of extras thrown in.

I wanted to compare it to the SVF filter I built

In which case you may wish to build a ParaEQ circuit and compare it with your SVF.
Build 2 and run in series if you want to get some idea of how the SF2 circuit might sound.

My understanding is the SVF and para circuits are very different.
Have fun with it.
Phil.

[Jarno]

The SF-2 is actually 2 State Variable Filters. A parametric eq doesn't give you the LP,BP and HP options the SF-2 has.

[phatt]

The SF-2 is actually 2 State Variable Filters. A parametric eq doesn't give you the LP,BP and HP options the SF-2 has.

Arrh Good point. Obviously I've got the wrong end of the Bull ay?
Phil.

[Jarno]

Came across some interesting info, apparently the sf2 does have state variable filters, but of the monolythic kind.
UAF42 is a specialised ic for filter applications.

I have a couple thinking that they could be interesting (for synth vcf applications), but when i tried to find info, the consensus was that they are not very suited for audio applications.
Apparently not so, Alembic is quite known for fidelity.

Pricey too, i think i paid something like 3.50usd per ic. Bollocks really, for a quad opamp with some caps.

[PaulBass]

Came across some interesting info, apparently the sf2 does have state variable filters, but of the monolythic kind.
UAF42 is a specialised ic for filter applications.

I have a couple thinking that they could be interesting (for synth vcf applications), but when i tried to find info, the consensus was that they are not very suited for audio applications.
Apparently not so, Alembic is quite known for fidelity.

Pricey too, i think i paid something like 3.50usd per ic. Bollocks really, for a quad opamp with some caps.

TI sent me a couple of UAF42's as samples and I didn't care for the sound. they were kinda noisy and hissy. I got better results with LT1359 and TLE2074's.

[Jarno]

Apparently it is what Alembic uses in this preamp, is there anyone here with an original unit who can confirm?

Maybe they'll just filter out the hiss, it is a bass preamp unit. Maybe they'll roll it off early, 6-7k or so? The noise thing they could cope with by having a large gain up front, and run a large signal, rather than keeping line level throughout. This is also done in modular synths, VCO's have a 10v p-p voltage (at least in some synth formats, Buchla IS line level for audio).

[tube-exorcist]

Now lets summarize the infos that we have:

We have a unit with 2 identical circuits/knobs and an added "Preamp".

One knob is mystically called "Damping Ratio" - normally this is called "Q" or "Bandwith"

Alembic reveals on their website:

The SF-2 filter circuit is a resistively tuned two-pole universal active filters, which yields 12 dB/Oct ultimate slopes in the high-pass and low-pass modes, and 6 dB/Oct ultimate slopes in the band-pass mode.

"resistively tuned" = with a potentiometer
"two pole .... filter" = stereo potentiometer
"6db/Oct ... in the band-pass mode": really ? at which "Damping Ratio" ?

Now let´s have a look at the guts:



We can see 3 identical 14-pin chips with the same logo on it. For two filters there would only be needed 2 identical mystery chips. So the UAF42 is not used.

Let´s have a closer look at the chip:



We can see from the tantals that pin 4 carries V+ and pin 11 carries V-.
Pin 3 and 5 (non-inverting inputs) are connected to gnd/bias via a resistor.
Between pin 1 + 2 and 6 + 7 (output and inverting input) there is a capacitor.
This is enough for me to know that this is a pretty standard quad opamp in a pretty standard state variable configuration with some mixing/routing/switching around, which can be found in the block diagram of the user manual:

http://www.samadhilounge.com/esa/bass/A ... (v2.0).pdf

http://www.bandatempo.com/baixos%25202/ ... MANUAL.pdf

End of brainstorming.

BTW, recently I bought two of this units on evilbay for $ 25,-- each:

It doesn´t look so bad inside (beware of the new SMD-junk !):

I think by modding them I can save a lot of money.....

[PaulBass]

I think by modding them I can save a lot of money.....

wow a Badringer, they are known to use the cheapest junk components because all their products are disposable. I would gut out the entire PCB board. maybe the pots aren't that crappy

[tube-exorcist]

I think by modding them I can save a lot of money.....

wow a Badringer, they are known to use the cheapest junk components because all their products are disposable. I would gut out the entire PCB board. maybe the pots aren't that crappy

I don´t know which components you normally use, but Rohm metalfilm resistors, Wima capacitors and LM833 aren´t that bad. As you can see in the picture in this old series there were used these high quality components. So don´t believe all rumours you read on the net.

[arcanek]

The SF-2 is actually 2 State Variable Filters. A parametric eq doesn't give you the LP,BP and HP options the SF-2 has.

They're basically the same. Depends on where you take the output. They just put in a switch to control that.

[arcanek]

Alembic reveals on their website:

The SF-2 filter circuit is a resistively tuned two-pole universal active filters, which yields 12 dB/Oct ultimate slopes in the high-pass and low-pass modes, and 6 dB/Oct ultimate slopes in the band-pass mode.

"resistively tuned" = with a potentiometer
"two pole .... filter" = stereo potentiometer
"6db/Oct ... in the band-pass mode": really ? at which "Damping Ratio" ?

No, 2 pole means 2nd order filter, not stereo (ganged). That's where the 6dB/Octave comes in. A first order filter is 6dB/Octave. 2nd order would be 12bB/Octave

[george giblet]

It's going to be a state-variable filter for sure. It's going to be a second order filter for low-pass, high-pass *and* band-pass.

The second order low-pass and high-pass are naturally 12db/oct

The 6dB/oct band-pass doesn't make sense. The slope of a band-pass filter depends on the Q. If the Q is less than 0.5 the slope is 6dB/oct as the band-pass is like a first-order low-pass + first-order high-pass. For Q. > 0.5 the slope depends on the Q. High Q band-passes have narrow band-widths and high slopes - think of a 1/3 rd octave equalizer.

I read Damping Ratio *Reciprocal* as being related to the Q. The term Damping Ratio Reciprocal is not conventional and it creates an unnecessary ambiguity for the meaning. Damping = 1/(2Q), so mathematically damping ratio reciprocal could mean 2Q. If the control is 1 to 10 then Q could be 1/2 to 5. If the control was just Q outright then the Q would be 1 to 10.

[arcanek]

It's going to be a state-variable filter for sure. It's going to be a second order filter for low-pass, high-pass *and* band-pass.

The second order low-pass and high-pass are naturally 12db/oct

The 6dB/oct band-pass doesn't make sense. The slope of a band-pass filter depends on the Q. If the Q is less than 0.5 the slope is 6dB/oct as the band-pass is like a first-order low-pass + first-order high-pass. For Q. > 0.5 the slope depends on the Q. High Q band-passes have narrow band-widths and high slopes - think of a 1/3 rd octave equalizer.

I read Damping Ratio *Reciprocal* as being related to the Q. The term Damping Ratio Reciprocal is not conventional and it creates an unnecessary ambiguity for the meaning. Damping = 1/(2Q), so mathematically damping ratio reciprocal could mean 2Q. If the control is 1 to 10 then Q could be 1/2 to 5. If the control was just Q outright then the Q would be 1 to 10.

State variable bandwidth isn't determined by Q. It's really not a bandpass filter. There's a band that gets passed, but the slope of the filter sections do not vary. Think of it more like 2 overlapping filters, rather than a single stage where the reactance/resistance ratio changes. Like 2 shelving filters, except they have gain.

[george giblet]

State variable bandwidth isn't determined by Q. It's really not a bandpass filter. There's a band that gets passed, but the slope of the filter sections do not vary. Think of it more like 2 overlapping filters, rather than a single stage where the reactance/resistance ratio changes. Like 2 shelving filters, except they have gain

Not true at all. The state variable filter isn't not like two overlapping filters. It is *one filter* with separate output for low-pass, high-pass and band-pass. The band-pass output of a state variable filter is often used in parametric equalizers. Parametric equalizers allow the Q to be varied.

Here's an example of the band-pass output. High-slope, high Q outputs.

The transfer functions of each section are as follows, (alpha = damping ratio) [original link]

You can see the band-pass isn't the cascade of the low-pass and high-pass. The band-pass is the high-pass output passed through an integrator. An integrator has transfer function K/s. The low-pass is the band-pass filter passed through an integrator. This is exactly what the state-variable filter does as you progress from left to right on the circuit.


MODERATOR NOTE: I apologize for modifying this post - this is really useful information and I'm trying to make it available on an as long as possible term, hence I copied images here.

[mauman]

I recently completed a pedal version of the Alembic SF-2, and I found this thread helpful in the design process, so I’m posting my results here rather than starting a new thread. There’s still no authoritative schematic of the SF-2 available, and I didn’t have an original unit to trace, but I’m pretty confident I’ve come up with a decent work-alike, or so says the bass player I made it for. I’m posting the schematic and a verified vero for the pedal adaptation I built. The original should be similar, with a few additions listed below.

In addition to this thread, I am indebted to the following sources:
Alembic SF-2 specs, description and block diagrams, http://alembic.com/support/SF2_Manual.pdf
Rod Elliott’s state variable filter (4 op amp version), https://sound-au.com/articles/state-variable.htm
Universal Filter by Uwe Beis, https://www.beis.de/Elektronik/AudioMea ... ilter.html
State variable filter for guitar by Vigstrom, http://blogg.vigstrom.se/2017/11/19/fil ... ic-guitar/
Alembic-like state variable filter schematic, https://www.musikding.rocks/gallery/ind ... schematic/

The original SF-2 was a two-channel second order state variable filter with constant gain, voiced for bass, rack mounted. It had an instrument-level input with preamp, two line level inputs and two outputs. It had three filter types (low pass, high pass and band pass), adjustable center frequency from 45 Hz to 6 kHz, and adjustable Q from 0.625 to 5. (The Q knob was marked in “reciprocal damping ratio” units of 1.25 to 10, which are just 2 times Q.) A mode switch selected either stereo (A and B channels independent with separate inputs/outputs) or mono (one input, split to A and B channels, which were mixed again at the output.) The clean input signal (called Direct) and filtered signal were mixed at the output using a pot for each. Direct signal level could be adjusted from zero to unity, and filtered signal level adjusted from zero to +20 dB. Power was 10.5 to 19.5 VDC, presumably regulated and inverted for a dual rail supply.

In my pedal adaptation, I’ve kept the instrument level input, associated preamp, and one output, eliminated the A/B line level inputs and buffers and the second output, and retained everything else. The input preamp gain (up to 20 dB) is pot-adjustable rather than the 3-position switch in the original. The mono/stereo switch now selects either channel A, or both A and B mixed, there’s no independent use of B. This A/B switch is a 3PDT toggle, the three sections are widely separated in the schematic and marked “A only” or “A+B”. The odd routing of the clean signal in A+B mode is retained from the original SF-2. I also added a notch filter for a total of four filter types, selected by a 1P4T rotary switch. Power supply (not shown on schematic) is +18VDC, regulated and inverted to +/- 15VDC (the max recommended for the op amps) with a total current draw of 80 mA including a 1 mA LED. Bypass switching is true bypass.

Schematic of pedal version, page 1 of 2

Schematic of pedal version, page 2 of 2

The nonstandard R and C values in the filters are made up by hand-selecting two series resistors or two parallel caps for each. Those caps should be high-Q, COG or similar composition. The filters have the same frequency range (45 to 6k Hz) and Q (0.5 to 5) as the original. The Q boundaries are set by R4 (low) and the value of the Q pot (high). The impedances within the filter are workable for TL074 op amps throughout the adjustment range of the F and Q pots. I started with 50k Frequency pots, but that put R7 & R8 around 392R which seemed low. Using 100k pots for Frequency and scaling the caps gave me a minimum op amp load of 755R which seemed safer.

The tapers of the pots were selected for intuitive response and to avoid bunching-up at either end of the range, as much as possible. Frequency pots are dual, and I could find slightly better matches between the two decks with B taper but the spread is much better with C taper. The Frequency pots were hand-selected for the best match between decks. The labels on the Freq pots are a little off the actual settings, 45 and 6k are correct, but 440 is actually 1k. I forgot that the tracks on reverse-log pots are actually multi-segment approximations… but there’s a chart in the user doc that gives the correct settings.

SuperFilter Guts 1 of 2

SuperFilter Guts 2 of 2

SuperFilter Front

I put this in a 1590XX enclosure to ensure plenty of working room, but it could easily be reduced to a 1590BB with a few changes. I think I’d swap the Q and Freq pot positions next time, since the Freq pots are taller and they prevented easy mounting of the main vero to the backs of the other pots. The Filter Type rotary switches are C&K A10405RNZQ 26mm, but any 1P4T will do. Knobs are 15mm and pots are 16mm Alpha. Since this was a one-off, I didn’t design a PCB, but you may be interested in doing that. The vero could also have a smaller footprint, you could integrate the preamp vero, etc. I like to keep charge pumps away from signal leads, so the power here is on a separate vero in the corner. I’ve been told my power protection and filtering practices are overkill, so you could simplify that portion. The heat sink of the MOSFET polarity protector is common with the drain which is +18V, so if you use one, insulate the tab with some heat shrink. I wasn’t sure if the final current draw would be over or under 100 mA so I used a 1A 15V regulator, but you could probably get away with 100 mA. I regulated the 18V down to 15V before the inverter. Q3/R2 on the power board prevent the op amp loads from pulling pin 5 of the LT1054 positive on startup. The circle around C1 is a grounded cylinder of aluminum tape to suppress square wave radiation from that cap, it’s very effective and eliminates any need for shielded signal wiring, the noise floor is very low. The LED +18V supply includes an RC circuit (Rpop, Cpop) to slow the LED turn-on and reduce popping.

About The SuperFilter Dual Bass Filter pg1.pdf

User doc, page 1 of 3

(45.17 KiB) Downloaded 110 times

About The SuperFilter Dual Bass Filter pg2.pdf

User doc, page 2 of 3

(60.03 KiB) Downloaded 101 times

About The SuperFilter Dual Bass Filter pg3.pdf

User doc, page 3 of 3

(413.38 KiB) Downloaded 97 times

Main and preamp vero

Main and preamp vero cuts

Power vero

Power vero cuts

Switch wiring diagram

SuperFilter Pot Wiring Chart.pdf

Pot wiring chart

(111.45 KiB) Downloaded 101 times

[mauman]

I forgot to provide the bench-test frequency response of the clean (unfiltered) signal in the pedal adaptation. It's quite flat. With a 200 mV p-p input, relative to 1 kHz, with the Preamp level knob fully counterclockwise (gain = 1), the -1 dB points at the output are 13 Hz and 41.7 kHz. With the Preamp level knob fully clockwise (gain = 10), the -1 dB points are 14 Hz and 26.5 kHz.

[Ferniff]

I forgot to provide the bench-test frequency response of the clean (unfiltered) signal in the pedal adaptation. It's quite flat. With a 200 mV p-p input, relative to 1 kHz, with the Preamp level knob fully counterclockwise (gain = 1), the -1 dB points at the output are 13 Hz and 41.7 kHz. With the Preamp level knob fully clockwise (gain = 10), the -1 dB points are 14 Hz and 26.5 kHz.

I made an account just to comment on this.


Why hasn't anyone commented; this is amazing. I've been playing around with the Craig Anderton Super Tone Control, and I think I want to tackle the SF-2 next. I have two super tone controls inside my bass guitar, so I'm looking to do the same with the SF-2.