Maestro/Oberheim Sample & Hold. Built. Fun.

[Mark Hammer]

Bit by bit, I'm getting around to finishing things that were sitting in the bin-o-boards.  The last two days were the Maestro S&H unit, using the General Guitar Gadgets layout, and 3080 OTAs, rather than the Tonepad layout using a 13600.  Nothing against that one.  I justr happened to have the other one stuffed already.  Unfortunately, it appears I etched the board back when I couldn't readily differentiate a PnP style pattern from a more traditional one.  So, this one ended up being etched "backwards", and the chips had to be installed from the copper side, with all the other parts on the normal side.

Took me a bit to get all the transistor orientations right.  I used BF245C JFETs, and they worked just fine.  My noise aource ended up being a metal can 2N2222.  The one change I implemented was a variable resonance control. R7 in the GGG layout is shown as 1M.  This sets the resonance of the filter.  I used a 200k fixed resistor in series with a 2.5M pot.  But the pot feels too stiff for my liking, and I don't see much utility in the in-between positions, so I think I may just replace R7 with a 2M2, and use a 3-position toggle to switch in other parallel resistors to get a high, medium, and low resonance setting.  Some sort of glide control would also be nice.

The trimpots take some tweaking to find the sweet spot, but once you find it, it's sweet indeed.  Once I put in a decay pot, I think I'll leave it at that: sensitivity, resonance, decay, clock rate, glide.  I have enough other swept filters that making this one up/down sweep just seems like a lot of extra work.  The centre-piece is, after all, the S&H, and sweep direction makes no real difference there.

I found the biggest challenge is really getting a S&H sound that isn't too high contrast.  Filter-frequency changes that are too high contrast can leave you with no audible middle.  Which is why nailing the trimpot settings is so important.

[Mark Hammer]

While I can't imagine it adds much to the S&H aspect, it strikes me that the autowah side is a bandpass filter, stock, but could be easily converted to lowpass.  has anyone ever done that, and if so, how did the unit sound?

[cortezthekiller]

I've built one (Tonepad layout) and a few useful mods I've found were:

-Blend control= Use a 500K Log pot with one outside lug connected to the 10k/100k junction after the input cap. The other outside lug is connected to the output side of the 0.15 output cap, and the wiper goes to the output (footswitch).

-Sample/Hold range control= Replace the 100K trimpot for a 50K (better control of range) panel pot for tweaking what frequencies the sample/hold filters. Can go really bassy or high. Fun.

-Self oscillation. This sound works best with a heavy distortion after the FSH (big muff is excellent). Use a spst switch to connect the input to the pedal with the -9volt supply. Not sure how safe this is, though it hasn't exploded yet, and in conjunction with the S/H range control can create interesting synth-drum-like textures.


Having different filter ranges would be fun too. Also interested to see ideas on that.

[Mark Hammer]

Hi Dave.  Nice to see you here.  Have you tried to add some portamento?  That wouldn't do very much for fast clock speeds, but it might make some of the slowest speeds more interesting.

[cortezthekiller]

That sounds awesome. How would it be achieved?

[Mark Hammer]

Well, I just tried the idea out.  It works, and it sounds interesting.  

There is a 47nf storage cap that holds the sampled voltage.  It goes between the gate of the JFET labelled as Q6 and V-.  I just tacked on a resistor between the gate of that JFET and ground.  That's it.  

Remember that the two JFETs - Q5 and Q6 - effectively isolate that storage cap by providing ultra-high-impedance paths to anywhere that the cap might conceivably drain off.  The cap is small enough to charge and discharge quickly, so that a stairstep function can be achieved with instant retuning of the filter's centre-frequency.  The resistance I'm adding provides a lower-impedance path for that storage cap to drain off before the next sample.

Rather than providing a portamento-like glide from one centre-frequency to another, as I'd hoped, it has a different effect, equally musical IMO.  Essentially you get a retuning of the filter, but a quick downward sweep from each new centre-frequency setting as the cap drains off.  So far I have tried it out with a 1M and a 390k resistor to ground.  From the sounds of it, going above 2M2 likely wouldn't yeild very much of interest except at the slowest clock speeds, and going much below 220k would not get anything pleasing.  But that could be me.

In the absence of a sound sample, I'll just note that instead of going ah-ee-oh-eye-oooh, it goes ow-eeewww-eyeoo, etc.  That is each new setting has a descent tacked on.  Neat!

Note that since it would be difficult to score the right sort of pot to make it work in stock form simply by turning the pot,  I'd suggest implementing it with either a rotary switch (6 position with one being no glide ought to do it), or else a 3-position toggle for stock and two different "swoop" settings.  If only two swoop settings, you might use 1M and 270k.  If a 6-position switch, then you might use 2M2, 1M, 470k, 220k, and 100k, the latter being more useful at fast clock settings.

Actually, you know what?  Get a damn 2P6T rotary and use one set of contacts to select a glide-resistor value for the S&H, and another to select different decay-resistor values for the envelope follower section.  That way you have 6 different decay times for the autowah, and 5 different glide settings for the S&H.

It's a worthy mod that adds an additional feel to the unit.  I'm glad I pursued it.

[cortezthekiller]

I tried that out and it's quite interesting. Removing the .047 cap altogether also produces some interesting sounds.
Kind of more laser beam-like filter sweeps.

The self oscillation mod I described above actually works much better by connecting the input signal to the 22k/47k/+input junction on the LFO op amp.
It gets quite loud and speeds up dramatically. Using the input grounding bypass method would be good for this if its left on when the pedal is off (LFO bleed).

[Mark Hammer]

Okay, postscript.  I installed an on-off-on DPDT toggle switch to provide three decay-time settings for the envelope-follower portion, and stock plus two glide settings for the S&H section.

For the envelope follower I put a 120k (instead of 100k) on the board, and used one set of toggle contacts to switch in 100k or 68k in parallel.  Not a huge range of decay times, but audibly different.  The S&H section uses either nothing or a 1M or 220k to ground.

Note that the glide-resistor value interacts with the clock rate.  If the clock rate is slow, then the drain of the storage cap needs to be slower, or else it drains so fast you don't hear much variation in filter tuning.  So the 1M resistor is better for slow settings, and the 220k better for fast settings.

But it works like a charm.  I have a sample on Soundcloud that is currently processing.  I'll post the link when its done.

[Mark Hammer]

here goes.  My first time using soundcloud: https://soundcloud.com/mhammer1/s-and-h-sample

Its the P90 bridge pickup of my Coronet into a ZVex Octane clone (essentially a modded Superfuzz), and into the S&H, set for slightly higher resonance and several different variations of clock rate and glide.

If it seems to hang onto a single pitch, that's because all you are hearing is a single open A string, sustaining because of the fuzz.  It's not the result of filter resonance or anything like that.

[Morocotopo]

Nice sounds Mark. How´s the thing in respect to ticking? I never could get the ticking out of my Tonepad build in S/H mode. Part of it is just the filter stepping, that with a high resonance makes a "tock" with each transition, but I always suspected that it could be a bit more subdued. Always thought that the difference was the use of 136000´s versus 3080´s, not that I could give any real explanation, but...

[alparent]

Oh no! I feel a new build is about to be added to my list.
I would build the Tonepad version (I have 13700s but no 3080s)

I'll etch the board and read/re-read these mods.

[Mark Hammer]

Nice sounds Mark. How´s the thing in respect to ticking? I never could get the ticking out of my Tonepad build in S/H mode. Part of it is just the filter stepping, that with a high resonance makes a "tock" with each transition, but I always suspected that it could be a bit more subdued. Always thought that the difference was the use of 136000´s versus 3080´s, not that I could give any real explanation, but...

I would think that the ticking is just like any other LFO-based ticking.  So, be sure to decouple IC3 in the Tonepad layout, and be sure to use a low-current op-amp, like a 358 or TL022.  (Though something tells me that since the clock is simply switching the JFET on and off, one might borrow some tricks from the JFET switching circuitry, and slow down the turn-on of that JFET juuuust a little bit with a cap to ground between the clock and JFET gate.)  Having used the GGG layout, I used 741s, which restricts my use of dual op-amps (   ).  And quite honestly, I've been so preoccupied with mods that I hadn't paid much attention to audible ticking.

The other thing I tried before going to bed was attempting a lowpass-filter output instead of the stock bandpass.  It's also a pretty simple change,  The 3k3 resistor on the final output takes its feed from the JFET between the two OTAs.  If you cut that path and connect the 3k3 to the output of the other OTA, you have a lowpass output, apparently.  At least that's the way it worked on the DOD FX-25.  I will note that it doesn't sound all that different from the bandpass (using a guitar through a 6" speaker), and also forfeits some volume, so I'm on the fence about including it.

If the cap discharge mod catchesd on, I think we should call it "droop", rather than "glide".  "Glide" generally refers to a smooth upward or downward transition.  This mod just makes the sampled voltage give up, so "droop" fits better.  Finding out how to add glide would still be a nice thing to achieve, however I suspect it would involve adding some buffer circuitry to permit slowing down the chargeup of the .047 cap, and that complicates an already complicated board.

If one hasn't already attempted this, you're better off making the Tonepad layout.  Not just because of the cost/availability of 3080 chips, but it will actually fit into a standard-sized box, while the GGG layout will not.  The Tonepad layout also includes up/down sweep for the envelope part.

If you include all the mods, that starts to turn into a pretty complicated pedal, with a lot of knobs and switches.  I would suggest something like a single 3P4T rotary switch to adjust resistances and produce changes in attack, decay, and droop simultaneously.  That way you have Sensitivity and Rate controls, switches for up/down sweep, an Envelope/S&H switch (which could be a stompswitch), either a toggle for a few resonance settings or a full variable Resonance pot, and a master-"Time" rotary switch that controls attack/decay and droop.  I suspect for most players, a two-attack-by-two-decay-time arrangement (4 settings), similar to the regen/width combinations switch on the MXR Phase 100, would be adequate.  The 3rd set of contacts in the rotary switch is then used to have either stock S&H, or 3 degrees/rates of droop, one of which will be more suitable to your clock rate.  So that could be a package as compact as 2 pots, one rotary, and 3 toggles - a ton of flexibility in a 1590BB.

[Mark Hammer]

Ignore the part in blue i the above post.  Sitting at a red light today, it dawned on me that the part I wrote in blue is utter nonsense.  The ticking comes from the current draw that comes from producing a square wave, and not likely from the time it takes to turn on the JFET.  I mean, it might contribute, I suppose, but LFO-based ticking tends to be primarily from spikes across the supply line shared with other chips, and that would not be addressed by making the turn-on time of the switched JFET just a tad less instantaneous.

[Mark Hammer]

I was working on somethng else yesterday, and had occasion to look over some analog synth circuits that incuded a glide/portamento option.  Typically, that involves the addition of a high input-impedance buffer after the "captured" keyboard voltage, with a cap to ground on its input, and a variable resistance in series with the input.  The variable resistance sets how long it takes the cap to charge up to the next voltage.

I thought about adding a little daughterboard for that.  Then it occurred to me, that we already have a high-impedance buffer in the FSH-1, except that it is used to hang onto the voltage sampled from the noise generator.  In this diagram, it is Q6.  If one cuts the trace that connects Q5 to C9, and inserts a resistance, you have essentially the same thing - a cap charging up a little more slowly.

Since I'm fond of such things, a simple way to implement it would be to bridge that cut trace with a suitable resistance value.  Estimating the average sampled voltage to be 1V, we'll say 470k, to give a cap charge-up time of roughly 22msec.  A 3-position on-off-on toggle has the common luh connected to the junction of that resistor and C9 and the gate of Q6.  Flick it to one outside position and we bridge that 470k resistor with a piece of wire so that it replicates the stock immediate-charge-up condition.  Flicking the switch in the other direction, and we bridge that 470k with a 390k fixed resistor, yielding a combined parallel resistance of 193k, and an estimated charge-up time of around 9msec.

This is all in theory, of course, but then so was the "droop" mod.  Having a portamento option, and a droop option should provide a bunch of different feels to this thing.  I have no idea how functional a combined portamento and droop would be...but I look forward to trying.

[Mark Hammer]

Update.

I thought this suggested mod would work, but it didn't.  I went as high as 3M3 and couldn't hear a damn bit of audible difference.  So don't waste your time on what I suggested.  I think if there IS going to be any possibility for a gliding change between filter frequencies, it will have to involve an additional circuit.

[aron]

[alexradium]

Oh no! I feel a new build is about to be added to my list.
I would build the Tonepad version (I have 13700s but no 3080s)

I'll etch the board and read/re-read these mods.

take care with the tonepads layout, it works but follow the schematic to place your jfets ,I used bf245 and I had to flip them 180. Plus one of them drain and source swapped.

[Mark Hammer]

I always feel a sense of dread when working with JFETs, for that very reason.  Lousy pinout diagrams on datasheets (Is that looking down from the top, or looking up from the bottom?) don't help, either.

Do yourself a favour, kids.  Even if you think you already have a datasheet for a given JFET, get all the datasheets from anyone else who makes that same number transistor - ON, Fairchild, NatSem, Toshiba, etc. - just to increase the odds you'll have clear and unambiguous pinout information.