Maestro FSH-1 (Filter Sample/Hold) mods

[funkiness]

Hi all,

I have just received the pcb (from GGG) and I am wondering what are the mods you can add to this effect (other that tonda's and moosapotamus ones) ?
And is it easy to change the filter to make a low pass ?

Unfortunately I am almost a newbie, the only effect  I have already "build" is the PT80 delay (which sounds great !!!).

Thanks for your answers !

funkiness

[funkiness]

So no one has tried some others mods ?

[Mark Hammer]

I have an etched and populated board sitting in a storage box with no wires attached, so I can't offer anything based on experience.

The mods that Tonda and Charlie have offered, address some of the more common functional controls of an envelope-controlled filter, namely envelope parameters and some of the more classic filter parameters.

If your intention is to get a little wacky, then the S&H aspect provides some fertile ground.  Keep in mind that the S&H aspect uses a noise source as a source of random voltages.  A clocked FET samples that randomly-changing voltage source, latches onto a voltage, and holds it until directed by the clock to do it again.  The range of random voltages available to be sampled is a function of the signal provided to the sampling circuitry.  So, the amplitude or bandwidth properties of the "random" source could be varied.  Hell, it doesn't have to be noise either.  It could be an external source like a radio. For that matter, you could even feed the sampling circuitry the envelope signal.  The S&H circuitry would then "stairstep-ify" the envelope-driven changes.  (I think some folks refer to this as "zippering").  The conversion is pretty simple.  You can see in the schematic that the envelope signal comes out of pin 6 of a 741 (or equivalent).  So does the noise signal.  Absolutely no reason I can think of why you couldn't reroute the output of the envelope follower to the input of the 10k trimpot that normally attenuates the noise source output.

Another aspect is what might be described as a portamento control.  Typically, the purpose of a S&H is to hold the sampled voltage in a stable form.  It is the stability of the sampled voltage and immunity from leaking that provides the step-like action.  If there is a path for the sampled signal to leak, though, such that it is held imperfectly, other sorts of sonic textures can be achieved that are sort of stairstep-like in nature, but each step has a bit of sloppiness attached to it.

C9 (.047uf) is the cap that stores the charge.  Because of the high impedance paths provided by Q5 and Q6, the charge has nowhere to go (or at least nowhere it can go easily) so it remains at the level it was sampled for quite some time.  Were that cap to have a path to drain, however, the filter would sound as if it was "falling" with each step, just falling from different heights.  I'm not the guru on such matters, but for the moment, try sticking a 1-2.2M resistor in parallel with that cap, and tell us if it does anything interesting.  Incidentally, I suspect a small amount of droop/slew is already inserted into the device, via C7/R20, so as to not make the change from step to step quite so jarring.  The change I am suggesting would make the droop more obvious and change the character of the sweep in a more obvious way.

As for the filter, it actually IS a lowpass filter already.  The thing about lowpass filters (in fact the very NICE thing about them) is that they sort of sound like bandpass filters when you add a bit of regen to them, since the regen accentuates the zone around the corner frequency.  You can alter the range of the filter, though, by changing the value of the caps that set the range - C3 and C5.

[soggybag]

Those are some great suggestions. The sample and Hold sort of bridges the gap betweeen effect pedal and modular synth, which is something I have always liked about this one.

As I read this I thought you might try feeding the clock signal through the envelope generator. To get an envelope sweep that happened on timed cycle rather than an envelope based on the input signal.

Of course on my FSH-1 there is a problem with the enevlope it has never swept correctly. Maybe I will have to take another look and see about if I can fix it.

[Mark Hammer]

As I read this I thought you might try feeding the clock signal through the envelope generator. To get an envelope sweep that happened on timed cycle rather than an envelope based on the input signal.

Not clear what you mean here.  Are you talking about using the clock pulse that would normally hit the gate of the FET (Q5) as a kind of impulse or trigger?  This would be periodic but I suppose a bit more like a kind of descending exponential sawtooth than like any sort of LFO waveform.

I might point out as well, that as long as you provide a reasonable point of entry, ANYTHING can be used to drive the envelope follower.  That could be a mono mix of the drummer in one's band, a tape recoding of a conversation with your preschool child, an AM radio tuned to the most outrageous sports talk show you can find, or whatever the heck you want, whether periodic or aperiodic.

The thing to keep in mind is that the predictability of sweeps, rate of change in sweeps, and extent of sweeps are jointly considered when determining the pleasantness of sweeps.  Huh?  :shock:

What I mean is that if the filter isn't being swept very wide, and if the next sweep in line doesn't come too fast and furious, the unpredictability of the sweeps isn't a problem.  If the next sweep stands a chance of being wild, then you won't want the next one and the next one to come too soon.  And so on.  It's a question of striking a balance between the degree of unpredictability that is attractive but not demanding, and the degree of predictability that is not demanding but easily boring.  

Ultimately, the filter sweeps are in service of music.  If the music itself is highly unstructured (think "modern electronic music"), then the sweps can afford to be complex, unpredictable and jarring.  If the sweeps are supposed to fit into a more organized framework (i.e., a song) then one needs to constrain how far the effect wanders in order to make it usable and useful.

[soggybag]

I was thinking the LFO which normally feeds the S&H could be used to trigger the envelope generator.

[Mark Hammer]

Ah....

In that case, it is worth considering how long the duty cycle on the clock is.  It will certainly be fairly brief, because the whole point is to turn the gate on for an instant.  Long enough to have something to sample, without being on so long that averaging out all noise yields the same voltage every single sampling period.  Think of it in statistical terms.  If I took a poll of 100 voters prior to an election, chances are pretty good that the average response would change with every single poll.  If I took a poll of 1,000,000 voters, the chances are very slim the overall result would not change over polls despite how much noise and variation of opinion is built into each poll.  Transposed to this scenario, a sampling period of 1msec may insure that just about every sample is different from the last, but a sampling period of 100msec may not.

That may well mean that the clock pulse doesn't last long enough to provide a viable trigger or gate pulse for the envelope follower.  That, in turn, implies that you either need to find a way to extend the duration (duty cycle) of the clock pulse, or find a way to use the clock pulse to trigger an actual envelope generator.  The latter may be attractive as a way of deliberately shaping the sweep in interesting ways.  The former may be simpler than I think it is.  Personally, I am unfamiliar with the workings of the op-amp based oscillator used for the clock in the FSH-1.  It might be as easy as changing the value of R32 or something like that (Am I at least close, guys?).

It is also worth considering that you may want to extend the decay time of the envelope follower if the transient you are feeding it has a very short half-life (and a pulse would obviously be much briefer and punctate than a plucked note).  Here, C6 may need to be bumped up substantially from the current/stock value of 4.7uf.

[Ry]

This is a great thread!  I've been fooling with my FSH since I built it a few years back.  What I don't like about it is that the noise source is too predictable.  It does goes to the low frequencies, then back to high at a steady rate, rather than doing the strange blips and burps that I would expect it to.  The faster I set the speed, the more predictable it is, at the fastest, it is completely rythmic.  At the medium speed settings, it has some randomness, but tends to latch up (continue to repeat) on the bass side of things.

I've tried numerous transistors as well as hours spent tweaking the trimmers.  I have been concidering building a digital noise source to replace the transistor, but that seems like a lot of work.

Ry

[soggybag]

Ry, I think the power supply or other LFO is leaking into the noise source. I notice a similiar situation in mine though not as bad as what you have described. I looked at the noise pattern on the scope and I can see what looks like a noise but it rises and falls over time.

http://webdevils.com/stompbox/IMG_2851.JPG

Here's a link to a picture of what I see on the scope. The fuzzy line is the noise drain of Q5. The other brighter line superimposed on top is coming from the emitter of Q3. This second wave form shouldn't be there. I'm not sure where it's coming from. My FS-1 has a problem, the envelope has never worked. This wave seems to be what I get instead of the sweep that should modulate filter in the filter mode.

The other wave form seems to be leaking into the noise source. Making the noise rise and fall over time. The noise is still random but over time you get a feel for this other wave that is modulating it.

I'm guessing you may have similar problem.

[soggybag]

After a few tests it appear that my power supply is the problem. It should have been obvious since I waas using a cheap RS wallwart. Here's a picture of +9v where it enters the board.
http://webdevils.com/stompbox/IMG_2853.JPG

[Ry]

I notice a similiar situation in mine though not as bad as what you have described. I looked at the noise pattern on the scope and I can see what looks like a noise but it rises and falls over time.

Yikes!  That's a lot of noise in that trace.  I built an ulta clean power supply (I think the schematic is at GGG) and used a MAX1044 to get -9v from it.  I will scope it and the noise source to see if I can find any correlation between the two...now that I think about it, it could possibly be the 1044 chip.

Thanks for the tip, I wouldn't have thought to trace it back to the power supply.

Ry

[soggybag]

I used the layout from ustomp.com to build mine. It has a MAX1044 on the PCB for the -9V supply. When I trace the power from the MAX1044 the ripple is not as bad.

[Ry]

I'm thinking out loud here, but I wonder if the current draw of the pedal and Max1044 is enough to send the power supply into oscillation by demanding more than the 9v regulator can gracefully supply.  The limit is supposed to be 100ma on a 78L09.  I find it hard to believe that the circuit would pull that much current.

Have you taken a current reading?  I'm in the middle of building Joe Gagan's Decade, but I will check this out in my pedal when I'm done (hopefully this weekend).

[funkiness]

Thanks Mark and the others for your answers !

[ExpAnonColin]

Some excellent suggestions.  Don't forget trying normal boring old LFOs to the input of R22.  I think I am going to breadboard this S&H circuit and see if I can do anything fun with it.

-Colin

[funkiness]

Hi all,

I have just finished the effect this week-end... unfortunately it is not working.  :cry:

Bypass is okay but there is no sound at all in filter mode and in s/h mode.

I have done a lot of checks and everything looks fine. Maybe the pb is caused by the parts is use ???

Here they are:

Q1: 2N5088
Q2: F245C
Q3: 2N3906

IC1: 3080T (i have tried 3080E too)
IC2: 3080T (i have tried 3080E too)

Schematic:
http://www.generalguitargadgets.com/diagrams/fsh1asc.gif

[Mark Hammer]

Unfortunately, "bypass" is not a very unique sound.

It is possible that only one part of the circuit is not working.  What you have to do now is identify which parts ARE working.

For example, does the envelope follower work?  Does the LFO work?  Does the noise generator work?  If you can eliminate those parts of the circuit as being a problem, that will help you to identify what MIGHT be the problem.

[funkiness]

Unfortunately, "bypass" is not a very unique sound.

Lol !

It is possible that only one part of the circuit is not working.  What you have to do now is identify which parts ARE working.

Ok.
Filter mode alone (without s/h) is not working so I will focus on filter and envelope parts.
For tonight checks I will try to follow the audio signal path and find the point where I lost it (sorry for my bad english...).

[Mark Hammer]

Your English is good enough, but your solder joints might NOT be.  Make sure to check them too.

[funkiness]

Your English is good enough, but your solder joints might NOT be.  Make sure to check them too.

After some tests, here what I found:

1) about the S/H:
_ noise generator works (a lot of noise at IC4 pin 6)
_ clock part works (i can hear pulses at IC5 pin 6)

2) about the filter:
_ envelope generator works
_ sound arrives to IC1 pin 2 (input of the first 3080)
_ but NO sound at the output of IC1 (pin 6)

A strange behavior for Q2: near +8V for each pins...

I used F245C for Q2.
I dont know how FET works but I think it should be the bad part of the circuit...

What do you think ?