Has anyone successfully recreated the Roland AS-1 Sustain as per RG's circuit?

[stryker]

Hi,

A mate has a real one, has had it since new and loves the effect, but it's physically huge, heavy, and it doesn't take external power.  So we thought we'd take on the project and recreate it in a smaller enclosure, based on RG Keen's circuit.  We've had a couple of attempts with a test rig, trying both cheap chinese vactrols and one we rolled ourselves.  In the first we used an external voltage boost module but after measuring it needs under 30mA I've added an ICL7660 to double the voltage so it can run off his 9V pedal board supply for this latest test rig. 

But reading existing posts in this forum and elsewhere we seem to have got to a similar place, with the output oscillating.  Using PRR's advice in this post we've tamed that noise, but no guitar signal gets through to the output regardless the vactrol we've tried so far.

I'm wondering if this one is actually achievable with 2019 parts, and if anyone has made it work what substitute they came to for the vactrol part of the circuit?

Thanks
Geoff

[Mark Hammer]

1) Such simple oversights are not uncommon to any of us here: check the pinouts of all transistors used.

2) Assuming all pins are present and accounted for, verify that the LED is doing what it should in response to input transients.  I'm in the process of making a bunch of LED/LDR units for myself and it is all too easy to mistake anode and cathode once those suckers have heat shrink over them.

3) Assuming the LED is doing what it should, verify that its activity results in a suitable change in LDR resistance.

[vortex]

Seeing that you have an original...

Why not document with some photos, parts id, voltage measurements etc.

A bit of give and take...

[stryker]

Thanks guys this gives hope. 

I found errors in my first board, which I optimistically created with SMD passives aiming for the smallest PCB to create a micro pedal.  The 2nd rig uses all throughhole, to be closer to the original.   Both test rigs were ordered as PCBs designed in KiCad which means any mistakes in the footprints will be common to both so I will cross check the pins again.

I will have the original back here within a week or so and was planning to compare measurements at similar points in the circuit, but did want to be sure this wasn't a dead end idea.  I have DMMs and a scope so if there's any particular areas you think it would be worthy to document from the original please let me know.

[stryker]

Returning to this project after the real world intervened...

I've verified the circuit, the transistor footprints and connections, the vactrol led polarity and that I have the correct recommended substitute transistors where they belong.

Sadly it all checks out so far.

Should I try to roll my own vactrol or is there a preferred Chinese alternative that's recommended?

[bartimaeus]

Are you sure that you didn't accidentally kill your current vactrols when you were testing it out initially?

Perhaps best way to find a suitable replacement vactrol would be to measure the range of voltages that the pedal generates in normal use, and find vactrols that suits that range

[Eb7+9]

Should I try to roll my own ?!

no need ...

seems obvious to me this high-gain circuit is expecting an LDR with low "off" resistance ... Keen usefully points out 70ohm "on" resistance, which is super low as far as LDRs go ... and since on/off values tend to trend, Roff should be lowish as well ...

since the LED sees no DC at rest and only starts firing up past a given threshold means we can separate the two limits embodied by the original cell ... as well, it's safe to say we don't need cells that go below 100 ohms at extreme squash - so, in principle, many modern optos should not be a problem here if that's the route taken (the NSL-32 or the faster NSL32-SR3 opto's would be good candidates to try)

two things to look at ...

1. simulate a low Roff idling value by sticking a 1M pot across the LDR and adjust to need

2. if your pre-rolls/opto's are acting aggressive and moving too fast you can always curb their potency by increasing the resistance (100r) at the emitter of Q4 ... in fact, once made variable, that resistor acts as a rough "ratio" control ...

self oscillation is a partly a product of gain, in this case open-loop gain ... the 1M pot is meant to drop max loop-gain at idle

good luck

[stryker]

Eb7+9 there's a lot for me to unpack there but you've given me a heap to work with.  Thanks!

[R.G.]

It's worth noting the copyright date on that schematic - I traced the only Roland Sustainer I ever actually touched, back in 1999, twenty years ago. Hmmm. Maybe that qualifies for a Sergeant Pepper's moment. 

At that time a friend was running a used-guitars shop and he would let me borrow pedals from the display case to trace. The AS-1 was one of them. I've had reports of successful builds from the schematic although none in the last decade that I remember.

Oscillation is gain-phase problem. The Nyquist criterion says that anything that has a gain of over unity at a phase shift giving 0 or a multiple of 360 degrees back at the input where the feedback goes will oscillate. This particular circuit gets tricky because there are two feedback loops to deal with. One is the emitter of Q2 back to the base of Q1. The second is the variable gain loop with the LDR, from the collector of Q1 back to the emitter of Q1. Q1 is acting like a differential amplifier, with the positive input at its base and its negative input at its emitter. The amplifier then has voltage gain of Q1 and Q2 cascaded for its open loop gain. The Q2E-Q1B loop also sets the DC bias conditions. The Q1/Q2 setup is a variant of "operational amplifier" that was widely used before opamps were affordable.

From visual inspection of the circuit, I think that the Q2E to Q1B loop is setting the max gain of the inner amplifier, with dominant pole compensation by that 47pf cap on Q2. That leaves the outer loop with the LDR to lower the gain further for compression.

The threshold for the beginning of compression is when the signal voltage at Q2's collector exceeds the Vbe turn on for the Q3-Q4 discrete darlington. This will probably be on the order of 0.9-1.0V. When the signal peaks hit this level, the darlington is turned on and the LED gives a pulse. The LDR gets the pulse of light and due to its slow response, "integrates" the pulses into a general lower resistance.

Given that you are having problems with the thing oscillating, if I were messing with it I would temporarily replace the LDR with a pot and put a scope on the input and outputs. Then I could mess with the pot and see whether ANY value of pot setting from a few ohms up to a meg or so would start and/or stop oscillation. If no setting makes it stop, the LDR is not the problem. This also lets you measure the necessary LDR resistance specs for operation by seeing what X pot setting does to the gain.

If all pot settings leave it oscillating, you can go to work on the inner feedback loop to make it stable. Dominant pole feedback, as in that 47pF cap, is usually a way of forcing the open loop high frequency response to drop at a single pole rate until the high frequency gain is below unity before the high frequency phase shift hits 0 or 180 degrees. If the inner loop is oscillating all by itself, I'd first make sure the power supply and ground and bypassing and signal path wire routing are pristine. These are not things that can go on schematics - they're heavily layout dependent. Then I'd try slugging the high frequency response with a bigger value than 47pF - maybe up to 1000pf. That may stop the oscillation. If so, you can test the audio response for "OK enough".

You didn't mention what frequency the oscillation is. If it's audio, that's one clue, if it's RF, that's a different issue. If it's super low frequency, that's another syndrome as well. Let us know what you find and we can go from there.

[stryker]

Thanks RG. I'll make up another board as you describe and report back.

You didn't mention what frequency the oscillation is. If it's audio, that's one clue, if it's RF, that's a different issue. If it's super low frequency, that's another syndrome as well. Let us know what you find and we can go from there.

It's audio.  It's been many months since I had it on the scope so I won't pretend I remember the exact number, but I'll measure again and report here.
Merry Xmas everyone!