Info wanted: Yamaha GC-100

[Mark Hammer]

I picked up one of these today, 2nd hand.  http://www.effectsdatabase.com/model/yamaha/100/gc100

The pedal packs a 3-control compressor and 6-band EQ into one pedal.  I haven't brought it home to take apart yet, but was wondering if there was any technical info on it, especially a schematic.  Curious about whether the two functions are easily separated/defeated.

[Mark Hammer]

Well, partly dissected.  Switching is electronic, via a 4016 quad CMOS switch.  The EQ section uses a Rohm BA3812L 5-band graphic equalizer chip, and a TA75559 dual op-amp.  The compressor section uses a trio of 4559 dual op-amp SIP chips, and a Hamamatsu P873-13 dual optoisolator, which probably accounts for why it seems to be a pretty quiet unit.

On the surface, it seems like the compressor precedes the EQ.  I'll need to trace things from the compressor Level control output to the EQ-section input to determine if the two section might be independent selected/cancelled.

[digi2t]

Hamamatsu P873-13 dual optoisolator

Woah.... isn't that what was in the Mutron III?

[Mark Hammer]

Perhaps.  I know that when I searched for info about that optoisolator, some of the hits that came up were about the Q-Tron.

But it's made me curious: just what would a compressor do with two LDRs?

[RickL]

Sorry to resurrect a dead thread, but I just got one of these too. Did you ever find out anything more about it? I almost didn't get it because the compression seems so subtle and the attach control seems to barely have an effect. Since it uses a positive tip adaptor I wonder if any damage could have been caused by someone connecting a standard negative tip adaptor.

It does have a fair amount of gain (or at least volume increase) so it's at least partly working.

[Mark Hammer]

Hi Rick, Phoenix-affected retiree Mark here.
Jeez, I'd forgotten about that pedal.  It got moved to the back of the shelf and disappeared into the shadows back there.  I'll have to dig it out again.  As I recall, the "Attack" control worked fine.  But like all such labelled controls, the impact is not all that noticeable unless you pick fast, since it dictates gain-recovery speed.  You would not be the first person to twiddle such a control and mutter "Jeez does this even do anything?".

I'll try looking inside mine for clues later in the week.

[RickL]

Yeah, I finished an acting position and lost almost 30 years of accumulated sick leave. I've got it back now but what a pain 

It turns out that this compressor actually wants something to compress. Single coils with the guitar volume at about 5 aren't enough to get to the pre-set threshold. Volume at 10 and now the Attack slider works and I can actually hear it compressing.

It also turns out that one shouldn't assume things. The label is missing from the back of the pedal and I figured "old Yamaha, must be centre-positive". After I replaced the open protection diode that I probably blew, it works fine with a regular Boss centre-negative adaptor.

[Rob Strand]

Go here, click on the images until you get the largest pic,
https://www.ebay.com/itm/293006270219

For those skilled in the art like yourself there's enough to work on.

It's a feed-forward side chain so you probably only need to increase the gain of the first section in the detector.  Maybe tweak the time constants.

[Mark Hammer]

A couple of observations:

1)  Thanks, Rob.  Very helpful, and much appreciated.

2) Partly what I expected and partly not.  I knew there was an equalizer chip in there and an optoisolator.  The opto is a Hamamatsu dual.  One half is used oretty much the way that Craig Anderton uses it in the EPFM compressor (and Jack/Tom use it in the Whisper), but the other half is used to attenuate the input to the LED driver.  Weird.  An unusual arrangement.

3) Unlike the vast majority of stompbox compressors, the Attack control actually IS one!  The sidechain also appears to use a different sort of full-wave rectifier than one normally sees.

4) If sensitivity to single coils is a problem, it would seem that the simplest fix is to tack a second resistor in parallel with the existing input resistor in that first gain stage, the one labelled pre-amp after the input buffer stage.  It's a reasonably straightforward inverting amp, so calculating the change in min/max gain should be pretty easy.

(As a completely off-topic aside, Rick and I were both affected by a change to the payroll software application that was implemented - quite prematurely - in early 2016.  Briefly, the Canadian federal government had been needing to harmonize HR information systems for years, since the various federal organizations tended to use completely different and incompatible systems.  It could often take months for information about things like accumulated sick leave and vacation time to be forwarded to another department/agency, if one changed jobs and jumped organizations, preventing new staff from being able to take vacation or sick leave.  Unfortunately, the hasty introduction of the new system-wide payroll system followed hot on the heels of laying off a great many compensation advisors - to "save money" - who were familiar with the twists and turns of their agency's various collective agreements.  Some job groups are unique to a specific organization, like border guards, prison guards, lighthouse keepers, meteorologists, et al.  The people tasked with getting the system running were not well-trained, and the erroneous data management resulted in tens of thousands either being underpayed or overpaid, or not being paid at all.  Most affected were people who changed jobs or were recently hired.  Universities had to set up emergency funds for students who had worked the entire summer without pay and were unable to pay their tuition or living costs.  Some permanent employees found themselves missing mortgage/car payments and in jeopardy of losing houses/cars, so emergency funds were set up for them as well; all of which will require gobs of accounting fixes to rectify.  It has been a mess, and for something that was supposed to save hundreds of millions a year has ended up costing several billion so far.  I was unaffected until I retired, which is a change in status, and Rick was unaffected until he subbed for someone in a higher par scale - a temporary change in status.  Eighteen months post-retirement, I am still awaiting a five-figure payout for unused vacation and other things, that was due.  One of my former co-workers has now waited two whole years.  Don't cry for us, Argentina.)

[Rob Strand]

3) Unlike the vast majority of stompbox compressors, the Attack control actually IS one!  The sidechain also appears to use a different sort of full-wave rectifier than one normally sees.

The rectifier is a linear one (in that the output magnitude is linear with the signal).   The Dynacomp rectifier is exponential.   The Yamaha is sidechain Feed-forward whereas the Dynacomp is Feed-back.  The different rectifiers produce similar final results in each case (without getting caught-up in too many finer points).  You will see that rectifier on a lot of the later era Limiters like Boss LM-2 and LMB-2.

One thing about the Feed-forward is the RC time constants match-up with the Attack/Release time constants.   The Feedback types tend to have a Faster Attack time constant and a slower Release time constant than the RC time constants imply; see the NE570/NE571/NE572 app notes.   In other words you need different RC values to get the same final Attack/release behaviour.

[Mark Hammer]

As I understand it, feed-forward is more for limiting, where feedback is more for compression.  That is, feed-forward tames the peaks but doesn't do that much for soft passages, where feedback tends to dynamics at each end of the range, yielding a more constant level.  Now that you mention it, I can easily spot the feed-forward aspect.  What I don't quite understand is how the Dynacomp circuit - used in so many other compressors -  is exponential, while this circuit is linear, as well as what difference that would make.  I need more insight into this.  And is the linear aspect a result of the combined use of the two LDR halves in different parts of the circuit?

[PRR]

At some level of sophistication, FF and FB are equivalent. No difference "compression" or "limiting" (which are the same thing to different degrees).

The other thing you can do is Expansion. Noise Gate.

Without infinite gain and perfect gain laws: in compression/limiting, FF does not correct gain-device nonlinearity, FB tends to. However FB can not make "infinite" (brick-wall) limiting because it is killing its own error signal.

In Noise Gate, FB kinda don't work, FF is universal. That is, if the default is NO signal passing, and the control comes after that, how can it know when to open-up? Even if the gate is not to-nothing, it is reducing the signal needed to open-up, and that is usually annoyingly small already.

[Rob Strand]

This whole topic is quite complicated so I can only show the cases by using some examples.

As I understand it, feed-forward is more for limiting, where feedback is more for compression.  That is, feed-forward tames the peaks but doesn't do that much for soft passages, where feedback tends to dynamics at each end of the range, yielding a more constant level.

In practice that's kind of how it pans out.   However, feedback doesn't necessarily imply a compressor.   It is possible to create a limiter with a feedback structure.  For example, see figure 12 of,
http://experimentalistsanonymous.com/diy/Datasheets/SA571%20AN.pdf

You can see the detector is *very* non linear it is either on when the output voltage is above the threshold or off when it is under (the detector here is "full wave" in that detects both positive and negative peaks).   It is a hard limit with an infinite compression ratio.

Technically all feed-forward mean is the detector/rectifier is controlled by the input voltage, and feed-back means the detector/rectifier is controlled by the output voltage.

So if if you now look at a feedback structure, for example figure 17 of,
http://experimentalistsanonymous.com/diy/Datasheets/SA571%20AN.pdf

You can look at this as a simple inverting amplifier with a variable feedback resistor,

where,
    Rf = The gain cell
    Ri = 20k  (the Chip's R3 resistor)
    A = overall gain = Rf / Ri   ; ignore sign for convenience.

For the NE570 (with chip resistors R1, R2, R3) the gain cell looks like a resistor with value,
    Rg = 22k / VG_pk = 20k  where VG_pk is the peak voltage at the input to R1.

You can see the rectifier goes to the input (as required by feedforward) so the rectifier input VG_pk is the input signal Vin_pk.

For the opamp example the connection are such that Rf  = Rg and Rin = R3.

So look what happens with that feed-forward connection as the input signal changes,
At Vin = 1.1V peak , Rf  = 20k,  A = 1,     Vout = 1 x 1.1V pk = 1.1V pk
At Vin = 2.2V peak,  Rf = 10k,   A = 0.5,  Vout = 0.5 * 2.2V pk = 1.1V pk.
At Vin = 0.55V peak, Rf = 40k, A = 2.0,  Vout   =2.0 * 0.55V pk = 1.1V pk

So what is going on here is if we increase the input voltage the gain cell resistance is reduced (as it is controlled by the input voltage) and the gain is adjusted in such a way that the output remains constant.  It achieves this by nature, we don't need to monitor the output like the feedback case.

The natural limiting behaviour of feedforward is why you see this used for limiters.

The example is for an ALC not a limiter.  The difference is as the input voltage gets very small Rg gets dialled-up up more and so the gain is dialing up more and more.  That keeps the output constant.  For a limit we don't want it to keep dialing the gain up so one way of stopping the gain going up is to put a fixed resistor in parallel with the gain control (Rg)

So take now look at what happens with a Feedback connection with linear rectifier.   This is basically figure 2 of,
http://experimentalistsanonymous.com/diy/Datasheets/SA571%20AN.pdf

Here the rectifier is fed by the output signal and does not have a hard threshold detector like figure 12.
For the opamp example the connection are the same as feedforward Rf  = Rg and Rin = R3.

At Vout = 1.1V peak , Rf  = 20k,  A = 1,     Vin = 1.1V pk  / 1 = 1.1V pk
At Vout = 2.2V peak,  Rf = 10k,   A = 0.5,  Vin =  2.2V pk / 0.5 = 4.4V pk.
At Vout = 0.55V peak, Rf = 40k, A = 2.0,   Vin  =  0.55V pk /2 = 0.275V pk

So in this case for large changes in the input voltage the output voltage changes less ie. compression.  The point here is the output cannot be kept constant as the input voltage varies but the output does change less tan the input.  In fact the compression ration is only 2:1 which is quite low.    So with a linear rectifier the natural behaviour of a feedback compressor is to produce only 2:1 compression.

So now we look at the Dynacomp.   The rectifier is formed by the two transistors that connect to the filter cap.   The recitifer connects to the output making it a feedback sidechain structure.  When the output voltage is below Vbe = 0.6V the transistors are off so the gain is on maximum.  However as soon as the output voltage hits Vbe=0.6V they turn on and then the gain control action starts.  So in a way the rectifier has a built in threshold of +/- 0.6V at the output.   This is a lot like the previous example of figure 12.

The main difference with the Dynacomp vs figure 12 is if the output voltage was 0.55V the transistor might turn on a little bit and if the output voltage was 0.65V the transistors would get turned on harder.  So unlike figure 12, which is a hard-limit, there is small amount of softness which gives the output voltage freedom to be other than 0.6V.     The overall effect of this is the compression ratio can be less than the infinite compression ratio of figure 12 but more than the 2:1 ratio of the feedback + linear rectifier case.   The transistors provide the exponential rectifier.  This comes about because transistors are exponential in nature, just like diodes.  It is the exponential behaviour that lets the transistor switch on soft or hard.