MXR Noise Gate Tonepad Revisited Need Help

[Ajay]

[...] add a 68pF across the 1M on pin 1+2 IC1
Changed 680R//1uF on pin2 to 4k7//2u2
changed the 1M from the transistor collector to the fet gate to 2k2.

Followed Duck's mod and then finding out I had messed up the footprint of the FET  
Got it to work with a j201 tonight, so I will put it in an enclosure tomorrow during lunch break and add it to my fx board

Thanks for the hints Duck!

[Kipper4]

how did you impliment the 68pf
from pin1 to 2 of IC?
or in paralell with the 1M resistor?

[duck_arse]

the 68pf, the value isn't critical, is parallel with the 1M across pins 1 and 2. it is there to roll off the high frequency gain of the opamp.

[edit] exactly as shown in zippos pic. above ^^

the 2 transistors are doing nothing much, and the low gain type specified is good enuff. don't waste time with matching them, just grab.

[Kipper4]

Callgate Version1
(not the modded version2)


Heres my findings
I implimented All of Ducks mods
modded values
Q1 Q2 2n3904 as original
Q3 j201 more on this later
2k2 between transistor collector and fet gate
68pf in parralell with the 1M pin 1 to 2 of IC
10nf (replaces 1uf NP in series from Q2 base)
4k7  // 2u2 electrolytic    (replaces 680R //  1uf  from pin2 of IC)

Not just any old J201 will work I found the one with the smallest switch on of voltage i could less than 0.15 worked for me

How do i know which ones are which i recently tested all my jets with a matcher (long story involving a mini phase 45 that Duck also helped me with)

Edit: Op amp is TL072, I've yet to try the lm1458. Its on the cards.

Its working fine now always plenty of range on the pot and a nice smooth decay on the gate shutting.
Thanks for the Help everyone especially Duck Arse.

[Ajay]

Its working fine now always plenty of range on the pot and a nice smooth decay on the gate shutting.
Thanks for the Help everyone especially Duck Arse.

Great that it works now! Mine's working too btw, also with the help of Duck

It still puzzles me why this design won't work as intented with the given values. Some kind of design update that wasn't documented properly? Some sort of design protection? Or...?Therefore this afternoon I built a second board with as much stock values and part types where it could matter (with one major exception: the opamp is an OP275 of which I have plenty available and which is my favorite general purpose audio opamp). I intend to do some additional testing/modding with it to see if I could improve this design somehow.  
I also wonder what the design criteria were; for instance why didn't they use some sort of rectifier? In the current setup I measured a lot of ripple on the control signal to the base of the second transistor which I think could rather easily be prevented. I simulated the circuit with a rectifier diode (1n914) in Pspice and it seems to do the job better. I may add that to this second board as an experiment.

[duck_arse]

ahhh, well, now that we are redesigning this wheel instead of just re-rounding it ....

I thought it would be better with a low noise single at IC1b, and use a cmos low power dual or tl062 for the recitfier and buffer the Vref. and use a diode rectifier. and possibly a small value fixed resistor in the "attack" spot, takes away the nasty gate opening sound.

and I think it was mark hammer I was reading somewhere recently that said there was 2 types of gates: one shuts hard off, the other just reduces the volume to a less obnoxious level. dunno if that will get us anywhere with this, though.

and good work, kipper, I'd leave the tl072 if you're happy with it. the 1458/4558 would be a step backwards performance wise.

[Mark Hammer]

Dumb question, but does this thing really use a momentary switch?  I used to have one in the late 70's, and while I never looked inside much, except to change the battery, and 35 years may have dulled my recollection, I remember it as having a distinct CLICK when I stepped on the stompswitch.

Note that the 10uf cap connected to Q3 provides a "bleedoff" path for the entire signal.  Seems to me that one could turn the unit into a sort of "de-hisser" by reducing the value of that cap so that only the upper mids and treble are attenuated when you stop playing, and not the entire signal.  That would let the tail of a sustained note die out more naturally, instead of being cut out.  And since there isn't much treble to that part of the note, the action of the circuit would be largely "invisible".  Note that this would still let any hum that makes up part of the "noise" still pass through unaffected, so this suggested mod would really oly be useful on a signal whose major crime is added hiss.  I'll suggest starting out with a value of .039-.047uf, and seeing whether it needs to be made larger or smaller.

Another potential mod is a "range" switch to allow for better dialability of mic vs line level signals.  As shown, IC1a is set for a gain of just under 1500x.  Make the 680R a 2k7 unit instead.  This will drop the gain down to 370x.  If you use a SPST toggle to put a 910R resistor in parallel with the 2k7, that drops the effective resistance down to 680R, and bumps the stage gain back up to stock (taking 5% resistor tolerances into account).  With the gain reduced, it should be able to handle much hotter signals without forcing you to stay in the 7:00-7:30 range of the Threshold control.

[duck_arse]

mark, is Q3 the fet? do you know what the purpose of the 47nF from the output of the follower to the 1M at the gate is? and do you rekon a resistor in series with the 10uF would provide "lowered volume" rather than "gated output", dividing w/ the 22k?

[Mark Hammer]

On the PDF I'm looking at, yes Q3 is the FET.  No clue what the .047uf you refer to is for.

RE: series resistor.  Keep in mind that the drain-source resistance of the FET, plus whatever the resistance setting of the suggested pot in series with it might be, is placed in parallel with a 1M resistor to Vb.  This is NO different than the way a 2N5952's drain-source resistance is placed in parallel with a 22k resistor to Vb on each of the phase shift stages in a P90.  The 1M resistor essentially sets the max resistance to Vb, and the FET+pot determines how much it drops below that.

The 10uf cap in between those components and the 22k resistor keeps pesky DC out but needs to be in between the 22k and the 1M so that the FET+pot can be placed in parallel with the 1M.

But let's ignore the 10uf cap for a moment and pretend the 1M+FET+pot are tied directly to the 22k resistor.  What we have there is exactly analogous to the 82k resistor and 2N5457 we see on the Armstrong Orange Squeezer...except that the two are intended to work in opposite fashion.  That is, the 82k (22k) and JFET are intended to serve as "automatic voltage dividers".  In the case of the OS, the FET resistance gets lower in response to signal peaks and attenuates the signal.  In the case of the NG, the FET resistance is low, and attenuates (in conjunction with the 22k) until the signal reaches the appropriate turn-on threshold.

Based on that premise, the degree of attenuation produced during the "off" and "on" states can be played with by varying the 22k resistance, the 1M resistance, the setting of the "attenuation" pot, and what you apply to the gate of the FET.  And the spectral range that this attenuation is applied to can be varied via the value of the cap that is presently 10uf.  But I'm guessing that the degree of attenuation produced in the "off" statecould be varied by means of parallel paths.  Let's say that you stuck a 47k resistor between the 10uf and FET, BUT placed a .022uf cap in parallel with the 10uf+47k.  The low end would now still have some attenuation applied in the off state, but more attenuation would be applied to the higher frequencies.

I suppose in an ideal world, one would want one FET attenuation for the bottom, and another for the top, and each would have different time constants, such that the release time for the treble would be faster than for the lower mids and bass.

[Fender3D]

Why don't you use a 100KC pot instead of Drain's 1M resistor and connect FET source to wiper?

[Ajay]

Interesting stuff guys!

So, where are we going with this? Are we -as Duck put it- going to reinvent/redesign the triangular wheel into something that actually does the job? I wouldn't mind doing another board spin for a decent noise gate Especillay if I could fit it in my 1590A case again 

[Kipper4]

Just when I thought it was sorted. I spent ages redrawing the schemo with the mods in eagle)
This will be intresting. I'll leave it to the experts. Is parts count still going to be a redesign factor?
Thanks guys

[Ajay]

AFAIC, parts count is not limited, as long as I can squeeze them in a 2" x 2" board. Granted, 1590A case will not fit anymore then, but I could live with that.

But seriously, if we would go for this, there are several factors that should be weighted: parts count, board size, vero vs. pcb, single/double sided, regular/smt etc.
EDIT: and most importantly: features!

I'm all ears  

[hangingmonkey]

Cool, ill try these changes. I never got my tonepad pcb to work

[duck_arse]

can one of you fellas with a unit "working good" and a high-gain noisey something do a short audio of your gate-ing, please? maybe gating good, and bad, so I can compare. I wanna know what to listen for.

[peterg]

duck - I won't subject anyone to my (lack of) guitar skills but can give you a run down of my "working good" NG. There is stuttering if the note is left to ring. I don't think it can provide clean staccato notes without muting.  It's good for removing noise when hitting a note and then muting i.e. reggae or AC/DC. The pot setting needs to be precise - around 8:00. A higher setting around 3:00 will remove hum from noisy pedals. I haven't tried an MXR Noise Gate but from the promo videos my is working as it should.

[Ajay]

They way you describe it Peter 'sounds' familiar and is exactly what you'd expect one you've seen how the gating works.

[Mark Hammer]

Cool, ill try these changes. I never got my tonepad pcb to work

This is why I suspect the use of a momentary switch is incorrect.  If you think of it, when not bypassed, the circuit is supposed to remain "off" until a sufficiently high-enough signal turns it "on".  "Off" consists of the the FET gate being turned on such that the FET acts like a low resistance to ground (Vb, actually).  When the gate is "on", that means the FET has been turned off and presents a high resistance to Vb.

So, anything that interferes with turning the FET on will cancel the gating effect by leaving the gate permanently "on"/open.  And closing the switch shown will do just that...as long as the footswitch stays closed.  There is nothing in the circuit that would lead to the gate remaining in an open/on state as a consequence of the footswitch being momentarily pressed.  And since the SPST footswitches people likely use are normally open, bypassing the effect would require you to hold the footswitch down with your foot for as long as you wanted to remove gating.

That is certainly ONE of the potential reasons why different people may not have been able to get theirs to "work".

[Kipper4]

I couldnt see the point of the momentary defeat switch either and did not include it in my latest build.
I've redrawn the schematic in eagle and done a perflayout with the above mods in DIYLC too.
I'm waiting for some improvement and updates before though
I'm sure this gate could be better. I just wished i had the knowledge to do it.
I'd be guessing and hacking.

[Ajay]

My understanding of the MOM switch was to prevent the gate from closing (by keeping the gate voltage low and preventing the 1u cap from building up charge). I assumend this is intended by the author of the schematic when for example at the end of a solo where you'd like a note to go into feedback instead of the gate. As this was not my intended use (i'd rather use a real bypass switch), I left that switch out.

But still, there is room for improvement...