MXR Distortion plus bahaves strange!

[sajy_ho]

Hi guys, hope you're OK.
Finally got a tima to build a Distortion plus. I used GGG's schematic but built it on a perfboard:http://generalguitargadgets.com/pdf/ggg_dist_plus_sc.pdf
The problem is when the gain knob at almost max; suddenly the distortion becomes very ugly.
I removed the clipping diodes and found that max gain pot ugly distortion comes out of the chip!
I'm using an ic socket and tested two UA741s but the problem didn't disapear.
Voltage readings are ok except two pins:
Pin 3 (non-inverting in): 1.8V (supposed to be 3.6V according to GGG's build instruction).
Pin 8 (NC): 1.2V!!    I tested this pin and when the unit has no power, this pis is not connected to any of other pins!
I tested all the connections almost 10 times with a DDM and everything seems to be ok...
Does anyone have an idea what could be wrong here?

Thanks
Sajad

[duck_arse]

sajy, all the voltages please! although we don't really care about pins 1 or 8 or 5 in this inst. it is probable that your meter is loading the input voltage reading, so we look at the output pin instead, which has grunt enough to ignore the meter load. if the output volts is good, it means all is well in DC world at the input pins.

pin8 has internal connections, apparent in the internal circuit diagram, usually shown in the datasheet.

[dschwartz]

Check your Vref values..maybe you got the resistors wrong..

[sajy_ho]

Thank you guys.
I checked all the voltages again:
Vcc: 9
Vref: 4.5
Pin 1: 0
Pin 2: 4.48
Pin 3: 1.8
Pin 4: 0
Pin 5: 0
Pin 6: 4.48
Pin 7: 9
Also I checked all the resistors and caps and everything seems to be OK. I don't know if the meter is loading the input pin; how can I find it out?

[ashcat_lt]

Thank you guys.
I checked all the voltages again:
Vcc: 9
Vref: 4.5
Pin 1: 0
Pin 2: 4.48
Pin 3: 1.8
Pin 4: 0
Pin 5: 0
Pin 6: 4.48
Pin 7: 9
Also I checked all the resistors and caps and everything seems to be OK. I don't know if the meter is loading the input pin; how can I find it out?

You kind of just did!  Pin 6 is sitting right at Vref, and pulling Pin 2 with it.  That means the voltage at Pin 3 must be right during normal operation, but when you put your meter on there it pulls way down because your meter is apparently super low impedance.  You could look at whatever documents you can find re: your meter, or measure it with something known to have a much higher impedance, but I'd just replace it because it's going to mess with you all over the place.

[dschwartz]

The meter will load the pin, but around 0.4 volts...try changing the two 1meg bias resistors for a much smaller value, like 100k, and increase the cap to 10Uf

[dschwartz]

Btw, the dist+ distortion is very ugly, fuzzy and dirty when maxxed...maybe it is right...

[ashcat_lt]

The meter will load the pin, but around 0.4 volts...try changing the two 1meg bias resistors for a much smaller value, like 100k, and increase the cap to 10Uf

With the equivalent of 1.5M as a source impedance, a 1M meter impedance would pull 4.5V Vref down to exactly 1.8V.  You'd really want 15 or 20M at the meter to have anywhere close to accurate measurement there.
Except that's wrong!  The numbers work out, but...

The Vref divider has 1M as the top resistor, and 1M || (1M + (meter resistance || opamp resistance)) for the bottom resistor.  That gives you the voltage at that end of the 1M biasing resistor.  The voltage at the other end would then be divided by 1M for the top and meter resistance || opamp resistance for the bottom.  I'm not going to bother to work that out, especially since there are two unknowns.  I still think that meter needs to be replaced.

[Mark Hammer]

I don't know if it's the chip itself, or maybe it's the biasing scheme (I mean, really, a 1M+1M voltage divider?), but nobody gets a clean sound from this thing when the clipping diodes are lifted, even at minimum gain.  So it's not you.  It's the design.

The Dist+ is kind of a double-clipper in disguise.  That is, the signal hitting the diode-pair is not only boosted, but has already been rendered dirty.

Having said that, there is always the possibility that something is amiss with your gain pot.

[Electron Tornado]

I don't know if it's the chip itself, or maybe it's the biasing scheme (I mean, really, a 1M+1M voltage divider?), but nobody gets a clean sound from this thing when the clipping diodes are lifted, even at minimum gain.  So it's not you.  It's the design.

The Dist+ is kind of a double-clipper in disguise.  That is, the signal hitting the diode-pair is not only boosted, but has already been rendered dirty.

Having said that, there is always the possibility that something is amiss with your gain pot.

An opportunity to learn something -

Mark, how does using 1M+1M in the voltage divider cause the chip to clip? I recall one of the differences between the MXR Dist+ and the DOD OD250 is the resistor values in the voltage divider.

[ashcat_lt]

I don't think the Vref divider affects the clipping much.  We can see that opamp is idling right around half the supply.  The issue is that the thing has about 200x gain in the passband.  A healthy 1V peak will try to go to 200V, but even with a rail-to-rail opamp, it can only get to about 9V, and with most will fall short even of that.  Even a 100mV sustain portion of a note will try to swing more than twice as far as the supply rails will allow when the gain pot is all the way up.

[Mark Hammer]

An opportunity to learn something -

Mark, how does using 1M+1M in the voltage divider cause the chip to clip? I recall one of the differences between the MXR Dist+ and the DOD OD250 is the resistor values in the voltage divider.

I don't know.  But point me to anything else that uses a voltage divider pair to derive a Vref with that high a resistance.  Oh, you'll see 10k pairs, and 22k pairs, and 47k pairs.  But 1M?  Only thing I've ever seen with that much current limiting is the Dist+, so I've convinced myself that it is doing something.

The DOD250 has essentially the same design, with 3 major differences:

• Uses a 4558 instead of a 741
• Uses silicon diode pair instead of germanium
• Uses a 22k+22k pair to derive Vref and a 470k resistor from Vref to the chip input

It does not aim for greater gain.  But the 250 will clean up at low gain without the diodes in circuit.  So is it the use of silicon vs germanium?  Not when thediodes are out of circuit.  Is it the 741 vs the 4558?  The 741 isn't that bad of a chip, and the 4558 not that great of a chip; with not that big a difference between them.  So the only thing I can figure out is that the biasing of the input, and the current starvation has something to do with it.

The bias current reaching the non-inverting input on the 741, is a mere fraction of what it would normally be on most other designs using a 741.  Keep in mind that the current being tapped from V+ has to pass through one 1M resistor, to the junction of the divider pair, where V+ is dropped down to 4.5V, and then pass through another 1M resistor to the 741 input.  Contrastthat with the 250, where the same initial V+ passes through a 22k resistor, atthe divider pair, and then through a 470k bias resistor.  BIg difference.

There are folks here a whole lot smarter about such matters than myself, so I will ask them to step in and either show me where I'm dead wrong (which I am completely open to), or explain to me how the biasing and lack of cleanliness are related.  Is it something unique to the 741, and not amenable to duplication with other op-amps?  I have absolutely no idea.

Maybe some of you with a DOD250 on your breadboard can try introducing Dist+ style input biasing and see if it does anything unusual.

[Mark Hammer]

I don't think the Vref divider affects the clipping much.  We can see that opamp is idling right around half the supply.  The issue is that the thing has about 200x gain in the passband.  A healthy 1V peak will try to go to 200V, but even with a rail-to-rail opamp, it can only get to about 9V, and with most will fall short even of that.  Even a 100mV sustain portion of a note will try to swing more than twice as far as the supply rails will allow when the gain pot is all the way up.

Absolutely agree.  And I'm familiar with clipping circuits, like the Black Cat and Sans Amp, where no diodes are used, and the clipping is purely a product of headroom limitations in the chip.  The mystery to solve here is that even with gain turned down all the way, and the diodes out of circuit, the thing never gets clean, despite a combined input signal and gain that is hypothetically well within the chip's headroom.

[ashcat_lt]

Absolutely agree.  And I'm familiar with clipping circuits, like the Black Cat and Sans Amp, where no diodes are used, and the clipping is purely a product of headroom limitations in the chip.  The mystery to solve here is that even with gain turned down all the way, and the diodes out of circuit, the thing never gets clean, despite a combined input signal and gain that is hypothetically well within the chip's headroom.

Frankly, I'm not familiar with this pedal.  I'm trying to help with some of the theory, but I'm also working, so I don't always get to think through all of my answers.  I edited one of my replies above after drinking about it a bit...

[sajy_ho]

Thank you all guys for your help.
Actually I think Mark is right; the circuit is a bit dirty even with the gain knob at min, But the kind of distortion I'm talking about is when the gain passes around 4 o'clock. It sounds like opamp clipping, and you said that is normal in the case of Distortion+ ; so I guess the problem solved!

Thanks again
Sajad

[sajy_ho]

I did some tests on the circuit and found some interesting facts; I used my laptop supply (20V), to see if hitting the rail voltage is cause of the clipping, and the circuit acted exactly like before. I mean the chip starts clipping at around 4 o'clock, so I guess there is another reason for the chip to clip, maybe what Mark said about the biasing circuit. I should replace those resistors with lower values to find out...
I also used my old analog multimiter (a lot of digging in the basement!) and now the voltage reading on pin 3 raised to 3V, so the meter was deffinitely loading down the input!

[dschwartz]

I suppose you already tried changing the opamp?? Sometimes they fail..
Also the usual stuff, reflowing solder joints, changing wires..

741 has awful headroom, specially when is heavily loaded (low resistance from output to ground)

Are you using a reverse log pot? If not, try a 100k linear pot, or better yet a reverse log. Log pots will not work well in this circuit..

A nice mod is to change the gain configuration, change the 1M resistor for a 1 meg log pot, and eliminate (jumper) the pot from Feedback to ground..tailor the amount of distortion available by modifying the resistor to ground..i like 1k there (1001 times amplified at max gain)

[sajy_ho]

I suppose you already tried changing the opamp?? Sometimes they fail..
Also the usual stuff, reflowing solder joints, changing wires..

741 has awful headroom, specially when is heavily loaded (low resistance from output to ground)

Are you using a reverse log pot? If not, try a 100k linear pot, or better yet a reverse log. Log pots will not work well in this circuit..

A nice mod is to change the gain configuration, change the 1M resistor for a 1 meg log pot, and eliminate (jumper) the pot from Feedback to ground..tailor the amount of distortion available by modifying the resistor to ground..i like 1k there (1001 times amplified at max gain)

Thanks for the reply,
Unfortunately the only type of pot could be found here are logaritmics
I changed the Level pot to 100k and that didn't help the clipping.
Also the circuit already have a gain of around 213 and the opamp is clipping at max gain, so how could raising the gain help the clipping?

[dschwartz]

Your issue is the log pot.. You need a linear or reverse log for the gain pot..
If you only have log pots, you can reverse the pots connection, but the gain will work in reverse..

Or put the gain pot at the feedback, replacing the 1M resistor..

[Mark Hammer]

Log pots will work fine, if you wire them up in reverse so that clockwise is less gain.

The use of a 500k or 1M antilog pot in the Distortion+ was to be able to span the range from as clean as the unit could go up to the maximum gain of 213x.

But that was in the days when players had maybe 3 pedals.  There is no requirement these days - especially among DIYers - for a single pedal to have to cover slightly-dirty boost up to distortion.  So, Dschwartz's suggestion to use a 100k pot is entirely reasonable.  Keep in mind that, with a 100k pot (+4k7 resistor) and 1M feedback resistor, your minimum gain is around 11x, which isn't really enough to do any serious distortion.  And since you don't need to progress through all the subtle degrees of slightly-dirty boost, even a linear 100k might be fine.