MXR M-203 Stereo Flanger buffer or compandor noise?

[space_ryerson]

I have an old MXR M-203 Stereo Flanger (the 2000 series one), which has always been noisy in bypass mode, and very slightly less than unity gain. I've had it since around 1995, and to deal with this, I simply built a bypass box to handle it. Tired of the bypass box taking up pedalboard real estate, I'm trying to figure out whether it's the buffers or the compandor (or both) that are creating the noise, and whether there's a not horrible invasive way to correct this. Here's a schematic of the input/output sections with the compandor and switching:

What do you guys think?

[Pablo1234]

So in bypass mode it becomes noisy and quieter.
is it noisy with the effect in or just when in bypass?
a TL072 to replace the 4558. the high impedance of the jfet input may make a difference.

I would guess that you loose volume from the effect being taken out and no makeup volume in bypass, that's typical in almost any pedal that uses a disable on the effect but no makeup on the dry to bypass.

[space_ryerson]

It's noisy both with the effect on, and in bypass.

I should mention that when I said the pedal is slightly less than unity gain in bypass mode, I was comparing that to the pedal being true bypassed by the external bypass box. Into a distorted amp, you can definitely hear a difference between the pedal being true bypassed or just being in bypass mode.

I'll look into swapping the 1458 with a TL072.

[Pablo1234]

For the attenuation I would say just add larger feedback resistors to the output op amps.

Also the pinouts are Identical with both the TL072 and the 4558 and the 4158.

[Fender3D]

1458 is a noisier op amp than 4558, so swapping it will give some advantage...
I'd use a 072 or 5532, I guess "better" op amps (OPAxxx) wouldn't be noticeable here...
you may try lifting the 4.7uF cap from 571 and check if noise comes before mix section...

[Thomeeque]

you may try lifting the 4.7uF cap from 571 and check if noise comes before mix section...

If it does, you may try to add pre/de-emphasis too, it helps to suppress noise a lot - but it would mean 4 new resistors and 3 capacitors to squeeze there

Btw. 1M at input buffer's (+) is probably "grounded" to V_ref = V+/2 (V_S? Label should be at 10k+10k junction then.. As it is now, V_S = V+).

T.

[space_ryerson]

1458 is a noisier op amp than 4558, so swapping it will give some advantage...
I'd use a 072 or 5532, I guess "better" op amps (OPAxxx) wouldn't be noticeable here...
you may try lifting the 4.7uF cap from 571 and check if noise comes before mix section...

Thanks! I'll try lifting the 4.7uF cap first from the 571 to see if that makes a noticeable difference. If not, then I'll put a socket in and try out a 072 or 5532. I'm pretty sure I have both laying around.

you may try lifting the 4.7uF cap from 571 and check if noise comes before mix section...

If it does, you may try to add pre/de-emphasis too, it helps to suppress noise a lot - but it would mean 4 new resistors and 3 capacitors to squeeze there

I think I have at least one example of pre/de-emphasis in my collection of circuit designs, but could you point me to one just to be sure? It's a very dense, double-sided circuit board:


I don't use a battery, so if I needed to, I could probably cram those parts into the battery compartment.

Btw. 1M at input buffer's (+) is probably "grounded" to V_ref = V+/2 (V_S? Label should be at 10k+10k junction then.. As it is now, V_S = V+).

Good catch, T.! That was an error on my part. I've fixed the schematic. If you reload the page, it should update to the corrected version.

[space_ryerson]

Well, I pulled the 4.7uF cap, and there was not any noticeable change in the noise. Tomorrow, I'll swap the 1458 for a TL072, and see what kind of difference there is.

[Thomeeque]

you may try lifting the 4.7uF cap from 571 and check if noise comes before mix section...

If it does, you may try to add pre/de-emphasis too, it helps to suppress noise a lot - but it would mean 4 new resistors and 3 capacitors to squeeze there

I think I have at least one example of pre/de-emphasis in my collection of circuit designs, but could you point me to one just to be sure?

In this case it would be easiest to combine active pre-emphasis with passive de-emphasis at the very output (both of them):



Green is pre-emphasis filter charecteristics,
Blue is de-emphasis filter characteristics,
Red is whole input-to-output characteristics*.

So you'd need to:

• interrupt 6+7 connection on input buffer and add C2+R3/R4 network
• replace output 1k resistors by 10k (R5)
• add R6+C5 to both outputs

T.

*Slight error is caused by C1+R1, in reality there will be yet more error added by rest of the circuit (e.g. by those 1nF caps), but it should not be marginal..

[Fender3D]

Pre-emphasis and de-emphasis are somewhat tricky 'cause the wet signal phase inverting.
I have this on my chorale pedal: www.wizardinside.it/foto/schemi/chorale_mix.pdf. You may adapt it to suit MXR's parts. (sadly you'll have to disconnect first op-amp pins 6-7).

[Thomeeque]

I have this on my chorale pedal: www.wizardinside.it/foto/schemi/chorale_mix.pdf. You may adapt it to suit MXR's parts.

De-emphasis cannot be done same way here, it works only if all signals are mixed at inverting inputs of opamps - it's the main reason why I suggest passive de-emphasis at the outputs.

Actually I don't think wet (BBD1 & BBD2) signals led by 100k resistors to non-inverting inputs are de-emphased correctly in your chorale either (but that was maybe not the goal there).

T.

[Fender3D]

De-emphasis cannot be done same way here, it works only if all signals are mixed at inverting inputs of opamps - it's the main reason why I suggest passive de-emphasis at the outputs.

Actually I don't think wet (BBD1 & BBD2) signals led by 100k resistors to non-inverting inputs are de-emphased correctly in your chorale either (but that was maybe not the goal there).

T.

Upper 1458 has in-phase dry/out-of-phase wet mixing (it's the out-of-phase version of my dry/R100k mixing), whereas lower 1458 has in-phase dry/wet mixing (It's the out-of-phase version of my dry/R82k mixing).

I left signal levels purposedly uneven (it enhances stereo separation and it's a chorus, flanger will need a better balance...) and the 82k resistors will compensate BBDs/filters/gates attenuation (you may simulate schematic as they were 100k).
They're within +-1 db input signal, out of phase signal is ~85mdB higher and has a slight enhance (less than +1dB) around 100Hz.

Passive de-emph. at outputs also will be slightly influenced by input impedance of the following effect/amp.

[Thomeeque]

 Hmm, I still don't see how it could be done. Please, if you know how to update network around 1458's to get proper de-emphasis, draw it into the schematic.

They're within +-1 db input signal, out of phase signal is ~85mdB higher and has a slight enhance (less than +1dB) around 100Hz.

I was talking about those signals led into non-inverting inputs, which get de-emphased "twice" (and drop by cca 8dB) in your build:

Passive de-emph. at outputs also will be slightly influenced by input impedance of the following effect/amp.

Yes, that's valid note, I should have noted it. On the other hand with "normal" guitar gear loads (hundreds of kiloohms) it should be negligible. Output could by made yet stronger by adjusting R5/R6/C5 values, but I would not worry about it.

T.

[Fender3D]

Doh!  
I drew this:



(R18 1Meg, is there just to provide biasing)
which lead me to this:



PS
How did you managed to get those smaller LTspice windows (yet readable)?

[Thomeeque]

Doh!  
I drew this:

In "out1" emulation branch you cannot ignore 150k and 82k resistors leading from various signal sources (input buffer, BBD1, BBD2) to inverting inputs - each signal source acts as a ground from AC point of view. Once you will add these resistors there ("grounded" to VB1 in your case), IC2 gain for non-inverting input will not be constant 1 anymore.

PS
How did you managed to get those smaller LTspice windows (yet readable)?

It's a print-screen of LTspice window with 50:50 horizontal split between schema and graph (which happens automatically when I run simulation with schema maximalized), manually cropped in graphics editor (IrfanView). And, well, my schema is bit smaller that yours

T.

[oldschoolanalog]

@space_ryerson: Quick Q; Is this flanger the black 3 knobber w/blue print and heavy enough to use as a doorstop? If so, let me know as I have one that fits that description. It's also R5106 based (notorious for being a very noisy BBD). If it is, I'll dig mine out & plug it in to see if it has any of the same symptoms yours displays.
Peace!

[space_ryerson]

I gave a TL072 a try in place of the 1458, and it definitely cuts down the noise, but it's still there to some extent. If a sound clip would help, I can put one together. The plus side is that the flanging definitely sounds a little cleaner with the TL072 in place.

I'll see how feasible it is to insert the pre- and de- emphasis. I'm about to go into rehearsals for a show next week, which I'm going to use the flanger in, so I probably won't try this until after the show for the sake of sanity

@space_ryerson: Quick Q; Is this flanger the black 3 knobber w/blue print and heavy enough to use as a doorstop? If so, let me know as I have one that fits that description. It's also R5106 based (notorious for being a very noisy BBD). If it is, I'll dig mine out & plug it in to see if it has any of the same symptoms yours displays.
Peace!

Yep, that sounds like the same flanger. If you have time, I'd definitely be interested in hearing what yours sounds like. It's my favorite flanger, but it's just not the quietest the pedal by any means.