MXR phase 100

[jrc4558]

Hey everybody!

Two questions about the abovementioned pedal.

Do you know how many phase-shifting stages it uses?
Do you know if the current reissues are true bypass?

Thank you all.

[MAXIMUS]

10 stages.... dunno if it is true bypass or not tho... You looking to buy or build?

[Boofhead]

10 but only 6 are swept - so it's more like 6.

[jrc4558]

Yeah, I want to buy one. Phasers seem to be too somplicated to perf-board.

[Mark Hammer]

Although only 6 stages are actively swept, the 4 fixed stages are not doing nothing.

Consider that the RC combination in each allpass (phase-shift) stage dictates that the phase of the input signal is shifted by an increasing number of degrees (according to input signal frequency) starting from frequency X.  So, it may be the case that as the FET or LDR is swept, at some point there is, say, 20 degrees of phase-shift per section at 500hz (80 degrees for a typical 4-stager) and 30 degrees per section at 800hz (I'm just making the numbers up for illustration purposes; I'll leave the more accurate math to the EE's).  Frequencies higher up would have more phase-shift applied to them up to a maximum of 90 degrees per section.

The cancellation you get depends on the [cumulative] amount of phase shift applied at that frequency at the end of the wet/effect path.  The swept stages will change how much phase shift is applied at each frequency, but any fixed stages will add to that.  So, if I had an additional fixed stage that added an unchanging 20 degrees of phase shift at 500hz by virtue of the resistor and cap used, the hypothetical state described above would result in 4*20 plus 20 more, or 100 degrees of phase shift at 500hz.

Whether you get the math or not, the point is that additional fixed stages are a cheap and easy way to increase the intensity of a phaser without having to shell out for additional control elements OR risk the limitations on headroom they may otherwise provide when things like FETs are used.  In the case of the P100, MXR went with the fixed stages more for cost/packaging concerns than for any headroom ones, since LDRs impose no headroom risks.  I might point out that Eventide used fixed stages in their phasers, and one of the ROSS phasers (one of the orange ones, I think) also used a fixed stage in the feedback loop.

THe P100 has a number of reall interesting mods that can be done, some of them replacing the lame 4-position preset switch.  If you're game, I can explain those too.

[JT]

Mark,
        I would like to hear the Mods. i was about to order a 4 pos rotary switch for this project . Thanks in Advance!
JT

[Mark Hammer]

Before I go any further, here is the reference schematic, so we are both referring to the same thing: http://www.generalguitargadgets.com/v2/diagrams/mxr_p100_sc.gif

Second, whenever you have an effect which is modulated by a low-frequency oscillator (LFO), there are a number of parameters that can be changed:
1) The speed/rate of oscillation
2) The range or width of influence that LFO has on the effect (essentially the amplitude of the LFO output, variously called "sweep", "width", "depth", etc.)
3) The starting point of the sweep, often called "manual", "initial", "start", etc., which is essentially a DC offset that the LFO voltage is added to.  (Note that there are limits to a) how much sweep/control voltage you can produce, based on the power supply, and b) how much control voltage can be sent to the control element in question.  For this reason, many manufacturers make the sweep amount (#2) and this manual adjustment interactive, so that the two combined never exceed the maximum feasible.)
4) How much of the effect and straight signal are combined, often referred to as "blend", "mix", "depth", "effect", etc., depending on how it is done.  Sometimes it involves a single or dual-ganged control that increases one while it decreases the other.  Sometimes it involves a single control that allows the effect to be equal to the straight, or attenuated to varying degrees.
5) How much of the effect signal is fed back or recirculated through the effect path all over again.

The MXR Phase 100, in a manner typically of its 1 or 2-knob era, kept controls to a minimum.  Even though these could have been separate panel-mount controls, the 4 position switch essentially duplicates the 2 functions of sweep width and regeneration as presets and packs them together.  Two regeneration amounts, and two sweep widths gives 4 combinations.

The regeneration control is easily turned into a continuously variable control.  All the switch does is feed the shifted signal back through R1...or not.  In essence, this is a pot with two values: 0 ohms and infinite ohms.  This yields either no feedback, or a modest amount.  Replacing R1 with a 10k resistor and  250-500k variable resistor in series could do a very nice job carrying out this function, providing everything from negligible amounts of feedback, to over-the-top near oscillating feedback, and everything in between.  It is common for phasers to replace the equivalent of R1 with a trimpot, adjusted so that the circuit just nearly breaks into oscillation.  The panel-mount control can then be used to further reduce that, but you have assurances that no setting of the knob will ever reach a point of oscillation.

The other function of the switch is to adjust sweep width.  Here, I am guessing as to the appropriate replacement of the switch with a pot, but let's say 1meg for starters.  When the switch is closed, you essentially have a zero-ohm resistor bridging R41 and R45.  The question is how much resistance is needed to mimic the effect of opening up that connection.  You may find that a 3-position toggle switch that either bridges it with one of two different resistances or leaves it open can provide enough sweep width options if it turns out that sweep width depends on a wider range of pot values than you can provide.

The centre frequency trimpot is essentially the same thing as  the "initial"/"manual" control described above, and can be easily moved to a panel-mount control.  A 25k pot should be just fine.  There is enough tolerance/play between the 47k and 39k resistors on each side of the trimpot, and the trimpot itself that a 25k pot will not exceed the range of possible resistances that circuit fragment could take on.  Keep in mind that those 3 components essentially form a 106k pot which is only allowed to rotate a bit from its midpoint.  Of course, if you bought a 106k pot, it would likely have an actual measured value anywhere between maybe 95k and 112k so don't sweat the difference between a 20k and 25k pot.  If you find that the range of control it affords is not to your liking, feel free to tinker with the values of the resistors to the side, as long as you aim for the overall resistance value.  For example, it may be the case that all the adjustment you really want can be achieved by having R46 be 56k, R48 be 47k and R47 be 10k.  Let your ears guide you.

The dry and wet signals are passively mixed via R18 and R19, a pair of 6k2 resistors.  If you increase R18, the wet/phase-shifted signal dominates.  If you increase R19, the  effect recedes into the background.  Lift R18 or make it a very large value, and you will get vibrato.  As JC Maillet has pointed out, when you produce vibrato by means of phase-shift stages, the amount of pitch bend depends on the number of phase-shift stages, so the vibrato function here will be exaggerated, relative to a Univibe or other 4-stager.  That will make the sweep width control come in handy, the same way it does on a chorus.

Depending on your tastes and needs, there are different ways to accomplish the mix/blend function here.  One way is to insert a 50k-100k pot (wired as variable resistor) in series with R19 so that you can go from a 50/50 blend of wet/dry (yielding maximum notch depth) to just a bare hint of wet.  Being able to make the phase-shift signal fade a bit into the back is great for bass, and also great for those situations where you like the tone of a high regeneration sound, but don't want to rub it in your face.  This method does not permit you to use the control to achieve vibrato, however, since it does not affect the straight signal.  If you want vibrato, you'd need to install a toggle to lift one end of R18 out of circuit.  Or else you could replace R18 with a 1meg resistor and use the toggle to put another resistor in parallel and achieve a 6k2 resistance.

Alternatively, you could stick a pot between R18/R19 and C7, such that the wiper of the pot is tied to C7, and each outside leg of the pot goes to the end of R18/R19 that would have normally gone to C7.  This pot allows you to adjust the balance of each signal, with equal levels and max notch depth at the midpoint.

Since it takes a whole lot of resistance to make the dry signal go away, and takes much less resistance to make the wet signal recede into the background without entirely going away (if you want it to completely go away you step on the stompswitch, right?), the smart thing to do would be to use a high value linear pot, but put a parallel resistor between the wiper and the lug going to R19.  This way turning it in one direction imposes moderate resistance changes for the wet path and large resistance changes for the dry path.  I'm going to take a wild stab and suggest a 1meg linear pot with a 100k parallel resistor.

There are tons of other more exotic mods you could do,but there, in a chunky nutshell, is your basic 5-knob phaser.

Have fun.

[jrc4558]

Mark, I will erect a small monument in your honor right in front of Active Surplus on Queen street here in Toronto.

THIS IS FREAKING GREAT!!!!!!!!!!!!!!!!!!!!!!!!!!        

[Mark Hammer]

As long as it doesn't look like the old gorilla they used to have (although I miss it), and as long as you can arrange for Rachel from Much to come over and interview it. :lol:

[Mike Nichting]

Hey Mark,
I'm up for a couple more exotic mods  :wink:  :wink:

I am going to build another P100 and mod the hell out of it.
I would love to hear what else you have instore for it hehehehe

Peace and thanks
Mike N.

[bwanasonic]

Top of my list would be envelope control with the ability to assign the envelope to different parameters. Maybe someway of splitting the stages and using them as two seperate phase shifters in series, with different rates or envelope response.

Kerry M

[moosapotamus]

Just finished perusing my P100 notes from previous discussions on this topic. If you're looking for more, in addition to the mods Mark discussed, you might consider these...

If you replace R56 with a lower value, or a 10K trim pot, you can get higher speed settings.

Add a switch to select different tap points for the feedback (regen/resonance) path. Stock, it's tapped from IC2a pin 1 (stage 9). Tapping from the other odd numbered stages will give some different sounds; IC2b pin 7 (stage 7), IC4b pin 7 (stage 5)... and so on. You could also try leaving the tap point at stage 9, but instead of returning it to the input of stage 3 (IC6b pin 6), return it to a different stage, like stage 5 (IC4b pin 6) or stage 7 (IC2b pin 6).

Try adding a series capacitor in the regen path (high-pass filter). My notes indicate that something in the range 0.022 to 0.1 will provide some additional flavors. Maybe switch between a couple different values and none at all.

I also wonder what a low pass filter in the regen path would sound like. Instead of putting the cap in series with R1, just tie it down to ground.

Hmmm...

~ Charlie

[Mark Hammer]

Yeah, the selectable tap points is a nice option that only really becomes a possibility when you have that many stages to dicker with.

One of the things you see with some phasers is the use of a cap in the last allpass stage feedback loop to keep accumulated noise in check.  The noise issue won't be as bad with an LDR-based unit, but one would be foolish to expect that 11 op-amps in series contributes NOTHING to the unit's noise level.  A 1000pf cap in parallel with the 10k feedback resistor in the last phase shift stage will roll things off at just under 16khz.  A 1500pf cap will roll off around 10khz.  If you wanted ironclad guarantees of noiseless functioning, then you stick a similar cap in a few places (e.g., after stage 4 and 10) to accumulate sharper rolloff, use 1% metal film resistor wherever appropriate (buy a 100-pack of 10k), and use low noise op-amps.  The TL022's specified are low *power* but I have no idea about their noise specs, relative to other devices.  If you have any remote plans whatsoever to use a wall-wart, go for low noise instead of low power.  

Finally, consider upping the gain in the first stage a bit for a better S/N ratio.  Presently it is set at (10k+20)/20k = 1.3.  The LDRs are capable of tolerating a much heftier signal than FETs or OTAs are so go for it.   Crank up the 10k feedback resistor in the input stage to 22k-47k.  39k will give you a gain of (20+39)/20 = 2.95.  Given that the input signal is boosted, you will naturally want to stick an output level control on to provide whatever effect/bypass balance you wish.  The effect path is terminated by a 150k resistor.  You won't want to reduce the effect level that much below the bypass level anyways, so just use a 100k log potand stick a 47k fixed resistor between the ground side of the pot and ground.  This will provide whatever mojo the current terminating resistance yields, but let you achieve output levels from higher than to lower than the bypass level.

[saxtim]

Replacing R1 with a 10k resistor and 250-500k variable resistor in series could do a very nice job carrying out this function, providing everything from negligible amounts of feedback, to over-the-top near oscillating feedback, and everything in between.

The other function of the switch is to adjust sweep width. Here, I am guessing as to the appropriate replacement of the switch with a pot, but let's say 1meg for starters

I'm about to embark on a build on this and would like to try these mods - has anyone else used them, and if so, can you report what value pots worked well?

Would you recommend using linear or log pots for these controls?

tim

[Mark Hammer]

Linear should work fine.

There are plenty of classic cases where we know that audio taper is preferred or even necessary, such as volume controls, and other cases where we know linear is necessary (e.g., Baxandall tone cut/boost controls).

At the same time, there are plenty of other cases where the recommendation to use one taper or another is really a slightly educated guess about what part of the range will be most useful to the user.  Sometimes small variations at the lowest end of the resistance range of a pot produce noticeable changes that one assumes are of use to be able to dial in where changes at the high-resistance end produce negligible changes.  Other times, variations in resistance at any point across the range produce identifiable audible changes to that user.  At still other times, a user may find they can *hear* the changes produced by small changes in resistance at the low/high end of the resistance-range, but they are most interested in small variations at the opposite end.  Any of these circumstances may lead a user to decide that a linear, a log taper, a reverse wired log taper, an antilog, or even some custom-tapered pot (see the Secret Life of Pots at www.geofex.com) best suits their needs, in spite of recommendations that others make.

My advice is to use what you have and decide if it suits your needs or not.  From that point, you can switch tapers or design your own taper.  Really the only times we are dead certain that we need an audio/log taper are when we know in advance that something will take huge changes to render audible differences near the higher end of the range (as with volume), and the only times we are dead certain that a linear taper is needed is when we need to be able to accurately identify the "middle", as with cut/boost or mix/balance/blend controls.  Everything else is up for grabs in my books.