Simple Phaser – Going Back Full Circle

It’s time to go back full circle … and actually go back nearly to Phase 45 clone. Well, turns out I was more off than I normally like to be on the original pedal 😅. If this picked your interest, keep reading, or as usual just jump to the video and come back for details later.

In this post, besides going (not so) deep into theory, I added the improved schematic, and I also did a simple mod.

A Bit of Theory

While I got some nice sounding effects previously, I wanted to make it better. Little theory is a dangerous thing as the saying goes, and sometimes we all like living on the edge, right? (I’m sure the saying is about something else but let’s roll with this) I’ll be looking at how JFET behaves in omhic/triode region. This is important because we’re trying to use our JFETs as variable resistors and that’s where we want them to be.

Ohmic Region

There are three regions of operation of a JFET (I’m just reading this from Wikipedia 😊): breakdown (bad), constant current (amplifiers) and triode/ohmic/linear (which we’ll use).

If we look at a diagram of J109 (a handy one I had opened):

The area highlighted there is where V_DS < V_GS-V_P, In this area the drain current is directly proportional to the drain-to-source voltage and it’s regulated by V_GS , (well at least at the straighter part of it it’s linearish). Phew, that was a long sentence.

Less V_GS, steeper the line on the graph, less resistance. I’m sure you can see that from the graph 😎. Say we keep V_GS around -0.1V, if V_DS varies between 0.5 and 1V, I_D current varies between 60 and 110mA – exactly like a 8 to 9ohm resistor. If V_GS is -0.5V, we have I_D between 50 and 80mA, like a 12ish ohm resistor.

J109 is not a great example, but if we keep V_DS very low we can control resistance better. Since obviously I can’t really read the graph that easily and figure out resistance, let’s see some formulas.

JFET Resistance

Some relevant formulas for calculating resistance:

I used various sources for these (mostly this). Make sure to use absolute values for V_GS and V_GS(off). We start with formula for drain current in this region, transform it to get our resistance and if V_DS is very small we get that final simplified formula.

With this, since we measured our JFET parameters we can plot a resistance graph against V_GS:

r_ds is on y-axis and it’s a function of V_GS on x-axis. Or you can see it directly on desmos site. These are 6 pairs I tried out in the last post. I just shifted graph’s x-axis so they are all close together otherwise they wouldn’t fit on the image.

How to read this graph? Well, for the pair 1, the red graph shows that if V_GS varies from -1.3V to nearly -1.39V for example, the resistance goes from 1K to 9Kohm. The point of all these graphs being, the resistance change is non-linear close to V_P, and it shoots up pretty quickly.

Addition of 24Kohm resistor in Phase 90 in parallel helps limit the resistance (note that J113 & BF256B have different pinout).

But it does not change the response that much (I zoomed it in here because the max resistance is 24K). You can play with the graph here.

I used 10K because that’s what’s used in Phase 45, but that was a mistake … I’ll come to that 🙂 It still works, but some lower frequencies are not swept through. The thing with above is, BF256B was definitely the right choice (Green & Purple diagrams), but also, the sweep through lower frequencies is super quick.

This is what bothered me a bit so I went to research how to improve it, and maybe it will sound better. It is important to also note that the swing of the LFO is relatively large compared to less than 300mV swing in the above graph.

Improving JFET as Variable Resistor

Hmmmm, at this stage I should really point out that this is my understanding of all of this … and I might be completely wrong. Feels like I should’ve made this disclaimer way before in the post 😁.

Anyway, I wasn’t lazy, I knew Phase 45’s schematic is slightly different so I looked into some application notes. Namely, I looked into AN105 by Sliconix, the hint is in the title: FETs As Voltage-Controlled Resistors 😲. The document is from ’97, so maybe too modern for Phase 45, no? Also, it says J111 is a popular choice!!! Fools!

Here’s an excerpt:

The other one I looked at was AN-6603 from Fairchild named A Linear Gain Controller Amplifier. That one is from 1975 … hmmmm, some of these topologies look familiar:

This was all slightly over my head at this stage, but looks like the fellas who designed the pedal knew what they were doing, so I decided to go back and fix the schematic to be more like the original Phase 45.

One thing though, I don’t know if the engineers used any of these application notes when designing the pedal (Phase 45 was released in ’74 after all), but I know I would. Anyway, the whole exercise gave me some better understanding and appreciation for this pedal.

Also, I said in the video I won’t be using J111 … was I a fool? From the documents I read, greater V_P and lower I_DSS gives you better linearity and control (we saw that in the diagrams above with BF256B for sure). I would be careful with that though, all of the docs were in the context of amplifiers, maybe not so good for phasers, but definitely food for thought.

Fixed Schematic

All of that theory was too long!!! Here’s the fixed schematic. In reality, I only added 4 new resistors and 2 capacitors, but all of a sudden the breadboard became way more crowded!

The only difference from the last time is around JFET:

You can see here 2 extra resistors and a capacitor (per transistor) compared to before. Now, the minimum frequency for the notch goes to about 150Hz, which is in line to what it would have been if I used 24K resistor instead of 10K the last time. The same value as Phase 90 uses. Doh, I can’t believe I did not try that out on video! But we got to the same place eventually 😆.

Here’s the breadboard diagram, slightly more crowded this time around:

And if you want to play around DIY file here it is:

And here’s updated SPICE:

Easy Mod No. 1

Finally, I’ll add a mod. It’s a simple mod, a LED flashing at the rate of the effect frequency. It uses a bi-colour LED, so it flashes green (or red, or depends on your LED and how you wire it) when not engaged, and it’s red with yellow flashing (or whatever else you got) when engaged:

Trying it Out

Here’s the video, I cover theory just a bit and then it’s breadboarding and trying it out + the mod: