For this build I chose Fetzer Valve from runoffgroove.com. It is a simple effect, but it is quite different to previous ones. One of the goals of the effect is to emulate response from a Fender 12AX7 input stage. This brings with it certain limitations and requirements when it comes to the design on top of using JFET.
In this article I’ll get to know the circuit and measure JFET parameters to be able to get final version of the circuit. This is now needed to get particular response out of the effect as I’ll cover below.
I’ll then run SPICE simulation and then I’ll finish it off with breadboarding the effect. In the follow-up article I’ll prepare the enclosure and complete the pedal on a protoboard, but before doing that, I’ll need to do some improvements on the circuit (oh … that might’ve been a spoiler).
Getting to Know the Effect
There are several diagrams on the effect’s page and I chose “revisited” version of it. The page goes into more detail on some of the design choices, but it is good to do something similar myself. From educational perspective and also in case I need to do some adjustments to the circuit.
There are some peculiarities of the circuit. Because it has a specific goal in mind – emulating the response of a 12AX7 tube as used in a typical Fender tube amp, some of the resistor values depend on the JFET transistor parameters. This is why I did not mark Q1 transistor, nor Rs and Rd resistors.
Let’s go through all the sections of the effect real quick.
On the schematic there is only a single coupling capacitor – C2. The value is chosen so it does not affect the frequency response. High pass filter that it forms with RV1 potentiometer has about 33Hz cut-off frequency.
There is no input coupling capacitor. We can get away without it since the reference voltage is 0V, but in case we’re chaining effects and there’s a DC component coming from the previous stage that might mess-up our biasing. That’s why an optional 22nF capacitor was suggested on the effect’s page.
The gain of the amplifier is fixed, hence volume control. Just a 100K log taper potentiometer.
Input Low Pass Filter
Resistor R2 and capacitor C1 form a low pass filter on input, this filters out frequencies higher than 20KHz outside of audible spectrum.
That leaves us with the amplifier itself. JFET is biased differently from BJTs and MOSFETs I used previously and the configuration used here is self-biasing common source amplifier. Good source of information is Vishay’s AN102 application note. It is comprised of resistors R1, Rs, Rd and of course the transistor Q1.
It is called self biasing because it does not require an extra power supply. VGS is now determined by voltage across RS. Since gate is kept at 0V, VGS becomes -VRs which is equal to -ID*RS. Knowing IDSS and VP parameters we can determine operational conditions using equation:
ID = IDSS*(1 – VGS/VP)2 = IDSS*(1 + ID*RS/VP)2
So knowing parameters for my transistor is going to be used to set the operational conditions. And knowing ID I have all the operational conditions I need.
Measuring JFET Parameters
As mentioned above, to determine best values for source and drain resistors, I need to know JFET parameters. The effect’s page also has a neat circuit for doing the actual measurement of the parameters.
In the video at the bottom of the article, I go through this in depth, but I’ll go through details here real quick.
On the left hand side is the full circuit. It has an on/off switch, and a switch to choose between measuring IDSS and VP. Digital multimeter should be connected to (+) and (-) points to measure voltage between them. It has to be digital multimeter for this measurement since it has very high input impedance that should not affect the reading.
In the upper-right corner is the equivalent circuit for measuring IDSS. In it, VGS is 0V so the current through 100 Ohm resistor is IDSS and can be measured via voltage reading on the digital multimeter. If the voltage reading is 1V, for example, the current through the resistor is 10mA. So to get the current I need to multiply the meter reading by 10, and I get my IDSS in milliamps.
In the bottom-right corner is equivalent circuit for measuring VP. The leakage current going through transistor, when the transistor is in pinch-off mode, is very low – 1nA order of magnitude. This will not be enough to cause such a voltage drop over the very large resistor to actually keep the transistor turned off. Thus, VGS will be just slightly above VP and the transistor will be turned on, but only just. Voltage read directly on the multimeter will be very-very close to VP.
I have a bunch of transistors and among them J201, clear favourite for guitar pedals. The problem is that J201 nowadays is coming in SMD packaging (at least my supplier has them only in SOT23 packaging) and that is a real pain to measure. Thus for this exercise I used BF256B and J112 that are coming in TO-92 packaging (although with different pin-out, so be careful about that).
Application note AN-6609 – Selecting best JFET for your application is a good read on … well, how to select your JFET.
In reality, I got all THD JFETs from my local supplier that I could conceivably use, so that’s what I measured (I have a couple more devices that I did not show on the video). I can solder SMDs but it is a real pain for doing beforehand measurements like this. In the end of the day, if the sound is good and the effect meets all the goals of the build, what device was used in the build does not really matter.
It is easy to see why J201 is so prominently used. It has relatively tight range for VP and IDSS parameters and circuits can be designed such that using any J201 transistor will still work, despite the variation. BF256B has way wider range for the parameters, but it is still manageable. For J112, maximum values for IDSS are not even listed for example. All that being said, this is totally irrelevant if I’m measuring parameters anyway, but just thought to point it out.
Anyway, I did the measurements for a bunch of transistors and I chose one BF256B and one J112 with following parameters for my circuit:
VP is actually quite good that it is relatively high because it determines the maximum input range I can support. IDSS is relatively high, especially for J112 so that might not be so great when using it in a pedal, but we’ll see about that in a minute.
To finalize schematic, I need to plug in measured parameters into a handy calculator on the effect’s page and like magic I get the optimal resistor values for my source and drain resistors.
The above screenshots are calculations done on the effect’s page. For BF256B, I get 267 ohms for source resistor and 245 for drain resistor. Since I have 270 ohm resistor I’ll use that for RS, and for RD I’ll combine 200 ohm and 47 ohm resistors to get 247 ohms.
The screen also gives me expected ID – quiescent drain current of about 5.66 mA, and tells me that expected amplification is actually attenuation. The input signal will be roughly halved.
For J112, I was using 22 ohm resistor and 47 ohm + 22 ohm = 69 ohms, but if you look at the video, when I measured VS and thus ID, and VD , I got results that differ from what I got from the calculator. I rather got lower drain current which was actually in line with my hand calculated values. Anyway, since RS is what is supposed to get us desired output response and RD is there to optimize headroom, I opted for using 47 ohm resistor instead of what was calculated.
The final schematics that I’ll be using are thus:
I decided to leave out the potentiometer since neither of the circuits are actually amplifying the signal. Also, I included optional 22nF input coupling capacitor since I might be chaining effects so just in case. Also note the different pinout for BF256B and J112.
Again, I have to note that these schematics are specific to the devices I have. Other devices, the same parts but different physical transistors, might not work at all in this configuration.
The effect’s page has some simulation results presented there. I’ll go through the SPICE simulation but with my parts and suggested resistor values. Let’s do J112 version first. The only reason for this is that I have J112 already in the library.
When I run the AC analysis I get pretty flat response as I expected:
The results are roughly what we expected based on the calculator having in mind that I chose higher resistor value for RD from the recommended one. Actually, my calculation puts the gain around -9dB and the graph shows around -9dB – spot on.
Now, all models are probably slightly wrong, they can’t possibly capture characteristics of all physical devices. Since I have particular device in mind let’s see what are the actual parameters of the model and how does that compare to what I have.
Vto is threshold voltage. It’s not exactly my VP but it’s not either miles off. In the model it says: -2.057, that’s way smaller (in absolute value) than what I have. Also, Beta is transconductance coefficient (but not transconductance). 5.695m for Beta means absolutely nothing to me since I have IDSS so I need to somehow calculate it. My back of the envelope calculation gives me very roughly 3.7m.
I think these parameters are much closer reflecting my device than what came with the stock model. By no stretch of imagination this should be used for more serious than I’m doing here. For what I’m trying out this is pretty decent approximation. Here’s a guide on how to model the whole JFET in PSpice if anyone feels doing it.
I can override some of the stock model parameters using ako command in SPICE and plugging in my parameters. Thus, I used: .model my_J112 ako:J112 Vto=-3.73 Beta=3.7m
My new SPICE schematic now looks like this:
To use my J112 model instead of the stock one, hold CTRL and right click on the JFET, that will allow you to change the model.
Ta-da … and the graph looks … the same. What a bummer, having said that, this was expected, right? I did expect around -9dB. In fairness, it is slightly different, the gain is around -8.5dB. I did have a look at quiescent drain current though and the stock version of transistor is working off 10.7mA, my version of the model makes that 20.8mA, so the diagram might be misleading because operating conditions are certainly not the same.
Let’s look at transient analysis so.
Transient analysis J112
I tried to get the transient analysis graphs side by side:
Not much of a difference beyond higher drain current for my device – and that is along the expected values. It is a bit difficult to see anything in this graph. Any distortion of the signal is hard to spot. If the circuit was clipping for example, that would be visible immediately, but subtle colouration not so much.
The main take, I suppose is – the circuit works 🙂 I’ll go into some extra analysis in a minute.
For BF256B I went through a similar process. This time around I had to get the model online. I used one from LTWiki, another option is InterFET (sadly no models on ON Semi’s website). Here’s the circuit in SPICE:
All the above discussion stands, I had to calculate Beta myself, but ultimately, AC response looks pretty much the same for both versions of BF256B model:
Difference in operating drain current is smaller than for J112 as one would expect: 2.8mA (stock version) vs 4mA (my model version). Gain roughly the same about -6.9dB.
As for J112, for BF256B version of circuit there’s no noticeable distortion, etc. Let’s move on to something more interesting.
Sounds esoteric, no? This is just analysis of spectrum of frequencies in my guitar sound – with or without effect. I hooked up output of the breadboarded effect directly to my audio interface and recorded me plucking 4 E tones – open 6th string, 2nd fret on 4th string, open 1st string and 12th fret on 1st string.
I did this with clean sound, with J112 and then with BF256B version of effect. Once I had that recorded I used Sonic Visualiser (free software) to do some really basic analysis.
Click on the image to get full version of it that you can zoom in. Essentially, the clean sound just shows that my guitar has rich harmonic content – that’s why it sounds like a guitar. If it only had single frequency in it – it would sound like those old handheld video games from previous century.
Interesting thing is to see that Fetzer Valve emphasizes some of the harmonics. I extracted just E4 (open 1st string) from the above image – side by side:
As noted on the effect’s page both odd and even harmonics are slightly emphasized so there might be something in this. Notice how both versions of the effect have 2nd, 3rd and 4th order harmonic frequency more red than in the clean sound spectrogram. Anyway, let’s move on.
Characterizing the Effect
In order to describe the pedal, let’s list out some operating characteristics (all values simulated in SPICE):
|about 3.4Kohm (@1KHz)
|about 3.4Kohm (@1KHz)
|Fixed about -6.8dB
|Fixed about -9dB
(without LED indicator)
|4 mA on average
|20 mA on average
While these parameters look alright, gain is less than 1 (0dB) so I can’t really call it a booster.
Note on input and output impedance. Theoretical input impedance of self biasing common source JFET amplifier would be R1 (1Meg) and output impedance would be around RD (47 or 247 ohms). This effect has additional components (like capacitors and additional resistor) that will affect this impedance, thus the difference in the table.
Breadboarding the Effect
Finally time to do the breadboarding. Here’s one of possible ways to do it, and the one I used in the video below.
I found a non-standard breadboard lying around, but it was really handy, allowed me to organize my 2 effect neatly. I had to draw it out of vero boards in DIY Creator since the standard ones can’t be configured.
You can see it in action in the video below. Here’s the diy file:
The video is showing how I tested JFETs and then tried out the breadboarded effect. It’s a bit longer but I was kind of describing in more detail what I was doing. It is split in to chapters so you can skip parts you’re not interested in.