Transistor update! - AionFX Phobos Germanium Fuzz


In my previous post, I mentioned waiting on some nice Telefunken transistors. They finally arrived after being in transit for only 10 days (and I am still waiting for some PCBs from the US that were shipped on the 5th of June...) !

The original transistors

The transistors included in the kit were two Russian transistors (one MP16B and one MP21A), and a Tungsram AC125. They are relatively cheap and easy to find, and they perform admirably. 

Their hFE value were 62, 56, and somewhere around 90, respectively. These values are perfectly suited for the Tone Bender circuit.

Don't get me wrong: they sound great and the fuzz they generate is very clean, but I want a fuzz that is a little dirtier. I found a good article from Coda Effect where he built an earlier revision of the same PCB I have, and he used AC116 transistors. By some sheer luck, I managed to find the exact same transistors on eBay for about 20$, shipping included. What could go wrong?

The "new" transistors

Their heatsink looks nice. Standard 2N3904 for comparison.

They arrived yesterday, a good 10 to 15 days earlier than I expected. Compared to the others (and to a standard 2N3904 silicon transistor, as shown in the above picture), they are quite tall. I actually wondered how I'd make them fit in the limited space I have available. They are "new old stock" and were probably manufactured somewhere in the early 70's.

Turns out they fit just fine.

I first tested their hFE values, and they ranged between 54 and 98, which is right in the ballpark of where I expected them to be. I picked the 76 for Q1, the 84 for Q2, and the 98 for Q3, because more gain is better, right?

Well, no. Not really. Too much gain in the first stage will cause some weird problems with the saturation stage not reaching the correct voltage and gating heavily (i.e., blocking the signal since the voltage is out of tolerance). I spent the whole evening yesterday to try and figure out why Q3 barely reached 150mV instead of the 2V it is supposed to be. Even with the bias trimpot at 0Ω, almost no voltage gets through. Perhaps Q3 is bad (which I did not completely rule out yet), but I noticed an interesting behaviour.

Voltage issues

I removed the TFK transistors and reinstalled the original ones to check the voltages they have when in operation. The bias of Q1/Q2 was steady at roughly 3.5V, but the bias of Q3 is behaving weirdly. It starts up at roughly 1.8V and slowly rise up to 2V. As far as I know, this is somewhat normal since germanium transistor conductivity will change based on temperature. When in operation, its internal temperature will rise, and the conductivity will rise accordingly. If you put your finger or a screwdriver on the metal can, the conductivity will drop as the internal temperature cools down. It takes a very long time for the voltage to stabilize, as it keeps climbing 2-5mV every second.

In the 30 or so minutes I monitored the voltage, it never really stabilized and reached upwards of 3.5V. I'm not sure if that's normal or not, but it doesn't seem to cause any issues with the output. I put the TFK transistors back in, and monitored the voltage over the course of 30 more minutes. The TFK transistors are much more stable, but the voltage at Q3 never reached more than 300mV, except when I plug in an instrument and pluck the strings (where the voltage peaks at 3.5-4V). Q1/Q2 bias is rock-steady at 3.5V.

At that point, I'm assuming the transistor is bad, so I swap it with the AC125 that was in the kit. It works, but is doesn't sound quite right. It's still gating heavily, and the voltage is about 1.2V on startup with the trimmer potentiometer at 0Ω. Voltage still rise in time as it did previously, but even then, it never really stops gating out the signal.

I swap Q3 with the two other TFK I had left, and a similar behaviour is observed. Very low startup voltage with the trimmer potentiometer at 0Ω, gating, weird noises coming out of the amp. The two transistors I had to try had an hFE of 61 and 64, and a startup voltage of about 1.2V. With the heatsink attached on these transistors, the voltage is a lot more stable and doesn't rise as fast - it at all.

It's now late in the evening and I'm completely out of ideas, I let it sit for the night.


Even when on vacation, I'm still waking up way too early in the morning at 0430. While thinking over the entire situation this morning, I had an epiphany:
What if the circuit is made for low-gain transistors and starts misbehaving with mid- to high-gain transistors?

For some reason, this thought did not occur to me when debugging the day before. It makes sense when you think of it: the first stage is a Darlington-pair of transistors to amplify the input signal, and the second stage is a saturation stage, where all the nice fuzz is generated. Too much gain on the first stage will already saturate the signal which will cause gating and clipping and all sorts of strange things to happen in the second stage if the output voltage of the previous stage is not within the range of what the circuit expects.


With that in mind, I re-tested all the TFK transistors I had to confirm their hFE values. Out of the five in the pack, I picked the two with the lowest value for Q1 and Q2, with an hFE of 61 and 64 respectively. Then I picked the hFE of 84 for Q3. Once they were neatly installed, I plugged everything back in and checked the voltages.

The bias of Q1/Q2 is a little high at 4V, but a quick turn on the trimpot lowers it at 3.502V. I monitor that for a few minutes, and it's not going anywhere, which is great. Next, the bias of Q3 is at 4V! This is also great. I never really saw a voltage on startup much higher than 1.8-1.9V for Q3 with the trimpot at 0Ω. I dial it in about half way and the bias is now at 2.2V. Voltage is still rising, albeit slowly (which means the heatsinks are working as they should).

I plugged in my trusty P-Bass, and started playing around: no more impromptu gating or strange clipping. The sound has the slight imperfections you'd expect from a circuit like this, and the fuzz sounds great. At maximum fuzz it's thick and dirty, and it dials back nicely with either the fuzz potentiometer or the instrument volume knob (the latter can have some unintended side effects depending on the potentiometer installed in your instrument, but let's not get into that). I very much like the noises that comes out of this pedal.


As far as I'm concerned, I'm done working on this pedal. I don't think I will change the transistors again and I'm pleased with the results. It looks nice and plays great, even with the transistors bundled in the kit. Overall it was a great to build to do, and a good learning opportunity.


Popular posts from this blog

Overdrive, distortion, fuzz, hard clipping, soft clipping... What gives?

STM32 Development: Configuring Arduino IDE/Visual Studio to build and deploy to STM32-based boards

J.P.'s Gear Review v2.0 - Ep. 7 - Fender Hot Rod DeVille, or the conclusion of The Bassbreaker Saga