Germanium Transistors, Noise and Leakage

[squirrels]

I'm going to start by saying I have very little experience with Germanium Transistors, pretty much only having built a Crimsom Drive, Fuzz Factory, Oracle and a Sunface. The only transistors I've used are untested (going to be building a tester), using an AC128 in the Crimson, 2N404s in the Fuzz Factory and Sunface (same ones, didn't care for the FF) and TK3055s and 2N404s in the Oracle. All of my GE builds have been noisy regardless of how power was supplied (wall wart or battery). Would this be an effect of high leakage transistors? I know the lower the leakage the better but why? Is added noise the result of higher leakage? I know that there is probably a lot of info online but I think that getting actual feedback from people may be more beneficial.

Also does anyone have a good or even a decent source for Germaniums? Also any recommendations for Tonebenders and Fuzz Faces? Thanks.

[Travis]

The relation between leakage and noise would depend on the circuit the transistor is being used in.

In a general sense, more leakage does not equate to more noise

Low leakage is not always better depending on what you are building.

The Fuzz Factory and Oracle are both based on the Fuzz Face. In all 3 circuits, high leakage will generally alter the bias in a way that makes the pedal sound gated rather than noisy.

Leakage varies with temperature, and transistors with high leakage tend to be less temperature stable.

You should get a tester like the DCA55 to get a better idea of what's going on when you're selecting transistors

[rocket88]

+1 to getting a DCA55 for transistor testing. To add to what Travis has already said, there are some Ge transistors that are just noisy. Not to mention IME I've found that some of the higher gain pedals have happened to be noisier then lower gain ones, especially with Ge builds. Not saying that this is always the case, just what I've experienced. For example, in my oracle build I had used gt309 transistors in the gains of the original build that 112/304hfe respectively and it was a noisy fucker and sounded kinda like Si, as in harsher rather then smooth distortion. I swapped them for 2 mp41a transistors with both gains of ~100hfe. This not only gave me a good amount of fuzz, but it smooth, and damn near silent, and you can ask Travis he tried it during pedal-geddon 2016.

To make matters even more difficult sometimes it's not the model of th transistor, but the batch. I've had many Ge transistors that the noise varies with batch. One batch is great and quiet, while the next may sound great, but noisy hell. Not to mention in one build a set of transistors will be noisy, but in another they're not. Tbh, this is one of the reasons that got me first started on buying my Ge transistors in bulk, which was way before I started my company.

[Marbles]

I definitely agree on the peak atlas DC55. I got it last week and it's a breeze. I was intimidated a bit by Germanium and hated how my multimeter kept giving back different results.
I retested them with the peak and noticed my original measurements were way off.

Had another problem with silicons, tested the trannies and one said: there's a short between red and blue: Straight to the dumpster.
It's not the cheapest, but to see hfe and leakage that fast makes life a lot easier for me.

[squirrels]

Thanks for the info. So basically the higher leakage makes it harder to bias correctly, which could lead to more noise?

[Travis]

squirrels wrote

Thanks for the info. So basically the higher leakage makes it harder to bias correctly, which could lead to more noise?

No not at all. The role the leakage plays is completely dependent on the circuit in question. In the specific case of the the Fuzz Face, higher leakage will usually result in a gated sound (weak and farty with no sustain). Other circuits will react differently to leakage

[squirrels]

Ok, got it. I had totally the wrong idea then. The only experience I have with Germaniums is with what I have personally built, not even trying anything out in the store. Is background noise common for them? If not, how can I quiet them down?

[squirrels]

I did add a little extra filtering (100uf and 100r) to the Oracle's power supply which got rid of a whine but the white noise is still there.

[Beaker]

"Thanks for the info. So basically the higher leakage makes it harder to bias correctly, which could lead to more noise?"

No, higher leakage does not necessarily mean more noise.

Read Travis and Zach's posts again, as every point they make is true. You can make generalisations about Ge transistors, but there are few hard and fast rules that apply across the board. This makes selecting transistors tricky, and will always lead to a degree of trial and error, before you hit on the best transistor type for the circuit, and the best combination of gain and leakage.

As Zach points out, you can nail it on one pedal, then build another with the same type, gain and leakage, and have different results, if the second transistor is from a different supplier or batch.

Remember also, that Russian transistors usually read a lower HFE than they sound - if a circuit suggests 105 HFE for example,  a Russian transistor with a HFE of 60 might be a perfect match.

Good suppliers are our friend Maryam at bg-manue and ml electronics in Bulgaria, as well as Orpheus in Bulgaria. Other highly recommended sellers are SOVCOM and bird-SR71 in Ukraine, Elena in Russia and Valtek in Romania.

There are plenty of other sellers in Eastern Europe, some of which I have used, and I have yet to have had a problem with any of them. They all seem reliable so far.

[Travis]

Sometimes noise is just a side effect of having so much gain.. When you amplify your guitar a lot, you are amplifying any existing noise as well. So if you have a noisy guitar or amp that will make it harder to test.

You can make effects using germanium that don't have noise issues. It helps to have a variety of transistors to work with sometimes. It's all about selecting the hfe, leakage, and transistor type to do what you want. You can experiment with these different factors to try and get the best sound

Different transistor types tend to break up differently, have different frequency responses, etc in addition to varying hfe and leakage. Some transistors are darker sounding, some are brighter, some are cleaner, some break up sooner. You can use this to shape not only the noise level but also the fuzz character. In the case of too much noise, you may want to select a darker sounding transistor. Too muddy? Try something brighter, etc

If you're not testing the transistors and just popping them in, it's very possible you're not hearing the best that transistor type has to offer. You don't need to stick to RG Keen's hfe rules, but it helps to have an idea what you're putting in

[Beaker]

Look out for Russian GT and MP series transistors - the ones that look like flying saucers.

GT304, 307 and GT308 PNP
Most MP series PNP such a MP37 and MP39

For NPN, MP38 and especially MP38A if you can find some, and Tesla 104NU71 are the most useable in pedals.

[squirrels]

What about MP20? Any good?

[Travis]

I have a box of MP20 that are good but I tend to like the MP21A better

[squirrels]

I found some MP21A To try out from manue. I'm going to get some inverter chips, anyone deal with polida2008 before?

[Travis]

Inverters are a classic way to add noise problems to a fuzz face build

Sometimes they work ok, sometimes not.

Polida is not very well regarded but if you're not in a hurry you could always get your money back through eBay if they show up with a problem

[Beaker]

Yes, I'd either build PNP pedals, and live with the fact that they are the "wrong" polarity to the rest of your pedals, or build NPN versions, for which there are plenty of layouts available.

I've bought some good parts from Polida, but I've also bought some utter junk too!

[squirrels]

I could also flip the electrolytics and 9v and ground connections, right? A fuzz face circuit is so small it's worth a try. If it's noisy I'm pretty much where I am now with it.

I just put an order in for 21 MP21As, definitely looking forward to trying them.

[Travis]

You could reverse the electros and 9v and ground connections to use NPN transistors. You can try reversing the collector and emitter to use PNPs at that point but that doesn't usually work well.

To be honest it sounds like you're getting ahead of yourself trying to eliminate the noise issue by other means when it seems that you have used randomly selected transistors. I could be misunderstanding though

Generally speaking, introducing your wall wart will only make issues worse with the fuzz face

[Travis]

By the way, a couple tricks that work better than adding a cap from 9v to ground are...

Small caps between base and collector of the transistors. This rolls off high frequencies to eliminate noise. Too big of a cap and you're losing too much high end, to small and you're not cutting enough noise. Look at the DAM meathead as an example

Personally my preferred trick (for a fuzz face esque build with noise issues) is to add a cap to ground at the input, like you see in the Tone Bender MKII. I usually use 10n just like the MKII. This just dumps some noise to ground right at the front of the circuit and usually doesn't have a very noticable effect on the overall sound

[squirrels]

I didn't mean switching the caps and connections for noise reasons, I meant as opposed to using an inverter or a battery, like the Oracle and Silicon Fuzz Master. The transistors are random, which is why I wanted to know about testing and all. I just read that reversing the connections could make it noisy. I'll try adding caps to my Oracle and Sunface. Thanks for the tip.

Unless I'm still wrong (which I could easily be) the transistor gain can cause noise, not leakage. So lower gain, lower noise. It's no different than a silicon circuit. I'm going to read up on what RG says about it.