By Steve Daniels, Small Bear Electronics LLC
© 2010 By Small Bear Electronics LLC
- What transistor types are best suited for the FF?
They should be germanium if you want to get the classic 60's/70's "soft" distortion. Silicon has been used, and many people like the harsher clipping, but it is a different sound. Among germanium transistors, many type numbers (see the list below) are possible candidates for FF circuits and clones. However, careful sorting and testing are needed, because wide variations in gain and leakage among devices with the same part number are the norm.
- Why the wide variation?
Manufacturing techniques in the '50s and '60s when germanium was the material of choice were not nearly as precise as they ultimately became.
- What should the gains be?
The usual recommended range for the "classic" Fuzz Face is 70-85 for Q1 and 120-140 for Q2. However, if you like a sound that's more "squishy" and "compressed," I have found that a high gain device for Q2 (150-190) gives a good result with a Q1 in the range of 90-120.
- Do the devices have to be the same type number? Do they have to be any particular type number to guarantee great tone?
No and No. I have used pairs made up of completely different types; they work very nicely as long as neither device is too leaky and the gains are in buckets that permit correct biasing. Given the wide variation in characteristics noted earlier, the emphasis on getting particular type numbers is misplaced, IMO.
- I have a NNNN pedal. Will I get better tone if I replace the existing transistors with XXXX?
If the pedal already sounds good, it is hard to say. In the first place, your idea of "better" may not be mine, or anyone else's; personal tastes vary enormously. Beyond that, getting good results when modding a pedal requires knowing the characteristics of the pedal, the specs of the devices you intend to use, and often something about changing component values to make the marriage work.
On the other hand, I often get this question with respect to replacing cruddy, poor-quality or poorly-matched transistors in some "Name" commercial pedals. In that case, the answer is: Yes, if you know what you are doing. Be aware that the replacement is not nearly always a "drop-in"; it may involve trimming bias resistor values and/or adding transistor sockets to the board. If you are a beginner, you should do a lot of homework before proceeding or get knowledgeable assistance, and maybe both.
- How do I get a matched pair?
You can buy a bunch of devices (sometimes available from repair shops, wholesalers, surplus outlets or on-line auctions) and experiment, or pay a few bucks for a pair of devices that someone has matched for you.
- What should they cost?
If you get lucky (or you're very persistent), unmatched devices can still be had for surplus prices. However, just as you no longer find old Martin guitars going begging in flea markets, word of the value of certain electronic devices has spread among those who have them. For the moment, decent-quality old-stock parts are still available, if not as dirt-cheap as they once were.
In general, the price per transistor will rise commensurate with scarcity of the particular type. Beyond that, a reputable seller should tell you whether the devices are matched and audited, completely unsorted, or "rough-sorted." Caveat Emptor! Especially if you are buying a wholesale lot, know your seller! Fakes of famous type numbers have become rife, and many lots will turn out to have been "cherry-picked." It is depressingly easy to get burned even buying something you think ought to be a real score--like that lot of 2N404s that you thought would make great Tonebenders, but they all turned out to have gains in the 40s.
- What devices do you use to create your pairs?
My present U.S.-made raw stock consists of branded and unbranded, old-stock devices in TO-5 cans. The branded PNP parts are mostly 2N404, 2N404A or 2N1305, and the unbranded are roughly similar to the old 2N1307. My Chinese devices are 3AX31C. I sort and test to the following spec: With a 9-volt supply, collector-emitter leakage under 300 microamps at room temperature with the Base open. With 9 microamps base current, minimum "real" gains of 80 for Q1 and 110 for Q2. I also audit every pair to eliminate hissy devices, and I provide a record of the bias resistances I used when testing. See the Stock List
- Which of the pairs you offer is best for upgrading a Fuzz Face?
Any of the matched pairs will work, and will sound good to most ears if properly biased. "Best" is invariably a matter of personal taste. See my guarantee/warranty below.
- Which of the pairs you offer is best for building a Fuzz Face?
Any of the matched pairs will work, and will sound good to most ears if properly biased. "Best" is invariably a matter of personal taste. IMO, as long as the devices aren't excessively leaky and hit reasonable gain buckets, the make of the transistors is much less important to the tone than the values of the input and output capacitors. If you are seeking the holy grail, learn to breadboard (See the articles in How-Tos) so that you can see which ones grease your crank. Also see my guarantee/warranty below.
- Why would two devices with roughly the same gain and leakage sound different?
To paraphrase R. G. Keen: "There is no such thing as "mojo" or "magic" in the physical universe." However, since knowledgeable players have noted this difference, we can reasonably say that there are internal electrical properties of some transistors (internal capacitance has been suggested as one) that do affect the perceived tone, and don't appear in the published specs.
In my opinion, no. The difference in price is driven entirely by the relative scarcity, differing costs and yield of useable parts of the raw stock that I start with. That said, pairs made from devices that came from different manufacturing environments Will, often, sound different from each other. Knowing that these devices are as individual as snowflakes, and that customer tastes vary enormously, I offer the guarantee in the next FAQ.
I guarantee that your pair will sound "like a Fuzz Face" if used with the resistor values I recommend. The devices can be returned for credit or refund as long as they are not soldered and have full leads. Try them out on a solderless breadboard and/or use transistor sockets to make sure you're happy.
In any lot of raw stock, there are always only so many parts in the 70-85 and 130-140 buckets that everyone wants. A 20% yield is is typical. So, in order to keep these devices affordable for DIY, I don't guarantee to provide specific brands and my allowable range of gains is necessarily wide. Q1 may vary from 80 to 110, and Q2 may go as high as 199. We fill orders by picking from a "grab bag," so what you get is strictly luck of the draw.
If you have to have a pair in the "classic" range, consider buying a small bulk lot (SKU 0115 for the spec, SKU 0115A for the smaller bag) and sorting your own. You can also troll E-Bay, with the caveats that I mentioned earlier. I guess the bottom line is that you Can find the holy grail, but it will, inevitably, cost some money and probably time, as well.
This has come up a lot in recent years with the spread of small-volume pedal-making. I do not discount matched pairs, because there is no economy of scale in the testing. Whether I create one pair or a dozen, it takes the same amount of hand work per pair.
We cannot provide quantities of audited devices to commercial builders, and we purposely limit the number of pieces that we will sell to any one customer. For the boutique trade, I do offer rough-sorted, bulk lots of unbranded devices. Please see the Stock List for available types and specs.
Bulk lots also exist with many wholesalers, both here and overseas, but their minimum orders are usually out-of-reach for hobbyists. Those shops are sometimes alternatives for commercial builders who have the need, the ability to fund a large order, and a very high tolerance for risk. You will get some leads to sources if you do a Google search on "Germanium Transistors." But be aware that the market is totally unregulated, full of trash and very difficult to navigate.
This list of "candidates" may help. It is a compilation of things I have seen on other bulletin boards (Aron Nelson/Jack Orman/R. G. Keen), and my own experience with Japanese and American parts.
American JEDEC numbers
PNP: 2N404, 2N404A, 2N508, 2N527, 2N1305
NPN: 2N388, 2N1306, 2N1308, 2N1309, 2N1373
American "Replacement" Types
NTE (or ECG) 101, 102, 158
PNP: 2SA or 2SB 22, 54, 75, 77, 172, 175, 178, 187, 201, 303, 324, 370, 405, 439, 516
Of these, the types that have given me the highest yield of "good" devices are the 2SB175 and 178.
PNP: OC44, OC71, OC75, OC77, OC81, AC122, AC128, AC151, NKT275
According to R. G. Keen, the physical characteristics of germanium made it much easier at the time to manufacture PNP devices rather than NPN. The demand for non-leaky, temperature-stable NPN devices of consistent gains was one of the factors that drove germanium out of the picture in favor of silicon. The result is that, today, there are large, fairly dependable supplies of old-stock PNP germanium on which I can draw. I have secured some stock of U.S. and European NPN types, and I will make these available as I am able to get testing and auditing done.
I have a few OCxx types, and we offer matched pairs of some. If you do go looking for European numbers, buyer beware! The old-stock devices have become very expensive, and fakes are common. AFAIK, devices marked NKT275 are all re-labeled. They may or may not be decent quality--that depends on what raw stock the seller started with--but they are almost surely not the same devices that Newmarket sold back in the day. The modern re-makes of some AC12x have a poor reputation for quality.
I have no experience, but on-line sources report that many of these are both good and inexpensive.
Absolutely. Thousands of these were sitting unused for decades in warehouses in Shanghai, and I have a large lot of 3AX31C (PNP). They sound a bit more trebly to me than my U.S.-made stock, but I don't hear any "mojo" that would justify a huge price difference compared to other types. I suggest that you take any on-line hype with a large dose of soy sauce. However, they are very good quality, and suitable for any of your favorite classic builds.
We audit all pairs in the classic Fuzz Face circuit, and we guarantee proper operation in that circuit and with the bias resistor values that we specify. That said, any of our matched pairs can usually be made to work in any of the clones. However, be aware that the bias resistor values in the published schems usually presume that the transistors hit the "ideal" gain range. You can still get good results with pairs that don't fall in that range, but some tweaking of the bias resistors will be needed. If you are building something other than a classic Fuzz Face, testing on a solderless breadboard is highly recommended.
Some of these, like Joe Gagan's Easy Face, have become very popular. They offer a very wide range of tonal possibilities, and they tend to be less sensitive to temperature variations than the all-germanium FF. Regardless of type number, any device that meets the designer's recommendation for gain and isn't too leaky will work. I offer U.S.-made devices that have the right gains for the Easy Face and for Aron Nelson's Hornet. Again, as I am able to train people to do the auditing, I hope to offer a greater variety of specific device types.
The second and third stages are exactly the same as in the Fuzz Face, and the first stage needs one that measures from 75-100. Check the postings on diystompboxes.com for suggestions on how to tweak the bias. I offer a tested three-transistor set. See the Stock List.
If you get lucky. These devices (NTE, ECG, SK, etc.) are actually factory rejects from major transistor makers that have been sorted into rough (VERY rough) gain groups and re-labeled. In general, they are not a good buy; way too expensive, too variable in their characteristics and often excessively leaky.
Based on an explanation of the basics from R. G. Keen, I worked out a bare-bones method for getting an estimate of the static DC gain of a transistor, good enough for sorting devices for the FF and clones. You'll need the following:
The temperature sensitivity of older germanium transistors has to be seen to be believed; when setting up these tests, only pick up the devices using tweezers or wearing gloves. The polarities of the battery and meters are shown for PNP devices; they are reversed for NPN.
First, measure the leakage current:
It may take a few minutes for this reading to become stable. (I swear that some of the ones I have tested are reacting to variations in room temp and/or my body heat!) Record the figure on the meter. This is the leakage current.
Keen has noted that too much leakage makes suspect the long-term reliability of a device. How much is too much? My own spec has always been to reject any device that leaks more than 300 microamps at room temp.
Now apply forward base bias through the resistor:
Again, wait for a stable reading. Record the figure for collector current.
Do the following arithmetic to get Hfe:
(Collector Current - Leakage Current)
(all values in microamps)
The base current will be 9.0 (microamps) if the resistor is exactly 1 meg and the power source is really 9 volts. The normal 5% tolerance of a carbon film type will get you close enough, though R. G. Keen has pointed out that you can use the resistance scale of your DVM to select one that is closer.
Unless you know exactly how the DVM measures and calculates Hfe, you should presume that it doesn't take leakage current into account. This would give an incorrectly high reading of Hfe on a device that maybe you shouldn't be using at all.
The method described here is crude, but the only way to do better would be to spend money for a scope/curve tracer. The results you'll get from the bare-bones method are close enough to select pairs that will work well in the Fuzz Face circuit with no trimming. I have focused entirely on practicalities, so if you want an explanation of how the Fuzz Face works and why gain matters, go to the scriptures:
R. G. Keen "The Technology Of The Fuzz Face"