Recent Experience with Chinese and Soviet Germanium Transistors (Measurements Included!)

[PedalBuilder]

I've had a bunch of fun lately building (and playing through) silicon fuzz pedals, so I decided I'd branch out and have some fun with germanium, too. I picked up several different models of Soviet transistors from alexer1 on eBay and on a whim also bought a set of 100 Chinese 3AX31C germanium transistors from UM2part on AliExpress. All of them arrived surprisingly fast (10 days from China and 27 days from Ukraine despite an ongoing war). After they arrived, I tested them with my trusty Peak DCA55, and recorded the results.

First, a picture of the happy Germanium Family:


Now on to the overview.

3AX31C - Chinese PNP germanium transistor in a TO-1 or similar package. The ones that I received appeared to be an assortment of different lots, as there were several minor variations of the markings on the packages and the legs had varying degrees of oxidation. I paid ~$40 for 100 transistors shipped from Shanghai to the US. Out of the 100 transistors that I received, 1 was dead (it appeared to the DCA55 as a common anode diode network), 16 had leakage >500µA, and 9 had extremely low gains (hFE <10). The remainder had hFE ranging from 77-420, with an average hFE of 164 and an average leakage of 160. To me, the 3AX31C seems to have more in common with western transistors than the Soviet ones—the pinout is the usual EBC, the collector isn't connected to the case, and the transistors have more leakage. How will that translate sonically? Looking forward to finding out. Depending on how these work out, I might try a box of the 3BX31B, an NPN cousin of the 3AX31C, which the same seller had in stock.

P416B - Soviet PNP germanium transistor in what I'd describe as a "large UFO" package. These all had really low leakage; 11 out of the 22 measured as having 0 leakage and none with more than 17µA leakage. (The DCA55 does not seem to be able to detect leakage below 9µA, so they all probably have some leakage, just not very much.) All of the transistors that I received tested as functional, with hFE ranging from 62 to 120 and an average hFE of 86.

GT115D - Soviet PNP germanium transistor in what I'd describe as a "small UFO" package. These also had a some sort of plastic insulator at the bottom of the transistor body; not sure why. These were largely higher gain and leakier than the other Soviet transistors; the hFE ranged from 146 to 280 and averaged 203. Leakage ranged from 139µA to 311µA, with an average of 214µA.

1T320V - Soviet PNP germanium transistor, the military version of the GT320V, in the large UFO package. Higher gain than the P416B, but still very low leakage, with none measuring over 27µA. Gains ranged from 82 to 149 with an average of 120, and leakage ranged from 0 to 27µA, with an average of 19µA.

GT309B - Soviet PNP germanium transistor in the small UFO package. Similar gain and lower leakage relative to the 1T320V, with an average gain of 116 and an average leakage of ~11µA. Gains ranged from 78 to 163 and leakage from 0 to 15µA.

Individual measurements are below for those interested:

Spoiler: Individual Measurements

3AX31C			3AX31C(cont)			P416B			GT115D		1T320V			GT309B
hFE	Leakage		hFE	Leakage		hFE	Leakage		hFE	Leakage	hFE	Leakage		hFE	Leakage
420	339		143	112		120	10		280	139	149	23		163	15
417	258		137	93		112	14		252	237	138	20		161	0
396	345		137	84		101	14		209	164	138	27		141	14
323	159		136	128		100	17		202	177	134	20		140	14
297	124		136	166		96	10		200	256	125	25		134	9
280	108		135	67		93	15		196	228	120	16		128	10
270	125		134	164		90	15		196	214	119	16		116	0
265	237		133	117		88	0		182	236	114	19		115	13
258	211		133	168		88	0		167	311	110	13		111	0
205	419		132	83		86	0		146	178	110	11		103	0
204	182		132	135		86	14				106	27		94	10
199	110		131	103		86	10				82	0		88	13
197	113		131	89		84	10							86	14
195	199		127	71		82	0							81	13
195	143		127	112		81	10							78	13
194	155		127	114		80	0
189	128		126	218		74	0
182	133		125	97		74	0
174	82		125	324		68	0
174	183		124	107		67	0
167	128		122	66		63	0
166	109		121	150		62	0
165	133		119	102
164	85		118	333
161	374		118	118
160	145		117	243
155	131		114	107
154	108		114	77
154	264		112	72
152	143		111	50
151	113		109	141
151	107		108	465
148	133		107	42
148	128		102	84
147	118		95	56
144	176		90	454
144	121		77	481

 

[finebyfine]

Really cool results - especially for the 3AX31C as I don't have any firsthand experience w/ Chinese germanium transistors or see them discussed as much as Soviet transistors. I'm really curious about the NPN 3BX31B and might snag some up myself. I've bought from alexer1 on ebay as well - good seller imo.

The google sheet link is asking for me to request permission from you to view it. Unless you have charts in there it might just be easier to use a bbcode table generator and hide it behind a spoiler tag so it's not massive or something.

 

[PedalBuilder]

Really cool results - especially for the 3AX31C as I don't have any firsthand experience w/ Chinese germanium transistors or see them discussed as much as Soviet transistors. I'm really curious about the NPN 3BX31B and might snag some up myself. I've bought from alexer1 on ebay as well - good seller imo.

The google sheet link is asking for me to request permission from you to view it. Unless you have charts in there it might just be easier to use a bbcode table generator and hide it behind a spoiler tag so it's not massive or something.

That worked, more or less. I had to take rearrange a few things to get it under the 10,000 character limit (even blank cells take up a few characters), but the data's there now.

I'm surprised that I don't see the Chinese germanium transistors talked about as much, too. In terms of gain/leakage/pinout/package, they seem like much closer substitutes for western germaniums than their Soviet brethren. The gain range is interesting—I'm not familiar with any circuits that call for germanium transistors with gains of 250 or higher. Definitely some room for experimentation here, I'll probably look at some of the fuzz circuits designed for silicon transistors for inspiration.

 

[finebyfine]

@PedalBuilder definitely is surprising to me too given the results. I know I've seen the 3AX31C both for sale on small bear and discussed on his pedal guides but that's about it. Do you know of any resources cataloging Chinese transistors? I've had this Soviet transistor page bookmarked forever, would be cool to see something like that.

 

[PedalBuilder]

@PedalBuilder definitely is surprising to me too given the results. I know I've seen the 3AX31C both for sale on small bear and discussed on his pedal guides but that's about it. Do you know of any resources cataloging Chinese transistors? I've had this Soviet transistor page bookmarked forever, would be cool to see something like that.

I've found a few floating around the internet; I've attached the one that I found to be the most comprehensive. There's also a thread from 16 years ago on the DIY Stompboxes forum that lists some of the different Chinese germanium transistors and their parameters.

 

[andare]

Some juicy ones in there. You can build tone Benders, peppermint fuzzes and all kinds of cool circuits that require high leakage and high gain.

I'm Jelly!



I've also bought quite a few transistors from the same Ukrainian vendor but none of those you got. I'll add them to my next order.

 

[PedalBuilder]

Update:
After spending several hours playing through a breadboarded Fuzz Face and a Peppermint Fuzz, I can confirm that the 3AX31C sounds and behaves exactly like a good germanium transistor. No crackling, hiss, or other problems. I used two high gain (hFE ~265 for Q1 and ~419 for Q2) 3AX31C for the peppermint fuzz, and got a wide range of raucous, buzzy, spitty 60's fuzz tones. For the fuzz face, I paired a P416B with an hFE of 88 and no leakage with a 3AX31C (hFE ~130 and leakage ~160). The result was classic, wooly Fuzz Face tone that cleaned up nicely with the volume knob on my guitar. I swapped some of the other transistors in and out with similar results after adjusting the collector resistors, and everything sounded great. It's only one lot, but I feel comfortable at this point recommending that people give the 3AX31C a try. I'm sufficiently impressed that I've ordered a set of the 3BX31B NPN transistors; I'll post an update to this thread once they arrive.

 

[sonic_explorer]

I was having a hard time processing the stats @PedalBuilder posted, so I made a few quick plots to show the distributions - I thought others might find them useful also. The colors used for each transistor are the same in all the plots.

Scatterplot of leakage vs hfe:

Histograms of hfe and leakage (I cut the plots off to make them more readable - but the highest leakage and hfe outliers are not shown):

 

[andare]

Update:
After spending several hours playing through a breadboarded Fuzz Face and a Peppermint Fuzz, I can confirm that the 3AX31C sounds and behaves exactly like a good germanium transistor. No crackling, hiss, or other problems. I used two high gain (hFE ~265 for Q1 and ~419 for Q2) 3AX31C for the peppermint fuzz, and got a wide range of raucous, buzzy, spitty 60's fuzz tones. For the fuzz face, I paired a P416B with an hFE of 88 and no leakage with a 3AX31C (hFE ~130 and leakage ~160). The result was classic, wooly Fuzz Face tone that cleaned up nicely with the volume knob on my guitar. I swapped some of the other transistors in and out with similar results after adjusting the collector resistors, and everything sounded great. It's only one lot, but I feel comfortable at this point recommending that people give the 3AX31C a try. I'm sufficiently impressed that I've ordered a set of the 3BX31B NPN transistors; I'll post an update to this thread once they arrive.

Curious how you biased the transistors in the Peppermint.
AFAIK Q1c has a 100k trimmer on it, Q2c uses 1k + 3k3 + 25k trimmer. The rest is a stock Sunface, so a FF with a 1uF input cap and a 250k volume pot.

 

[PedalBuilder]

Curious how you biased the transistors in the Peppermint.
AFAIK Q1c has a 100k trimmer on it, Q2c uses 1k + 3k3 + 25k trimmer. The rest is a stock Sunface, so a FF with a 1uF input cap and a 250k volume pot.

I used the Dirtbox Layouts schematic for the Peppermint, so I just biased the transistors using trimmers. For the fuzz face, I was too lazy to put trimmers on the board and played around with the collector resistor values until they biased correctly. One thing to note is that the Peppermint is ridiculously temperature sensitive. It's a great sounding circuit though, so I'm going to try the RG Keen leakage matching trick to see if I can reduce the effect of temperature swings:

If you match the c-b leakage of two devices, you can use one as a normal transistor, and the other with the emitter open, connected collector to the base of the "normal" one and base to ground.

If the leakages are similar, the diode connected one shunts about as much leakage out of the base of the active one as leaks in from the collector. The active transistor then behaves more like it's normal.

Ideally, you have a good gain device with a leakage matched to a bad gain device; use the good gain device as the normal transistor, the bad one as the leakage-sucker. And be very careful in measuring leakage, putting devices in the test socket with pliers or tweezers or something. Your hand heat is enough to throw the leakage test off to the high side.

 

[andare]

I used the Dirtbox Layouts schematic for the Peppermint, so I just biased the transistors using trimmers. For the fuzz face, I was too lazy to put trimmers on the board and played around with the collector resistor values until they biased correctly. One thing to note is that the Peppermint is ridiculously temperature sensitive. It's a great sounding circuit though, so I'm going to try the RG Keen leakage matching trick to see if I can reduce the effect of temperature swings:

I have the same schematic for the Peppermint. I meant to say what voltages did you end up with? Same as a Fuzz face?
And yes, leaky transistors are very sensitive. All germaniums are, I've changed gain and leakage just by breathing on them

 

[PedalBuilder]

I have the same schematic for the Peppermint. I meant to say what voltages did you end up with? Same as a Fuzz face?
And yes, leaky transistors are very sensitive. All germaniums are, I've changed gain and leakage just by breathing on them

My favorite sounds were with the Q1 collector voltage between 0.5 and 0.7 volts and Q2 collector voltage between 4 and 7 volts. There's a very wide range of useable sounds depending on how you set the bias, so it's worth using external bias pots for the Q1 (B100k) and Q2 (B50k). The bias trimmers are interactive, so adjusting the bias on one transistor changes the bias on the other. I'd add that I also had great results with the Peppermint circuit using two darlington pairs of 1T308B transistors. They're military-spec Soviet transistors with gains between 40 and 55 and no measurable leakage, and they are not very temperature sensitive. I used a pair with a total hFE of 1,845 for Q1 and 2,805 for Q2; total leakages for both pairs were under 50 µA. For that version I added a 220pf cap between the Q1 base and collector to tame a bit of oscillation that occurred when the fuzz knob was turned all the way up.

 

[andare]

My favorite sounds were with the Q1 collector voltage between 0.5 and 0.7 volts and Q2 collector voltage between 4 and 7 volts. There's a very wide range of useable sounds depending on how you set the bias, so it's worth using external bias pots for the Q1 (B100k) and Q2 (B50k). The bias trimmers are interactive, so adjusting the bias on one transistor changes the bias on the other. I'd add that I also had great results with the Peppermint circuit using two darlington pairs of 1T308B transistors. They're military-spec Soviet transistors with gains between 40 and 55 and no measurable leakage, and they are not very temperature sensitive. I used a pair with a total hFE of 1,845 for Q1 and 2,805 for Q2; total leakages for both pairs were under 50 µA. For that version I added a 220pf cap between the Q1 base and collector to tame a bit of oscillation that occurred when the fuzz knob was turned all the way up.

Very useful info. I always forget about Darlington pairs. I have many low gain low leakage transistors sitting here. Will investigate.
Thanks

 

[PedalBuilder]

Update on the 3BX31B NPN transistors: they were almost all duds. Out of the 100 transistors, all but 8 were dead or had an hFE <8. AliExpress was quick to issue a refund, but I did end up with 8 more or less useable NPN germanium transistors:

hFE	Leakage
97​	201​
82​	112​
66​	85​
66​	101​
64​	160​
58​	299​
53​	217​
46​	41​

 

[finebyfine]

Hfes around/sub 8 if they actually still work could make for some interesting darlingtons

 

[szukalski]

You could also try them as diodes.

 

[MobyOctopad]

Update:
After spending several hours playing through a breadboarded Fuzz Face and a Peppermint Fuzz, I can confirm that the 3AX31C sounds and behaves exactly like a good germanium transistor. No crackling, hiss, or other problems. I used two high gain (hFE ~265 for Q1 and ~419 for Q2) 3AX31C for the peppermint fuzz, and got a wide range of raucous, buzzy, spitty 60's fuzz tones. For the fuzz face, I paired a P416B with an hFE of 88 and no leakage with a 3AX31C (hFE ~130 and leakage ~160). The result was classic, wooly Fuzz Face tone that cleaned up nicely with the volume knob on my guitar. I swapped some of the other transistors in and out with similar results after adjusting the collector resistors, and everything sounded great. It's only one lot, but I feel comfortable at this point recommending that people give the 3AX31C a try. I'm sufficiently impressed that I've ordered a set of the 3BX31B NPN transistors; I'll post an update to this thread once they arrive.

Sorry to ping an old thread, but was curious if the Peppermint requires/tolerates leakage higher than what's generally recommended on a standard Fuzz Face in addition to the higher than normal gain, or do you still need to shoot for high-gain but relatively low-leakage transistors?

 

[PedalBuilder]

Sorry to ping an old thread, but was curious if the Peppermint requires/tolerates leakage higher than what's generally recommended on a standard Fuzz Face in addition to the higher than normal gain, or do you still need to shoot for high-gain but relatively low-leakage transistors?

The leakiest transistors that I tried in the Peppermint had leakage of ~350µA. They worked, but the circuit got flabby if the temperature went up by a few degrees. I had much better luck with transistors with leakage closer to 100-150µA. One other thing—the 3AX31C are subtly, but noticeably darker sounding than the GT115D, so I ended up making a Peppermint and my own take on the Dark Peppermint. The regular Peppermint used GT115Ds—hFE/leakage for Q1 was 207/164 and 280/139 for Q2. For the Dark Peppermint, Q1 had hFE/leakage 199/110, while Q2 was 280/108. The leakier transistors worked really well for Q2 and Q3 of a Tone Bender Mk. II, so that's a circuit worth trying if you have leakier, high gain transistors.

 

[Coda]

MKI Tonebender and the FZ-1 both need leaky transistors. This is good news. Looks like I need to find me some leaky Chinese transistors…

 

[andare]

I have a batch on the way!