Probably the best and most thorough explanation in terms that most pedal builders and tinkerers can relate with, this clears up a lot of the discrepancies between different articles and information shared about Beta(hFE) choices in fuzz or other circuits that rely heavily on certain values. It definitely explains why my own results have varied or matched with theirs over the past two decades of my electronics exploration of music related circuits, including the fact that I have used so many different types and brands of testing devices and methods!😅 Love the whole channel and content.
Love your videos, dude. Straightforward, no BS, and well explained. And RG is the man. I remember skimming his site 20 years ago without really grasping a lot of it. Wish I'd buckled down and really learned it back then. lol
Thanks for this info! You’re an oracle of knowledge; the way you pedagogically summarize the the information shows depth and levels of understanding. Great!
Hi Joe, I have built a small tester based on R.G. Keen's method of Germanium transistor testing. If you like I can send you one for your test bench. Best Regards, Ray
I bought a DCA 55 and I was disappointed in it because it gave me reading that were much too high due to the 5ma current. I thought it was not taking the leakage into account and giving me false readings, until I figured out the RG Keen method is using closer to 1ma. Thanks for sharing this, it will be useful for beginners to know that the PEAK meters aren't necessarily useful for this and can save them a lot of money.
The peak can still be used, you just have to scale up the recommendations for other people. If we assume, for example, the recommend gains of Q1:90 hfe and Q2: 120 hfe for a fuzz face were done with the RG Keen circuit, then we can roughly guess that you would need something like Q1:120 hfe and Q2:160 for transistors tested with the DCA55/75. Also worth mentioning that I believe you can do a more controlled test with the computer program peak has.
I contacted Peak Atlas and expressed my concerns about the high current being used to test transistors, and suggested that they either lower it to the more standard 1mA, or make the current user adjustable. I also requested that they add a curve trace test the plots Vbe vs. Ic. Unfortunately, they were not receptive to either. :(
Great video!I'm working on a 'Deacy' type amp and the 'higher-gain' germanium are not as important.As you said,it's the application and how much 'distortion/break-up' you are after!
Hi Joe, great channel, I recently subscribed. We need more folks doing this sort of deep diving into electronics and pedals. I would like to respectfully point out one error however, and it is unfortunately a misconception (mis-information) that is being propagated by so many, and still being taught in school, and it is regarding the notion that BJT's are current driven devices. This is in fact incorrect. BJT's are actually transconductance devices (Vin/Iout), just like their FET and tube counterparts. BJT's do require base current to operate, but this is an artifact (non-ideal parasitic) of how they operate. Diodes too are forward biased by voltage, not current. This is why I asked Peak Atlas to offer a Vbe vs. Ic trace plot option in their software, but they were not receptive. I also asked for the option to select lower currents for BJT testing. Again, not receptive. Anyway, BJT's are biased by Vbe voltage, not current, as their is a direct mathematical relationship between Vbe and Ic, and this relationship is scaled by temperature as per the following: Ic = Is * (e^ (Vbe/Vt) - 1), where Vt is the thermal voltage (26mV @ room temp) that is dependent on the ambient temperature via: Vt = KT/q (K=Boltman's constant, q=electron charge, and T=temperature in Kelvin). So, not only can you see that Ic is dependent on the ratio of Vbe to Vt, but also that there is no "input" current considered whatsoever. Note that "Is" is saturation current, and is a fixed value and device dependent.
This care with methods and measurements opens a premium++ market for germanium fuzzes. I wonder how many boutique pedal builders do these tests to keep consistency on a pedal lineup.
It's essentially the same as assembling any other PCB, minus the attaching of the breadboards, which is just double sided tape. The PCB is the protoboard from PedalPCB.
This is great. I grabbed an M tester and was curious how I could find leakers, seeing as it doesn't... With this video, I now have a tiny board I can dedicate to leakage testing and complete the test list. Thanks man!
Here's the PNP instructions b/c this video is long. R1 2.2M at B R2 2.472K at C 9v: - at R1/2 ends and + at E DMM: DC V across R2 Note V w/ R1 (X Volts) Note mV w/o R1 (0.XX Volts) subtract small # from big and x100 = HFE
The RG Keen circuit uses a variable resister that you have to tweak because 9V isn't a perfect divider for ratio it's like 2.472k ohms or something. At exactly 9V not 9. something or 8. something. Then there's a simple ohm's law formula but that's the fun right?. I found the Soviet's have almost zero leakage but lack some character sound wise. imagine that.
Thanks for the video. I just wanted to say the M-tester I got does test for collector to emitter leakage (lceo). It's different than yours. It's a "TC1" model with rechargeable battery.
GRAY BENCH, You mention if you replace non leakage transistor in fuzz circuit designed that was designed for leakage current that it will cause more problems and won't work correctly. Can you make a video lesson on how to include the leakage current into the formulas for biasing a transistor amplifier? because you have to include the leakage current of the transistor and add this to the formulas for choosing the correct biasing resistor ratios. If you replace all the leaking transistor with Non-leaking transistor the biasing circuit will have a biasing ratio that won't work with non-leakage transistors.
Cool channel. I test my germs crudely with a cheap meter. I set resistance to the 10k range, auto won't work for this. Black to base, red to emitter for NPN, reverse leads for PNP. Anything over 2k resistance is leaky and rejected.
Schematic calls for hfe=175 @ Ic= 5.0ma. My DCA55 only measures at 2.5ma. When I measure I get hfe=174 @ Ic=2.5ma, Vbe=0.723V, Ib=4.210ma, 0 leakage. How would you calculate the hfe at double the ma’s? I can’t find a formula for this anywhere, just a lot of other dummies like me asking questions.lol I figured if someone knew how to handle this it would be you!
There isn't a formula. The rate of change of hfe vs collector current isn't a constant. Generally hfe will increase with Ic, but by how much will change for each transistor. Something doesn't make sense with your listed specs though. If your hfe is 174, and hfe is the ratio of Ic/Ib, that would mean 2.5mA/4.2mA, which isn't 174. Regardless, The hfe is almost definitely going to be higher than 175 at 5mA. If the circuit is really hfe sensitive, I would look for another transistor. Otherwise I'd just send it personally :)
Hey guys, I guess I need help. I'm new to germanium transistors, but not afraid of elaborate testing to learn. I'm of course using the breadboard technique, but I do vary both resistors, as datasheets tend to post data using higher currents than typical for guitar pedals. My actual problem is the timing of the leakage current test. If I let it stabilize for quite some time at a collector resistor of 100ohms (much lower than shown here, i know), then measure the "base-on" current, it gives me hfe values of several hundreds with Rb=100kOhm. but if I measure the falling egde when pulling the base switch to off, my multimeter shows gradual falling of values within a seconds or so. if I do use values immediately after the switch-pull (which I assume to be closer to actual leakage current under base-on conditions), hfe calculation goes down to a single digit number. similarly, If using the incredibly high base resistor value as in the schematic cited in this video (I haven't seen a guitar pedal that high a resistor?) , hfe also is in the single-digit region. Transistors in question are AC125's non labeled w.r.t hfe; now I know nothing as before, and dont know how to rate my seller on ebay^^. I would be grateful about a hint on whether there's something to my close-to-base-on conditions assumption, or if I'm just measuring the stabilizing feature of my multimeter? Any other help appreciated. Im happy to share spreadsheet if anyone 'd like to see. Peace to the Planet!
To get only the leakage of 30 pcs of AC126s I made the modified setup setting power rail to 10V lab voltage (current limited) and using 1k ohm collector resistor. That way I was able to read the milliamps directly being the same number as the voltage change over the collector resistor. Those AC126s are mostly bad only 6 having it under 300 (micro!)amps. Is it ok to use 1k in this case?
Thanks, this is a good start for me. I mess with tube amps but know nothing about transistors! I got a B&K model 162 transistor/ FET tester for free and wanted to mess with it and see if it works - but you have to have some knowledge to set it up. Lol How accurate are these vintage testers? Also what do you think of the DCA55? They’re pretty cheap on DigiKey now.
Vintage testers can be fine, but worth checking if any of the old components inside or out of tolerance. I used the DCA55 for a couple years before upgrading to the DCA75. It's worth the money.
I have one of those m testers and it works for silicon, but every single germanium i put in (npn or pnp) reads as being a resistor. Could that be because of high leakage or is my tester half busted lol
This is likely a really dumb and oversimplified question, and although I know very little about this I find it very interesting...so my question borne of naivete and ignorance is this... Can you take two lower-gain pnp germanium transistors, and solder the legs to each other (in series perhaps?) to increase the effective gain, by sort of making the two lower-gain transistors effectively operate as a single higher-gain transistor? So, for example, transistor A is 80 and transistor B is only 80 as well. Transistor C is 60. So, you use transistor A at 80 for your position 1 transistor in your fuzz face (or whatever fuzz circuit you're working on), ...can you then solder the legs of transistor B and C to each other in some way to add them together to make a 140 transistor (B 80 plus C 60) to act as one 140 gain transistor for your position 2 transistor in you fuzz face circuit?
Yes, it is called a Darlington Pair (look it up) the gain of the two transistors is multiplied together. The downside is that the leakage is also multiplied so you need to use a couple of low gain and low leakage transistors ideally.
Probably the best and most thorough explanation in terms that most pedal builders and tinkerers can relate with, this clears up a lot of the discrepancies between different articles and information shared about Beta(hFE) choices in fuzz or other circuits that rely heavily on certain values. It definitely explains why my own results have varied or matched with theirs over the past two decades of my electronics exploration of music related circuits, including the fact that I have used so many different types and brands of testing devices and methods!😅 Love the whole channel and content.
This is about the best run-through of the RG Keen method, thanks for simplicity and clarity.
Another option is Aideepen multi-function tester, available for under $30. Similar to M-tester, but returns hFE, Vbe, Ic, Ice0, Ices.
Love your videos, dude. Straightforward, no BS, and well explained.
And RG is the man. I remember skimming his site 20 years ago without really grasping a lot of it. Wish I'd buckled down and really learned it back then. lol
Thank you, agreed!
Hmmmm, I would love to design a pedal oneday it's very interesting. Great vid........cheers.
Thanks for this info! You’re an oracle of knowledge; the way you pedagogically summarize the the information shows depth and levels of understanding. Great!
Thank you!
Very valuable info. Thank you for posting.
Hi Joe,
I have built a small tester based on R.G. Keen's method of Germanium transistor testing. If you like I can send you one for your test bench.
Best Regards, Ray
I bought a DCA 55 and I was disappointed in it because it gave me reading that were much too high due to the 5ma current. I thought it was not taking the leakage into account and giving me false readings, until I figured out the RG Keen method is using closer to 1ma. Thanks for sharing this, it will be useful for beginners to know that the PEAK meters aren't necessarily useful for this and can save them a lot of money.
The peak can still be used, you just have to scale up the recommendations for other people. If we assume, for example, the recommend gains of Q1:90 hfe and Q2: 120 hfe for a fuzz face were done with the RG Keen circuit, then we can roughly guess that you would need something like Q1:120 hfe and Q2:160 for transistors tested with the DCA55/75.
Also worth mentioning that I believe you can do a more controlled test with the computer program peak has.
I contacted Peak Atlas and expressed my concerns about the high current being used to test transistors, and suggested that they either lower it to the more standard 1mA, or make the current user adjustable. I also requested that they add a curve trace test the plots Vbe vs. Ic. Unfortunately, they were not receptive to either. :(
Great video!I'm working on a 'Deacy' type amp and the 'higher-gain' germanium are not as important.As you said,it's the application and how much 'distortion/break-up' you are after!
Hi Joe, great channel, I recently subscribed. We need more folks doing this sort of deep diving into electronics and pedals. I would like to respectfully point out one error however, and it is unfortunately a misconception (mis-information) that is being propagated by so many, and still being taught in school, and it is regarding the notion that BJT's are current driven devices. This is in fact incorrect. BJT's are actually transconductance devices (Vin/Iout), just like their FET and tube counterparts. BJT's do require base current to operate, but this is an artifact (non-ideal parasitic) of how they operate. Diodes too are forward biased by voltage, not current. This is why I asked Peak Atlas to offer a Vbe vs. Ic trace plot option in their software, but they were not receptive. I also asked for the option to select lower currents for BJT testing. Again, not receptive. Anyway, BJT's are biased by Vbe voltage, not current, as their is a direct mathematical relationship between Vbe and Ic, and this relationship is scaled by temperature as per the following: Ic = Is * (e^ (Vbe/Vt) - 1), where Vt is the thermal voltage (26mV @ room temp) that is dependent on the ambient temperature via: Vt = KT/q (K=Boltman's constant, q=electron charge, and T=temperature in Kelvin). So, not only can you see that Ic is dependent on the ratio of Vbe to Vt, but also that there is no "input" current considered whatsoever. Note that "Is" is saturation current, and is a fixed value and device dependent.
Good info, thanks!
This care with methods and measurements opens a premium++ market for germanium fuzzes. I wonder how many boutique pedal builders do these tests to keep consistency on a pedal lineup.
Any chance you can do a vid on how you constructed the test bed. It's very cool.
It's essentially the same as assembling any other PCB, minus the attaching of the breadboards, which is just double sided tape. The PCB is the protoboard from PedalPCB.
This is great. I grabbed an M tester and was curious how I could find leakers, seeing as it doesn't... With this video, I now have a tiny board I can dedicate to leakage testing and complete the test list. Thanks man!
Thank you!
Another excellent, informative video. Thanks joe
This video clarified so many things for me, all the lights upstairs are on now. 😉
Glad to hear!
thanks for this very helpful for me as a newbie fuzz lover and now trying to build and fix some fuzzes.
Thank you!
Here's the PNP instructions b/c this video is long.
R1 2.2M at B
R2 2.472K at C
9v: - at R1/2 ends and + at E
DMM: DC V across R2
Note V w/ R1 (X Volts)
Note mV w/o R1 (0.XX Volts)
subtract small # from big and x100 = HFE
Thanks Joe .. for doing this !
Thank you!
Thanks for posting great content!
Thank you!
The RG Keen circuit uses a variable resister that you have to tweak because 9V isn't a perfect divider for ratio it's like 2.472k ohms or something. At exactly 9V not 9. something or 8. something. Then there's a simple ohm's law formula but that's the fun right?. I found the Soviet's have almost zero leakage but lack some character sound wise. imagine that.
Thanks for the video. I just wanted to say the M-tester I got does test for collector to emitter leakage (lceo). It's different than yours. It's a "TC1" model with rechargeable battery.
Good to know, thanks!
Thanks for the info regarding the TC1. Just picked one up from AMZ.
GRAY BENCH, You mention if you replace non leakage transistor in fuzz circuit designed that was designed for leakage current that it will cause more problems and won't work correctly. Can you make a video lesson on how to include the leakage current into the formulas for biasing a transistor amplifier? because you have to include the leakage current of the transistor and add this to the formulas for choosing the correct biasing resistor ratios. If you replace all the leaking transistor with Non-leaking transistor the biasing circuit will have a biasing ratio that won't work with non-leakage transistors.
Cool channel. I test my germs crudely with a cheap meter. I set resistance to the 10k range, auto won't work for this. Black to base, red to emitter for NPN, reverse leads for PNP. Anything over 2k resistance is leaky and rejected.
Schematic calls for hfe=175 @ Ic= 5.0ma. My DCA55 only measures at 2.5ma. When I measure I get hfe=174 @ Ic=2.5ma, Vbe=0.723V, Ib=4.210ma, 0 leakage. How would you calculate the hfe at double the ma’s? I can’t find a formula for this anywhere, just a lot of other dummies like me asking questions.lol I figured if someone knew how to handle this it would be you!
There isn't a formula. The rate of change of hfe vs collector current isn't a constant. Generally hfe will increase with Ic, but by how much will change for each transistor.
Something doesn't make sense with your listed specs though. If your hfe is 174, and hfe is the ratio of Ic/Ib, that would mean 2.5mA/4.2mA, which isn't 174.
Regardless, The hfe is almost definitely going to be higher than 175 at 5mA. If the circuit is really hfe sensitive, I would look for another transistor. Otherwise I'd just send it personally :)
The TC1 tester is also very cheap and it tests leakage. Mostly consistent with RG Keen's method and much faster.
Good to know, thanks!
Hey guys, I guess I need help. I'm new to germanium transistors, but not afraid of elaborate testing to learn. I'm of course using the breadboard technique, but I do vary both resistors, as datasheets tend to post data using higher currents than typical for guitar pedals. My actual problem is the timing of the leakage current test. If I let it stabilize for quite some time at a collector resistor of 100ohms (much lower than shown here, i know), then measure the "base-on" current, it gives me hfe values of several hundreds with Rb=100kOhm. but if I measure the falling egde when pulling the base switch to off, my multimeter shows gradual falling of values within a seconds or so. if I do use values immediately after the switch-pull (which I assume to be closer to actual leakage current under base-on conditions), hfe calculation goes down to a single digit number. similarly, If using the incredibly high base resistor value as in the schematic cited in this video (I haven't seen a guitar pedal that high a resistor?) , hfe also is in the single-digit region. Transistors in question are AC125's non labeled w.r.t hfe; now I know nothing as before, and dont know how to rate my seller on ebay^^. I would be grateful about a hint on whether there's something to my close-to-base-on conditions assumption, or if I'm just measuring the stabilizing feature of my multimeter? Any other help appreciated. Im happy to share spreadsheet if anyone 'd like to see. Peace to the Planet!
To get only the leakage of 30 pcs of AC126s I made the modified setup setting power rail to 10V lab voltage (current limited) and using 1k ohm collector resistor. That way I was able to read the milliamps directly being the same number as the voltage change over the collector resistor. Those AC126s are mostly bad only 6 having it under 300 (micro!)amps. Is it ok to use 1k in this case?
Sure as long as you use 1k as the resistance in your V=IR calculation for the leakage.
Hey Joe, have you ever done a teardown on an EHX Germanium 4 Big Muff ? Thanks for the show.
No, but I'll put it on the list!
BTW where did you score the little component clamp? Was that off of a Drok tester?
Those are called ZIF sockets (zero insertion force). You can find them from mouser, digikey, etc., also probably amazon/ebay.
Thanks, this is a good start for me. I mess with tube amps but know nothing about transistors! I got a B&K model 162 transistor/ FET tester for free and wanted to mess with it and see if it works - but you have to have some knowledge to set it up. Lol
How accurate are these vintage testers?
Also what do you think of the DCA55? They’re pretty cheap on DigiKey now.
Vintage testers can be fine, but worth checking if any of the old components inside or out of tolerance. I used the DCA55 for a couple years before upgrading to the DCA75. It's worth the money.
I have one of those m testers and it works for silicon, but every single germanium i put in (npn or pnp) reads as being a resistor. Could that be because of high leakage or is my tester half busted lol
Probably high leakage. RG Keen's tester should still work.
@@graybenchelec I was wondering if it was that, I'll have to look into that
Once i'v heard that Brittainium transistors are even better 😉
Leekwidge? lol 4:27
This is likely a really dumb and oversimplified question, and although I know very little about this I find it very interesting...so my question borne of naivete and ignorance is this...
Can you take two lower-gain pnp germanium transistors, and solder the legs to each other (in series perhaps?) to increase the effective gain, by sort of making the two lower-gain transistors effectively operate as a single higher-gain transistor?
So, for example, transistor A is 80 and transistor B is only 80 as well. Transistor C is 60.
So, you use transistor A at 80 for your position 1 transistor in your fuzz face (or whatever fuzz circuit you're working on), ...can you then solder the legs of transistor B and C to each other in some way to add them together to make a 140 transistor (B 80 plus C 60) to act as one 140 gain transistor for your position 2 transistor in you fuzz face circuit?
Yes, it is called a Darlington Pair (look it up) the gain of the two transistors is multiplied together. The downside is that the leakage is also multiplied so you need to use a couple of low gain and low leakage transistors ideally.