PNP leakage experiment

[mozz]

Scored some of these B714 transistors. Well,ok,400 of them for $30, surprised he took my offer. They are for sale on a few places on the net, these were on Ebay and the guy said "probably not usable for a transistor". I happened to find a single in my stash before these arrived from the post office. Same date code, kind of a weird coincidence. I don't know what it was out of.  Makes me think these were some late production PNP Ge that was using up some old crappy stock. 2SB714 is rated for 6w. Really high gain and really high leakage. Really high C-B capacitance,measured on a lab LCR meter,  like 150pf, most Japanese trannys are 50pf and some US stock are 10, 25, 50pf.



So recently i was building (experimenting) a few fuzzs (1 knob) with q2e as the only variable, so tore that out put a 1k there and went back to my standard q1c trimmer. Soldered in some B714's.




I got the smart idea to add another tranny on the back of Q1 B&E as E&B to counteract the leakage. These leak so much if you touch them it goes up to 1.5ma. I think the q1c trimmer was pretty low to get q1c at .57v and q2c at 4.6v, might have been as low as ~3k.





Well, browsing google about some high leakage i see a RG post, he talks about using C/B across B and ground (which would be E). Never saw that before as all talk about reverse transistor usage is usually B/E and E/B. So i figure make 2 fuzzes one with B/E attached and one with C/B attached. Both q2c at 4.6v. Sound exact the same to me with a strat. Transistors are pretty gainy, heavy, loads of compression at 10 (too much unless you are playing death has already taken your soul and ground it into a fine powder metal), super long sustain with only a tiny bit of gating at the end , has some cleanup but you still get a edge (which i like), and doesn't get real clean until 1-2. Could drop your pickups to the pickguard with these.

Heated them up (slightly) using some hot air. No jokes please. The BE reversed board Q2C went up to 5.5v while the C/B board only went up to 5.2v. I need to do more experimenting maybe let them cook longer and let the entire unit get warmer. More to come. Anyone ever hear of these transistors before? They are pulls, real short leads but have seen some with full length leads on Ebay. Pictures of some DCA-55 values.

  I think i have a FZ1 here with sockets and a MKII with sockets. Will try them.

[Rob Strand]

I got the smart idea to add another tranny on the back of Q1 B&E as E&B to counteract the leakage.
These leak so much if you touch them it goes up to 1.5ma.

You've done well making those things usable.  If you have 400 sacrificing a few isn't a bad trade-off.

I think the q1c trimmer was pretty low to get q1c at .57v and q2c at 4.6v, might have been as low as ~3k.

If the transistor gain is high you would be better off adjusting Q2's collector resistor.

On the FF the main reason Q1's collector resistor affects Q2's collector voltage is because of the voltage drop across the 100k resistor due to Q1's base current.   If Q1 is high gain the base current is small and the effect on Q2's collector voltage is small.   It's only when Q1 has a crazy low collect resistor that the Q1's base current starts to look the same as a the base current or a low gain Q1.   3k vs 33k is compensating for 10 times the gain.   Another effect of increasing Q1's collector current is Q1's VBE increases.

Well, browsing google about some high leakage i see a RG post, he talks about using C/B across B and ground (which would be E). Never saw that before as all talk about reverse diode usage is usually B/E AND E/B. So i figure make 2 fuzzes one with B/E attached and 1 with C/B attached.

Try both and see which is more stable with temperature.    The leakage depends on the junction voltage drop.   In circuit the BC and BE voltages of the active transistor aren't the same so the BC leakage from the main transistor won't exactly match the "dumped" leaked from the added transistor.  You might find either BE or BC junction works best.   At the end of the day you want the one that tracks best with temperature.  If you just wanted to dump leakage you can use a resistor but it probably won't track as well and may well affect the sound different.

Heated them up (slightly) using some hot air.

Heat  guns and hair dryers are great for this (although you have to factor in larger parts are going to heat-up slower than smaller parts).

[mozz]

"If the transistor gain is high you would be better off adjusting Q2's collector resistor.
On the FF the main reason Q1's collector resistor affects Q2's collector voltage is because of the voltage drop across the 100k resistor due to Q1's base current.   If Q1 is high gain the base current is small and the effect on Q2's collector voltage is small."

So, high gain transistor on q1, better off trimming q2c. Lower gain, better off adjusting q1c. I really like adjusting q1c and I've probably done that on 20 fuzzes. I'm going to have to find a cutoff gain point, maybe anything over 150 trim q2c, more experiments, thanks. Maybe i should drop the feedback resistor to get a bit more clean.

I think i need to go back to stock 33k and 8.2k and see where q1c is at. The way i have it trimmed now, my voltages looked great. q1 (e 0) (b .091 maybe tiny bit low) (c .571)     q2 (e .522) (b .571) (c 4.60) (don't mind all the decimals, got a new HP3456A).

[Rob Strand]

So, high gain transistor on q1, better off trimming q2c. Lower gain, better off adjusting q1c. I really like adjusting q1c and I've probably done that on 20 fuzzes. I'm going to have to find a cutoff gain point, maybe anything over 150 trim q2c, more experiments, thanks. Maybe i should drop the feedback resistor to get a bit more clean.

Having a gain cut-off is a good idea.  For low gains both methods will work but for high gains I reckon Q2's collector resistor is better (for reasons given).
 

I think i need to go back to stock 33k and 8.2k and see where q1c is at. The way i have it trimmed now, my voltages looked great. q1 (e 0) (b .091 maybe tiny bit low) (c .571)     q2 (e .522) (b .571) (c 4.60)

Yeah, there's no problems with voltages. You could run with it like that.   Be interesting to compare the tweaked 33k vs 3k.

The base voltage on Q1 is pretty much determined by the specific transistor and current.   A large transistor operating at 1mA is going to have a low Vbe; sort of like comparing the voltage drops of a 1N5404 vs a 1N4148 at 1mA.

(don't mind all the decimals, got a new HP3456A).

Great score.   I'm jealous    I miss the decimal points for checking drift.

[mozz]

Appears using the C-B of the reverse transistor, across the B-E of Q1works a small bit better than the reverse e-b/ b-e. You can actually make the Q2C voltage go up or down by using your thumb and forefinger on each q1. These are all out the same bag and lot number so i doubt i will get a perfect null with anything else.

Using stock value resistors, Q1C is very low .060v. Trimming Q2C will get me 4.6v (56k) but does basically nothing to Q1C. So for now i am sticking with Q1C trimmer, it was about 5.6 to 6.0k so not as low as i thought before. Only other way to get Q1C a bit higher was to add a emitter resistor of 1.8k. My boards have a provision for that and a cap but it changed the sound and i didn't compare. Still doing a few more things but i think the reverse transistor soldered under the board is going to work fine for this high gain high leakage deal.

[Rob Strand]

Appears using the C-B of the reverse transistor, across the B-E of Q1works a small bit better than the reverse e-b/ b-e. You can actually make the Q2C voltage go up or down by using your thumb and forefinger on each q1. These are all out the same bag and lot number so i doubt i will get a perfect null with anything else.

Your trick might be as good as it gets.  You can only go so far it's like matching two different things.   

Using stock value resistors, Q1C is very low .060v. Trimming Q2C will get me 4.6v (56k) but does basically nothing to Q1C.

Yes, that's expected since Q1's collector current is set by the Q1 collector resistor.   If you reduce Q1's collector resistor by a factor of 10 then Q1's collector current will increases by a factor of 10 and Q1's Vbe increases.   The amount of Vbe increase is roughly 26mV * ln(10) = 26mV * 2.3 = 60mV.     Q2's collector resistor only affects Q2's collector current has no effect on Q1's .   I

So for now i am sticking with Q1C trimmer, it was about 5.6 to 6.0k so not as low as i thought before. Only other way to get Q1C a bit higher was to add a emitter resistor of 1.8k. My boards have a provision for that and a cap but it changed the sound and i didn't compare. Still doing a few more things but i think the reverse transistor soldered under the board is going to work fine for this high gain high leakage deal.

Adding the emitter resistor will change the gain.   Technically you should bypass the 1.8k with a large cap to restore the signal gain.    Adding an emitter resistor helps the overall stability of the bias quite a bit but it's not often done on the ff.    Gus posted link to "the other forum" some time back to the Yack Fuzz schematic.  IIRC it used bypassed emitter resistors fuzz-face type of circuit.