How to Calculate Transistor Leakage?

[Henry89789]

Arcane:

Well, you either have a place to apply the probes on the side of the box that is not visible in the pic. Or, you insert the probes into holes in the perf board showing through the opening on the top of the box.

[Arcane Analog]

Try again. Both places to connect the DMM are visible from the top. The schematic really could not be more basic. Look at where the +9V and -9V connect.

Building pedals is not just about painting by numbers. Understanding the basics and critical thinking are key skills.

[Henry89789]

Kwai Chang Caine:  Old man, how is it that you hear these things?

Master: Young man, how is it that you do not?

I think I figured it out. You use the transistor's "E" leg test "socket" and that black ground lead you got coming out the top. I tested it on mine. Works. 

Okay, that was a fun exercise. I started another thread about getting a breadboard fuzz face circuit to work right so I can use it to manipulate resistor values to bias Q2C. How about giving me some clues on that one to see if I can figure it out?

[Arcane Analog]

Yes - you figured it out which is much better than just throwing you the answer I think.

[PRR]

> The gain ...95 when cool and 98 when warm.
> The leakage was 0.497 when warm and 0.332 when cool.
> Observation: Gain increased a little but leakage increased substantially.
> Conclusion: Heat increases gain and leakage? Or are my observations and conclusions completely wrong?

I can't know if your observations are correct.

However it is *generally* true that current-gain rises, slightly, with temperature; and leakage rises _significantly_ with temperature.

Leakage will roughly *double* for every 10 degrees C of temperature. Same basic physical/chemical law that says paint dries faster in summer, hot transformers die sooner, "cold-blooded" (really no-thermostat) reptiles are sluggish in the morning and active in the afternoon.

The reason current gain rises is more complex, or rather I don't have a quick explanation on hand, and it may be the difference of two phenomena so it may only *tend* to rise, but could fall.

If you are curious:

> the air conditioner cycled on and off

Don't fool around with little temp swings. Make a cup of hot tea and a glass of ice-tea. Tack long leads on the transistor, then stick it in a small plastic bag. Make your measurements in the hot and cold tea. Allow time for the hot/cold to soak through the bag and into the transistor.

Silicon can stand more heat/cold than you can. (Soldering-iron on the case is a traditional quick-check for heat troubles, though can be a bit hotter than is really wise.)

Germanium, especially vintage, you should limit the hot to tongue-safe. If you spilled it in your lap, would you be warmly annoyed or _scalded_? Likewise don't try dry-ice; regular ice-box ice is plenty cold for vintage Germanium.

From 1 deg C ice-tea to 65 deg C hot-tea, you should see very clear change of leakage. If double per 10 deg C, over 65 deg you figure 2*2*2*2*2*2 = 60 to 90 times the leakage (instead of 1.5 times in your very-small temp swing).

[DiscoVlad]

I've had a couple of queries from people wanting the spreadsheet I made for calculating leakage.

So here's a link.

This is an openoffice document so probably can't be opened in Excel. You will need something which can read them, like Openoffice, from www.openoffice.org (it's free!).

[R.G.]

Openoffice spreadsheets will save as an excell (.XLS) spreadsheet. Do "Save as" and pick the appropriate file format.

However, I still philosophically object to using a nuclear spreadsheet to squash a fly. It's very nearly the level of arithmetic that can be quickly done in your head.

[DiscoVlad]

I was going to write an app, but hey it's only two equations. I was a bit loathe to share the calculator because they always seem to end up with people relying on the numbers that come out but without understanding the implications of them.

I agree though, if you do the maths beforehand it's a pretty simple matter to calculate what  voltage readings the various gain/leakage bands fall into. Then all that needs to be done is measure and sort.

[R.G.]

On the other hand, if you have a pile of transistors to sort and match, spreadsheets do make sense.

You can number each transistor in the pile with those little sticky inventory dots from an office supply store. Then you run each one through the tester, entering only the device number in one column and its reading in the cell next to it in the spreadsheet. When the pile is entered, then you enter the calculation in the next column and drag-copy it for the whole column.  With that done, sorting is done just by highlighting the rows and sorting on the column of interest.

After that, sorting for matching or where to cut off your selections gets really easy. You can have the spreadsheet construct graphs of it, etc.

[mac]

I have a HTML page in my gallery that does the maths for you 

RG Test maths:
http://www.diystompboxes.com/smfforum/index.php?topic=45481.0

mac

[Henry89789]

I didn't want to waste time getting the breadboard to work so I built a kind of Biasing Pedal. It has sockets for Q1, Q2, the 470, the 2.2, and trim pots for the 33k and 8.2k. I test the transistor for gain and leakage with the tester discussed earlier and then bias them with this  "Biaser."  There's just one more thing I don't understand. You can see from the pic that the numbers on the meter show that each transistor is within the ranges considered desirable for fuzz pedals. These readings were obtained by adjusting the trim pots. So the question is :   

Once you adjust the trim pots to get the desired readings how do you determine the ohms to which each trim pot was adjusted so you can install  those transistors with the appropriate resistors  "pre-biased" in another pedal?     

[LucifersTrip]

You can see from the pic that the numbers on the meter show that each transistor is within the ranges considered desirable for fuzz pedals.

that's mainly just for this specific fuzz pedal...and for some others based on this topology

Once you adjust the trim pots to get the desired readings how do you determine the ohms to which each trim pot was adjusted so you can install  those transistors with the appropriate resistors  "pre-biased" in another pedal?     

you can only measure the ohms out of circuit, so you need to disconnect one leg of the trim. In my Fuzz Face test box, I have a switch on the trimmer so I can take it in and out of circuit

[Henry89789]

Luciferstrip:

Thanks for the reply. How do you connect such a switch?  Or is it two switches?   One on each side of the trim pot, so you disconnect the trimmer from the circuit on each side  and then measure the resistance between the two switches ?

[Arcane Analog]

Socket the trimmers.

[LucifersTrip]

Thanks for the reply. How do you connect such a switch?  Or is it two switches?   One on each side of the trim pot, so you disconnect the trimmer from the circuit on each side  and then measure the resistance between the two switches ?

you need only one spst switch on one side of each trimmer you want to remove from the circuit. a spst has only 2 lugs, so you should be able to figure out how to wire it. a socket would work if you're not putting it in a sealed enclosure.

[Henry89789]

I finally got Arcane Analog's Germanium Hfe/Lkg Tester put into an enclosure. I didn't want to use up a nice pedal box and I found this old kaput lantern lying around ....



Not very compact but it works and its cool, even has a sort of hidden message about a mad Diogenes carrying a lantern looking for an honest germanium transistor. Thank you Arcane Analog for your assistance.

[RobertJay]

Hi Hey Hello. I know this is old, but is anyone still here? I am having trouble understanding this but think I do understand based on this thread. I'm building old school tone bender circuits with germanium transistors. The hFE is easy enough to test using a cheap tester, but the uA leakage is another factor. Tranny 1&2 have to be less than 100uA and #3 between 100uA-300uA. Ok, so here's what I figured out... I took my measurements with and without the Base, subtracted the 2 and divided that by .00247. So,  .33v-.06v=.27v/.002472=109uA... Yes?

And YES. I DID use the geofex link instructions, I DID build the tester with the resistors and used a power regulator for 9v on the nose. It is THAT link that everyone wants to share as the almighty answer but the language in it is SO confusing. I want to know if my numbers and way of getting them are right as it's never really simply explained. It's not! Nothing against Sir Keen, I'm pretty sure he started this whole website and forum we are on. But why does everyone act like it's explained in simple english, it's not. Thanks

[mozz]

The hfe you are getting on a regular meter is not true, due to ................the leakage, not being figured in. For a silicon transistor, yes it would be a good reading. Go to RG'S website geoeffects and see how to build a simple germanium tester. That being said, even testing silicon transistors on various meters, you will get different readings, due to each brand of meter using a different testing current.

[Rob Strand]

The hFE is easy enough to test using a cheap tester, but the uA leakage is another factor. Tranny 1&2 have to be less than 100uA and #3 between 100uA-300uA. Ok, so here's what I figured out... I took my measurements with and without the Base, subtracted the 2 and divided that by .00247. So,  .33v-.06v=.27v/.002472=109uA... Yes?

Looks OK to me.  Also the values are reasonable.

Off hand,  RG's site (geofex) spells it out.

[mac]

I use a 1k resistor from V to Collector and a 1meg resistor from V to Base to measure Ge or Si.
Twice base current, easy to read leakage.
OTOH RG tester makes gain readings easy.
Easy leakage reading, pay with gain. Easy gain reading, pay with leakage... sounds familiar? 

The model:

ic = hfe*ib + iL

V: battery
iL: leakage

Rc: collector resistor
Rb: base resistor

(V - Vc*) : voltage across Rc with Rb not connected.
(V - Vc)  : voltage across Rc with Rb connected.

iL = (V - Vc*)/Rc
ic = (V - Vc)/Rc
ib = (V - Vbe)/Rb

Vbe: Base - Emitter voltage drop

hfe = (ic - iL)/ib

or
hfe = [(V - Vc)/Rc - (V - Vc*)/Rc]/(V - Vbe)*Rb

hfe = Rb/Rc*(Vc* - Vc)/(V -Vbe)

Vc*, Vc: collector voltages respect to GND

RG makes Rb/Rc/(V - Vbe) close to 100.
V - Vbe = 9v - 0.1v (typical) = 8.9v
2.2meg/2472/8.9v = 100/volt
so

hfe = 100*(Vc* - Vc)       easy
iL = (V - Vc*)/2472ohm

What I do is to set Rc = 1k and Rb = 1meg so

hfe = 112*(Vc* - Vc)
iL = (V - Vc)/1000ohm      easy

The eqs can be used to measure Silicons by making iL = 0.
And to know the gain under different working conditions.
For example, let's say you have a nice AD139 and want to build a Class A amp @ 100ma idle collector current.  If a priori you suppose a gain of 100 then base current is about a milliamp, way higher than 4 microamps used in fuzzes.
Since hfe is a function of ic you can expect gain is going be different at 1ma than at 4ua.
You can choose Rc and Rb to test the transistor under a base current of 1ma, for example, Rb = 9kohm, Rc=10-100ohm

mac