hfe doubling for low hfe germanium transistors

[Mark Hammer]

I can't vouch for it personally, but Osamu Hoshuyama knows his stuff, and if he says it works, it works.

http://userdisk.webry.biglobe.ne.jp/000/024/65/1/HfeDbl0506.GIF

[The Tone God]

I'm getting a Forbidden.

Can you rehost ?

Andrew

[albatross]

rightclick and save target as, works for me.

and a very good idea for low gain devices!

[Mark Hammer]

Yeah, link works for me in Firefox as right OR left click.  Failing that, you can go to the webpage and work from there: http://houshu.at.webry.info/  Scroll down about halfway.  You'll find lotsa good stuff on the way.

[The Tone God]

The link from the website worked especially when I translated on the fly using FoxLingo.

Andrew

[joelap]

"Influence of diode is negliible if Q3 is saturated" <--- does he mean "linear active" here?  I learned you want FETs saturated, BJTs linear active.

[R.G.]

It's worth noting that using this method, you can get hfe tripling, quadrupling, etc., depending on how desperate... er... dedicated you are. Just expand the current mirror with multiple secondary transistors. You can get better results in those cases with a Wilson current mirror as well, although it may not matter much to the sound.

[Paul Perry (Frostwave)]

I suppose there must be some subtle way to 'distort' the mirror, giving a ratio control? One problem is that people who do this stuff for a living, are able to assume perfectly matched transistors (because they are being made on the same chunk of silicon a few nanometers apart) and instead of two transistors in parallel, it' just one transistor made wider or something.
On the other hand, us discrete component guys have the best resistors

[caress]

i am using the ge/si darlington method to boost the gain on my low hfe ge's...can anyone explain to me the differences between these two methods?  also, the Q3/D1 - is that for another method, or is that integrated between Q1/Q2? -- i'm a little confused about the schematic...

[caress]

badump

[mac]

One question. Does it amplifies the leakage?
I recently posted about a Bazz Fuss with a Ge darlington pair and lakage was a problem because it was amplified a lot.
Errrr, how does it works?

mac

[Steben]

i am using the ge/si darlington method to boost the gain on my low hfe ge's...can anyone explain to me the differences between these two methods?  also, the Q3/D1 - is that for another method, or is that integrated between Q1/Q2? -- i'm a little confused about the schematic...

I agree. I don't understand it either. If the diode is another method, why bother for both the trannies, not?

And do these mirrors affect the alinearity of the germanium device?

[Iojan]

My English is poor, so I'll try to explain it 



if we assume:
-linear active
-hfe1=hfe2, Vbe1=Vbe2

we have:

Ic3 = hfe1 * Ib

Ic3 = Ic2 (Current mirror Q1-Q2)

Ic = Ic3 + Ic2 = 2*Ic3

Ic = 2*hfe3*Ic3 <- Doubled hfe

As RG said, if you connect more transistors pin to pin to Q2 you can get hfe tripling, quadrupling, etc.

Saludos desde Argentina (I need to use my mother tongue  )

[R.G.]

Y saludos desde Austin Tejas, Iojan.

Your English is much better than my Spanish - and entirely sufficient to the task.

And your electronics understanding is absolutely correct. Q1 senses the current through the emitter of Q3, and forces the base-emitter voltage of Q2 to be the same as its own voltage. This forces the collector current of Q2 to be equal to the total current of Q1, within the accuracy of the transconductance and Vbe matches of  Q1 and Q2. Also using additional paralleled "Q2s" multiplies the amount of current pulled.

The leakage of Q3 is multiplied like the desired current.

This trick can be done on either the collector side or emitter side of Q3, whichever is less critical for bias; usually the collector side is best. There is always a diode-drop lost across Q1.

Matching Q1 and Q2 is important if you want exactly 2.0 times the Q3 current, but that's usually not a big consideration. Probably any two devices of the same type number will give you some result.

I mentioned the Wilson current mirror. This uses a third mirror transistor to provide the base current for Q1 and Q2, and is simpler to use if you want multiple Q2s. It does introduce another diode drop in the sensing side, though.

[Iojan]

Thanks RG! I am student of electronic engineering but you are my reference on the technology of the effects. 
Nothing else that to say, I hope this helps to understand better the trick.

Greetings

[oldschoolanalog]

Note: It says untested on the diagram .

[Sir H C]

YOu could use transistor arrays to get good matching and be able to gain up pretty easily.  I guess this would be really only necessary on the first transistor in the FF circuit, the one that shouldn't be clipping.  problem with that implimentation is that you would want to put the second transistor on the bottom of the mirror, but if you do you now have to rework the biasing or you will always be close squashing out the first transistor (.7v drop for the current mirror device is right there near 2 vbe's for Ge devices).