So I searched for germanium designs on diyaudio and lots of people do use in various applications.
They becomes addictive. I obsessively buy germanium transistors of all sorts.
So now using in a hybrid.
Why do they sound good? They are leaky, unstable, poor batch control, etc.
Yes I read someone said no two sound alike. ( I think this is a positive! )
I don't think anyone makes them anymore(i think)
The high power parts are usually very low voltage and bandwidth limited below 10khz.
Darn I can't get a handle on it.
-
They becomes addictive. I obsessively buy germanium transistors of all sorts.
So now using in a hybrid.
Why do they sound good? They are leaky, unstable, poor batch control, etc.
Yes I read someone said no two sound alike. ( I think this is a positive! )
I don't think anyone makes them anymore(i think)
The high power parts are usually very low voltage and bandwidth limited below 10khz.
Darn I can't get a handle on it.
-
Sounds like you are making EFX boxes, lad.
Well one thing is for certain: the 'germanium transistor' no matter what its power, was essentially a “made by hand” Object d'Art. In that you have the veritable steampunk version of a sand-state amplification device. They are intrinsically hugely variable because of how they are fabricated.
Imagine…
50 millimeter diameter wafers were made out of N-type germanium, made from bulk N-type semiconductor which was almost the only kind the industry could inexpensively produce.
The boules were purified by the float-zone process, where after being grossly fabricated, an inductively molten region was induced, then gradually “sent to the bottom” from a suspended crystal. And done again, and again and again, until purity exceeded 5 nines.
Then it was sawn into wafers. And they were grossly lapped, then fine lapped, then super-fine lapped in batches. Then chemically polished from a dull to a bright sheen. If all was going well on the transistor line, the thickness of the wafers was about 500 μm or so (½ millimeter).
Now the fun part… by hand using 'jigs', itty-bitty round bits of germanium bearing a P-type dopant were literally stuck to the wafer at transistor centers, in arrays. They were loaded into a low-heat kiln having nitrogen gas with a bit of hydrogen and heated to where the 'sticky stuff' bonded well to the bits and the wafer. Then they were moved in trays to the hi-heat oven, where they were ramped up to several hundred degrees centigrade; near but not too near germanium's melting point.
The itty bits then completely welded to the germanium. The heat was kept up until the dopant atoms diffused from the surface toward the centers of the wafer. With some real-time testing and enough time, the dopant front would diffuse from both sides toward the other sides. The gap, the "base", would get thinner and thinner.
Problem was (and remains) that this bulk-diffusion method doesn't really have a terribly flat diffusion front. The (smaller) emitter and (larger) collector bits could easily have dendrites (not literally, but equivalently in a diffusion front sense) that'd short out the base entirely. Thus not just time, but temperature ramping was critical to making devices with sufficiently high gain, sufficiently high VCEO breakdown and sufficient tolerance to transients of over-voltage, over-current and so on.
One of my buddies, no long deceased, working at Fairchild said that when done, about 25% of the wafers were all but lost to too-high-and-undetectable-before-firing diffusion dendrites. Of the survivors, about 50% of the transistors were not just out of spec, but virtually useless for being so far out of spec.
The LAST steps were testing, dotting (putting red dots on the losers), dicing and then mounting on the transistor 'header' that was made for it/them.
And that was entirely by hand in the early days, because the wafer-dicer had this terrible tendency to make chipped edges of the wafer 'base' material.
APPRECIATE the germanium transistor, and remember it is NOT A SIN to accept the moniker “maker of EFX boxes” when working with these antiques.
GoatGuy
Well one thing is for certain: the 'germanium transistor' no matter what its power, was essentially a “made by hand” Object d'Art. In that you have the veritable steampunk version of a sand-state amplification device. They are intrinsically hugely variable because of how they are fabricated.
Imagine…
50 millimeter diameter wafers were made out of N-type germanium, made from bulk N-type semiconductor which was almost the only kind the industry could inexpensively produce.
The boules were purified by the float-zone process, where after being grossly fabricated, an inductively molten region was induced, then gradually “sent to the bottom” from a suspended crystal. And done again, and again and again, until purity exceeded 5 nines.
Then it was sawn into wafers. And they were grossly lapped, then fine lapped, then super-fine lapped in batches. Then chemically polished from a dull to a bright sheen. If all was going well on the transistor line, the thickness of the wafers was about 500 μm or so (½ millimeter).
Now the fun part… by hand using 'jigs', itty-bitty round bits of germanium bearing a P-type dopant were literally stuck to the wafer at transistor centers, in arrays. They were loaded into a low-heat kiln having nitrogen gas with a bit of hydrogen and heated to where the 'sticky stuff' bonded well to the bits and the wafer. Then they were moved in trays to the hi-heat oven, where they were ramped up to several hundred degrees centigrade; near but not too near germanium's melting point.
The itty bits then completely welded to the germanium. The heat was kept up until the dopant atoms diffused from the surface toward the centers of the wafer. With some real-time testing and enough time, the dopant front would diffuse from both sides toward the other sides. The gap, the "base", would get thinner and thinner.
Problem was (and remains) that this bulk-diffusion method doesn't really have a terribly flat diffusion front. The (smaller) emitter and (larger) collector bits could easily have dendrites (not literally, but equivalently in a diffusion front sense) that'd short out the base entirely. Thus not just time, but temperature ramping was critical to making devices with sufficiently high gain, sufficiently high VCEO breakdown and sufficient tolerance to transients of over-voltage, over-current and so on.
One of my buddies, no long deceased, working at Fairchild said that when done, about 25% of the wafers were all but lost to too-high-and-undetectable-before-firing diffusion dendrites. Of the survivors, about 50% of the transistors were not just out of spec, but virtually useless for being so far out of spec.
The LAST steps were testing, dotting (putting red dots on the losers), dicing and then mounting on the transistor 'header' that was made for it/them.
And that was entirely by hand in the early days, because the wafer-dicer had this terrible tendency to make chipped edges of the wafer 'base' material.
APPRECIATE the germanium transistor, and remember it is NOT A SIN to accept the moniker “maker of EFX boxes” when working with these antiques.
GoatGuy
And I know because i worked in the (Plessey, Swindon) Semiconductor factory that sold them to him!
By the thousands in plastic bags, bent leads and all. Tested in a jig called the "3 light box" - green=OK, Red=effectively short or very low Vceo, Yellow=Are you sure it is plugged in?
I was a junior techie on the line, soon to be superseded by a Planar licence from Fairchild.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.