We used to design circuits like that back in the Germanium days. Was they long tail pairs or was they short tail pairs?
It is just that we used one extra resistor that went from the base to the ground, so that the circuit could do something useful.
It is just that we used one extra resistor that went from the base to the ground, so that the circuit could do something useful.
Wavebourn said:
Do you see what it means?
In order to work effectively Scott's amp as drawn have to be powered by voltage twice of Beta. If Beta is 50, it has to be 100V. If Beta is 400, it has to be 800V. Also, in order to have an amplification factor marked as a "right answer" the source has to have zero output resistance, the load has to have infinite input resistance.
My father in law is a scientist, he worked in Academy of Sciences in USSR. Seeing our daughter learning the math he decided to take over while it is not too late. He teaches her to think instead of to guess answers according to rules given by teachers that they don't understand.
john curl said:Where are the fets?
FETs were invented before BJTs. Actually, BJTs are "bad" FETs that appeared to be useful.
One more device that was created, was called 2-base diode, or a single-junction transistor. It was used in oscillators, but I don't know if they are still manufactured.
My colleague, just built Scott's circuit using 10K resistors and a 2n3904 transistor. He got a voltage gain of 160, with a 10V supply. How about that? Seems that predictions don't mean much in this case. Some engineering!
The 1960 'GE Transistor Manual' on p. 37, gives this equation for the SAME circuit:
Ic = Beta * Vs / (R1 + Beta * R2)
Voltage gain is completely ignored.
The 1960 'GE Transistor Manual' on p. 37, gives this equation for the SAME circuit:
Ic = Beta * Vs / (R1 + Beta * R2)
Voltage gain is completely ignored.
john curl said:
Because this equations is given to roughly calculate an idle current.
john curl said:
What is the collector voltage? I assumed Vo is near the half of Vs. Your collegue got a bit more than 1 Volt.
Unijunction transistors, we used to make oscillators with them back in the old days!
My business partner still uses them in a strobe light. Try 2N491.
My business partner still uses them in a strobe light. Try 2N491.
john curl said:Unijunction transistors, we used to make oscillators with them back in the old days!
My business partner still uses them in a strobe light. Try 2N491.
Yep!
I found lot of them! 🙂
http://search.datasheetcatalog.net/key/UNIJUNCTION
I used one in my first electronic musical instrument. Kind of Joystick controlled Theremin.
Back to measurement mike preamps.
I have about a dozen B&K's and several HP mike preamps all from the early 70's. The B&K 1" preamps have heaters in them (not helpful for low noise but keeps the humidity out). However the HP's, with a much simpler circuit, have as low noise and much lower distortion. They all have unity gain. I presume that distortion was not critical for a measurement mike since the application was looking at levels, RMS or several types of peak levels. The HP's all work fine 40 years later. Many of the B&K's have serious problems. Sometimes additional complications don't bring better performance.
I have about a dozen B&K's and several HP mike preamps all from the early 70's. The B&K 1" preamps have heaters in them (not helpful for low noise but keeps the humidity out). However the HP's, with a much simpler circuit, have as low noise and much lower distortion. They all have unity gain. I presume that distortion was not critical for a measurement mike since the application was looking at levels, RMS or several types of peak levels. The HP's all work fine 40 years later. Many of the B&K's have serious problems. Sometimes additional complications don't bring better performance.
john curl said:
So is Vbe does the GE manual really set Vbe = 0? I guess they did that 40 years ago. Gain is ignored too, why build it? In the above equation for reasonable beta you just get Vs/R2 So what? In your circuit for about a beta of ~200 that's about the right answer, sort of 4 volts across the load the only thing first order that determines the gain (the useful part).
john curl said:We used to design circuits like that back in the Germanium days. Was they long tail pairs or was they short tail pairs?
It is just that we used one extra resistor that went from the base to the ground, so that the circuit could do something useful.
That makes sure it doesn't work. Yes you would put a resistor in the emitter so that no parameters of the transistor matter. WOW G =~ Ra/Rb now we're designing.
Again you aren't even looking at or thinking about any of this.
Wavebourn, don't be silly it's a thought experiment. Also your equation is wrong. This circuit works fine even at 1v with the right values.
scott wurcer said:
Like "What would you perform if Her Majesty The Queen falls down from your plane?"
scott wurcer said:
No, the equation is not wrong. Our assumptions of what is "fine" are different.
scott wurcer said:
No, he would put a voltage divider instead of one resistor from collector to base. Such a way he can adjust an idle voltage on collector, and it would less depend on temperature and beta variations than your amp. A resistor in emitter would not help to stabilize working point at all.
Wavebourn said:
... but that would only work with a sizeable resistor from the divider to base, or with an emitter resistor.
Jan Didden
I think saying that beta is not important in circuit design is silly. In most practical circuits there is some source impedance. You have have to take everything into account IMO.
janneman said:
Yes.
Say, we power the amp from 24V.
A first, select a collector load resistor according to output resistance you want.
Now, assume a voltage drop on it will be 1/2 of Vs. Say, it will be 10K. That means about 1 mA.
Now, calculate an emitter resistor to get needed voltage gain. Say, for gain of 10 it will be 1K, 1V drop.
Now, calculate an upper resistor in the divider, according to minimal Beta (that is say 100), with lower resistor omitted: 1 MOhm roughly.
Now, make it 10 times lower for desired thermal stability and to make the stage less dependable on Beta: 100K.
Now, calculate the lower resistor for 1.5V voltage drop: 15K
What do we have?
We have a stage with voltage gain roughly -10.
10K in collector, 1K in emitter, 100K from collector to base, 15K from base to ground.
Very rough calculations, but now we can forget about Beta and it's variations from device to device, and with temperature. However, due to a parallel feedback by voltage it will show lower gain depending on the source output resistance.
Edit: obvious typos corrected.
Wavebourn said:
This IS wrong. Unless the base resistor is very large Vo=~Vbe. The Schoeps transformerless mic uses emitter followers like this self biased.
Again this is a thought experiment using minimal information, to see if someone even has a notion of gm and and gain without looking for Re/Rc = 10. I'm surprised more people have not seen the self biased one transistor amplifier. It does really work and yes the exact gain varies due to non-idealities, that's not the point.
Johnloudb said:
I don't think Scott said that. He said that in that particular circuit the gain did not depend (to a certain extend of course) on beta. That circuit (R from C to B) is pure negative feedback, so of course it is insensitive to beta magnitude. Makes sense to me.
Jan Didden
scott wurcer said:
No, this is right. For your drawing where both resistors are equal. That output stage in that mic made my slippers laughing.
janneman said:
No, he assumed an input source with zero output resistance, otherwise a voltage gain will tend to be -Ri/R where Ri is an internal resistance of the source. The higher is Beta, the closer to it.
Edit: Scott's "fine" conditions mean close to saturation, mine mean half of Vs.
I prefer mine, for less distortions and higher output voltage swing.
Wavebourn said:
Are we looking at the same picture???? For infinite beta it's a diode connected transistor pulled up to the rail with a resistor (0V across the base resistor). Vce = Vbe. We don't consider that close to saturation.
It's not supposed to be a useful circuit.
- Status
- Not open for further replies.