Not true.
All transistors are voltage controlled devices, including BJTs, even though their base current may be non-negligible.
This discussion reminds me of my early years at Bell Labs in the Linear Integrated Circuit Department. My roommate and I looked at transistor operation in two different ways. I thought of a transistor as a transconductance device where re was 1/gm and stage gain was approximately RL/(re + RE). I viewed base current as an imperfection that just needed to be taken into account.
He, on the other hand, viewed a transistor as a current amplification device whose gain was beta. Of course, both views yielded the right answer as long as all of the issues were properly taken into account.
BTW, in the early years we had to design linear ICs with lateral PNPs with Beta of less than 10 and ft often less than 1 MHz. When Bell Labs developed the first junction-isolated complementary PNP process in the late 70s we jumped for joy and never looked back. We were able to make great op amps for use in active filters - op amps that incorporated class AB diamond buffer output stages and T compensation (TPC).
Cheers,
Bob
No, not really. With a BJT, base current is a bug, not a feature. Were it possible, they would make a BJT that got all the carriers across that base without any recombination for zero base current. From the very first BJTs, that's what they've been trying for.
Never thought of that, but I guess Steampunk does describe it
. That's how I design audio amps, and get excellent sonic performance. Don't forget: you can do solid state designs that incorporate OPTs as well. (attached)These SiC JFETs need to operate at some serious voltages in order to keep the internal device capacitances small and linear, and according to the spec sheet, 300Vdc looks good.
I like the isolation that an OPT provides between me and 300Vdc. You can also design with more conventional Si MOSFETs that use an OPT (Susan Parker's Zeus design) to avoid the necessity of "complimentary" N-Channel/P-Channel pairs that are a good deal less complimentary than NPN/PNP pairs, and the two floating supplies that the Circlotron topology requires.
Attachments
Miles, where did you learn about semiconductor physics? You are incorrect. Base current is absolutely needed for bjt operation. The recombination of carriers in the base region is a small fraction of the total base current. There is the injection component which is holes from the p type base (npn bjt) moving into the emitter region & recombining. Thirdly, there is the charge/discharge component of base current.
As frequency increases, more base current is needed to change the charge concentration at a faster rate. Eventually the base current needed will equal the collector current at some high value of frequency. At this f value, beta equals one since ic = ib. This is the transition freq, "ft".
The speed of a bjt is defined as the frequency where beta has fallen to one. Base current is not an imperfection any more than any other parameter. An ideal b-e junction when forward biased would exhibit zero forward voltage drop. Vbe can also be viewed as an imperfection.
Ic is created by Ie, but Vbe is needed for forward biasing the b-e junction because p-n junctions are imperfect. An ideal b-e junction would carry forward current with zero Vbe value.
In the real world, Ib, Vbe, are realities I have to design with. Ideally we can argue that Vbe should be zero, but it just doesn't make it so. Likewise with Ib. Also, If Ib were zero, the device cannot be bipolar. A forward biased p-n junction cannot have Vbe w/o Ib. The p type base has free holes mobile & easily move due to an externally connected generator's E field.
The only way to have no base current is to use an insulator for the base material. Such a device is no longer bipolar. The word bipolar literally means "two polarities". A zero Ib device is not a bipolar device. How can an E field exist base to emitter w/o moving base carriers towards emitter resulting in Ib?
People who propagate this "Ib is an imperfection" heresy simply have no understanding of fields or semiconductor physics. Every bjt maker describes them as current controlled when viewing them as a black box not regarding internal physics. When viewing the internal physics, every bjt maker calls them "charge controlled", which is spot on right.
Unless you are making the bjt, you really don't know enough to say that those who do are wrong. Those of you who insist that the bjt makers are wrong, how many courses in semiconductor physics have you taken & passed? Just wondering.
This is pure crackpot heresy. Enough already. Peace & nothing personal.
Claude
Miles is correct.....and my semiconductor physics background is beyond reproach....
I believe everyone is usually mostly "right".
I generally think of a bjt as a voltage controlled current source for circuit analysis, but every once in a while I do find it necessary to remember the currents and there carrier concentrations.
Miles mentioned trying to eliiminate the recombination current, as that is a nuisance current, but your still left with the injection current as Claude added.
So for me the base injection current however small needs to exist in order for the collector to sweep in the corresonding emitter current.
Thats my simple EE view, I'm fully aware of others true understanding (the people in the bussiness).
Thanks
-Antonio
I generally think of a bjt as a voltage controlled current source for circuit analysis, but every once in a while I do find it necessary to remember the currents and there carrier concentrations.
Miles mentioned trying to eliiminate the recombination current, as that is a nuisance current, but your still left with the injection current as Claude added.
So for me the base injection current however small needs to exist in order for the collector to sweep in the corresonding emitter current.
Thats my simple EE view, I'm fully aware of others true understanding (the people in the bussiness).
Thanks
-Antonio
That seemed too obvious to require pointing out, but I guess it wasn't. You always, always, always have at least some "injection" current with every active device, even vacuum tubes. You see all the time the statement that tubes operated in Class *1 require voltage only, not current. It's not just an oversimplification, it's just plain wrong. The control grid won't see the true signal level until Cgk + Cmiller + Cstray charge to that value. Charging capacitance requires current, and the faster you need to charge (higher frequency) the more current required. Yet you see designs all the time made by those who forget this. "My 300B is Class A1 and doesn't need current, so I'll drive it with this 12AX7" -- and they wonder why it doesn't sound right. (A lot of folks say they are disappointed by the sonics of the 300B, and in every case, inadequate grid drive looks to be the cause of that.)
This is a capacitance-like effect.
Miles is correct.....and my semiconductor physics background is beyond reproach....
Care to elaborate re your semicon phy background? The semicon OEMs describe the bjt as current controlled (external) & charge controlled (internal). If you, Michael, are "beyond reproach", then what are the semicon OEMs status.
I've taken 1 semicon phy course from phy dept undergrad, & 4 courses from EE grad curriculum. Still, I have more semicon phy backgroud than most, but I do not regard myself as "beyond reproach". That is a strong claim to make. Do you work in semicon production? What is your academic background. Claiming to be "beyond reproach" is something that needs to be backed up.
The views presented here that contradict semicon OEMs is sophomoric claptrap. Nobody has given reason to doubt the OEMs. I will elaborate any point I made so far. For those who dosagree with me, please address one or more specific points that you want explained.
BR. Claude
Hi Antonio,
Well stated. In the limit, all of these devices are ultimately charge-controlled. While it is clear that BJTs do indeed fundamentally require base current to flow, that does not prevent us, as circuit designers, from viewing base current as a bit of a nuisance. The transconductance of a BJT is a much more reliably predictable parameter.
As circuit designers, we may also be tempted to regard some of the charge-controlled nature of BJTs as a nuisance as well. Certainly, it might be a happier world if the junction capacitances were not there. But, alas, these capacitances are also a necessary nuisance because of the physics of the device.
In different parts of a circuit, different aspects of a transistor, e.g., transconductance vs current gain, may be the more important - often dependent on the node impedance in the base circuit.
There is no simple black-and-white answer here for the circuit designer, and the good circuit designer is aware of both effects. Perhaps the circuit designer is allowed to be more pragmatic than the physicist.
I usually like to design circuits so that current gain is the less important parameter. If you design a circuit that will work poorly if current gain doubles from nominal, that is usually not a good thing.
Cheers,
Bob
I think this discussion got here because of the initial comments around 'the whole amplifer being current controlled' (or words to that effect) from ostripper.
On a conventional voltage amplifer, I tend to think of the LTP to VAS as a transimpedance amplifer - you deliver current from the LTP and the VAs delivers output voltage - so the VAS stage is in fact an integrator.
it=CV
Any base currents or leakage currents not flowing into or out of Cdom are error terms.
Again, these comments are for conventional miller comensation designs.
On a conventional voltage amplifer, I tend to think of the LTP to VAS as a transimpedance amplifer - you deliver current from the LTP and the VAs delivers output voltage - so the VAS stage is in fact an integrator.
it=CV
Any base currents or leakage currents not flowing into or out of Cdom are error terms.
Again, these comments are for conventional miller comensation designs.
Hi Bonsai,
I generally agree. I also like to view the input stage as a transconductance stage and the output stage as a near-unity voltage buffer in a conventional arrangement. However, these views can change a bit according to frequency and circuit topology. This way of looking at it is quite valid at frequencies well above the open-loop gain crossover frequency in a Miller compensated amp with an output triple, for example.
But for some amplifiers, as with output doubles with low current gain, the output stage requires significant current from the VAS and, especially at lower frequencies, the output stage may look like a current amplifier driving the output load, and substantially load the VAS collector.
Cheers,
Bob
RMAF and AES
I just wanted to mention that Cordell Audio and Linear Audio (Jan Didden) will be exhibiting this month at RMAF in Denver and at The AES convention at the Javits Center in NY (Booth 755).
We'll have copies of "Designing Audio Power Amplifiers" and Linear Audio for sale, and would really enjoy seeing you in person.
To get a free Cordell Audio AES exhibits guest pass, just go to:
http://www.aes.org/events/131/vip.cfm?798
Cheers,
Bob
I just wanted to mention that Cordell Audio and Linear Audio (Jan Didden) will be exhibiting this month at RMAF in Denver and at The AES convention at the Javits Center in NY (Booth 755).
We'll have copies of "Designing Audio Power Amplifiers" and Linear Audio for sale, and would really enjoy seeing you in person.
To get a free Cordell Audio AES exhibits guest pass, just go to:
http://www.aes.org/events/131/vip.cfm?798
Cheers,
Bob
That was last year
.Seriously, Jan and I love to meet and chat with the folks that can get to RMAF and AES. This is a wonderful fraternity of great people. RMAF is a wonderful, fun event.
Cheers,
Bob
That was last year
Seriously, Jan and I love to meet and chat with the folks that can get to RMAF and AES. This is a wonderful fraternity of great people. RMAF is a wonderful, fun event.
Cheers,
Bob
I second that. And of course Bob will sign his/your book for you!
jan
