Bob's post is misleading. The bootstrap capacitance must be properly computed, before it makes any sense.
john curl said:
Bob's post is essentially correct. Defining the BJT junction capacitances as the sum of a space charge capacitance and a diffusion capacitance is the canonic approach in the Gummel Poon model.
Computing the two capacitances is almost impossible, due to the lack of essential values like the junction exponential factor, the transit times, transport saturation currents, etc... However, if one would measure C(V) of the B-E and B-C junctions, all these parameters could be extracted and allow a fair comparison between various BJT devices. See the excellent HP paper on this topic
http://eesof.tm.agilent.com/docs/ic...MODELING/3TRANSISTORS/1GummelPoon/GP_DOCU.pdf
What I would disagree on is making this a basis to compare BJT to MOSFETs. There is no common physical model that would allow us to compare their intrinsic parameters, simply because the physics beyond the two devices is different. Perhaps a generic Hij or Yij quadripole model would allow a basic comparation. But then the quadripole models are essentially linear, a very rough approximation at the high injection levels like in an OPS. And Hij does not reflect explicitely any bias (and hence gm) dependency.
syn08
lumanauw said:
So I've been told, and it's plausible. As a matter of actual fact,
when I developed circuits which did not use bipolar inputs, the
RF pickup complaints went to zero.
😎
syn08 said:
I've also got some info on this subject here, with examples for fitting the data of the MJL3281A and MJL1302A. I use the Excel solver to get CJE, MJE and VJE from curves of Cbe vs Vbe at multiple data points. There's a similar approach for the collector-base capacitance parameters.
You folks can also argue about' the number of Angels on the head of a pin' as well. What is being said here is incomplete and basically a 'half truth'. It might be interesting, but it is not very useful. Someday, when you build amplifiers for real, this will become apparent. For the record, use 9 pairs of these same devices in parallel and get virtually any rate of change that I might wish for. If I can EASILY get 100V/uS, then what does this input capacitance mean? Little or nothing.
Re: MOSFET and BJT Inter-electrode Capacitances
Hi John,
I think we kind of ended up talking at cross-purposes here. When you mentioned this:
I think you may not have noticed that Bob mentioned this:
Then when syn08 began talking about the junction capacitances again, I think he may have missed the point you were trying to make, namely that it's the bootstrapped capacitance that matters. I understand and agree with your point that since the gm of a BJT is higher than that of a MOSFET, the voltage gain of an emitter follower is closer to one than that of a FET source follower. So the "bootstrapping factor" (1 - Av) is closer to zero for a BJT than a MOSFET, thus partly negating the higher BJT Cbe as compared to Cgs of a FET.
So I think it's more of a communication thing, rather than any fundamental difference in point of view.
Hi John,
I think we kind of ended up talking at cross-purposes here. When you mentioned this:
john curl said:
I think you may not have noticed that Bob mentioned this:
Bob Cordell said:
Then when syn08 began talking about the junction capacitances again, I think he may have missed the point you were trying to make, namely that it's the bootstrapped capacitance that matters. I understand and agree with your point that since the gm of a BJT is higher than that of a MOSFET, the voltage gain of an emitter follower is closer to one than that of a FET source follower. So the "bootstrapping factor" (1 - Av) is closer to zero for a BJT than a MOSFET, thus partly negating the higher BJT Cbe as compared to Cgs of a FET.
So I think it's more of a communication thing, rather than any fundamental difference in point of view.
The capacitance is the capacitance. The transistor (or tube) does
not know which of the three possible amplifying connections
it is in. The amount of current traveling through the control
pin is the same for all.
😎
not know which of the three possible amplifying connections
it is in. The amount of current traveling through the control
pin is the same for all.
😎
Nelson Pass said:
There was a distinction being made by John and others between junction capacitance (Cbe, Ccb, Cgs, Cgd) and equivalent input capacitance of a source follower as compared to an emitter follower. That was one source of confusion that I was trying to clear up.
andy_c said:
I appreciate that, and it's not that I think the participants in the
discussion are confused, but the terminology tends to lead
neophytes astray.
😎
I think the terminology used ends up being appropriate to the posters of a thread, and not necessarily the lurkers. So there may be lurkers reading the thread, thinking "WTF?". But until they post, nobody will know that they may be confused about something. If a bunch of engineers are discussing something, it doesn't make sense to restrict terminology to that which non-engineers would immediately feel comfortable with. There might not be any non-engineers interested in or reading that subject.
If there are lurkers that are confused, the solution to that is making the transition from lurker to poster and asking questions. I hope the tone of the thread does not discourage that.
If there are lurkers that are confused, the solution to that is making the transition from lurker to poster and asking questions. I hope the tone of the thread does not discourage that.
lumanauw said:
Nelson Pass said:
Not because of bipolar forward gm characteristic ? I thought the FET is more immune because of smoother gm region than Bipolar
Hartono
I read it somewhere here:
http://www.ce-mag.com/archive/2000/novdec/fiori.html
I still don't understand many parts, so I might have misinterpret it.
http://www.ce-mag.com/archive/2000/novdec/fiori.html
I still don't understand many parts, so I might have misinterpret it.
syn08 said:
Hi Syn08,
First, I'm glad to see that you have taken the effort to build my design.
Second, I don't have any clou (at the moment) why your results are so poor, needless to say that I'm also rather disappointed. Maybe it's just a wiring error as a result of the low resolution of the scaled down schematic. However, this contradicts with the good results of your simulation. Anyhow, if you give me your email address, I'll send you the high res. version along with some notes on typical currents and voltages, so we are sure we are talking about the same thing.
BTW1, I haven't built this design. It is only a spin-off of a joint project of Glen and me, and meant as an example of NDFL and TMC. Maybe Glen is inclined to implement it.
BTW2, two pole compensation?, I thought that Glen has implemented the "transitional miller compensation" (TMC), but I can't check this right now, as his website has been blocked.
BTW3, thank you very much for explaining the peculiarities of the IRF9240. That was a really nice contribution.
Cheers, Edmond.
john curl said:
John:
I thought we are discussing here about comparing the intrinsic devices (BJT vs. MOSFET) rather than comparing two particular circuit instances (emitter follower vs. source followers). I apologise if I misunderstood the topic.
To me, the reason why emitter/source followers are so close in performance is not because BJTs and MOSFETs are comparable in any way, shape or form. I prefer to look at the emitter/source follower as a circuit a circuit with deep negative feedback, i.e., all of its output is fed back to become part of its input, see e.g. http://fourier.eng.hmc.edu/e84/lectures/ch4/node9.html As in any FB circuit, if the base amp is "good enough" then the final result is rather independent on it's intrinsic parameters. It's only the feedback network that defines the circuit properties and perfomances.
syn08
estuart said:
Actually I've tried both the TMC and a dual pole version, with essentially the same (good) results. Perhaps the differences are below the resolution of my equipment (Amber 5500)?
On the flip side, I've noticed that Glen's design is occasionally affected by some strange latch-up effect. When the output signal exceeds about half the rail, any slight change in the input triggers some sort of low frequency oscillation (a few KHz) with the output bouncing between rails.
This was for a high voltage version (+/-50V) and I was able to control this behaviour by adding a few clamp diodes, which also helped to better control the clipping.
Email sent.
Thanks,
syn08
Hartono said:
You can't approach any AM demodulation process with a device small signal linear model. You need to look at the large signal model of the device! In this respect, BJTs have an exponential dependency of the collector current on Vbe, while MOSFETs have a quadratic dependency of drain current on Vgs. The exponential characteristic makes a BJT a much efficient demodulator, even it's relatively low input impedance damps the "antenna" (e.g. a long input wire).
If you feel like dipping in some math, see the exact equations for a FET based demodulator at http://www.ece.ndsu.nodak.edu/~yuvaraja/EE421/E421-N-Amplitude Demodulator1.DOC You could substitute the bipolar exponential characteristic and find out how large the demodulated component is.
syn08
"The exponential characteristic makes a BJT a much efficient demodulator."
that's what I'm trying to say
that's what I'm trying to say
syn08 said:
Hi syn08,
Sure, THD figures should be essentially the same. It's the phase response of the global feedback loop that is totally different, see:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=94676&perpage=10&pagenumber=47 post#466 and 467
syn08 said:
Hmm.... very strange. Can you reproduce it by means of simulation, so I can also have a look at it?
Regards, Edmond.