The reason that I'm asking is because an experienced designer will know that although all of these circuits can be properly called "Darlington", they will have different performance characteristics. Furthermore these characteristics will have an impact on the sound quality.
So it might be useful to define our terms here so that when we use a specific word, we can agree what kind of "Darlington" we are referring to.
Nah...
That would make too much sense. Besides, then we couldn't play our game anymore (... the mulberry bush, the mulberry bush....)
Cheers,
Charles Hansen
So it might be useful to define our terms here so that when we use a specific word, we can agree what kind of "Darlington" we are referring to.
Nah...
That would make too much sense. Besides, then we couldn't play our game anymore (... the mulberry bush, the mulberry bush....)
Cheers,
Charles Hansen
Here's the Borbely quote, from JFETS: The New Frontiers, Part 2:
"JFETs as Followers
Just like tubes and bipolar transistors, JFETs can also be operated as followers, more specifically source followers. The basic circuit is shown in Fig. 15a. The drain is AC-grounded, and the output signal is taken out across the source resistor, which means it operates with 100% local feedback. The gain of the source follower is:
AV = gm x RS/(1 + gm x RS)."
se
"JFETs as Followers
Just like tubes and bipolar transistors, JFETs can also be operated as followers, more specifically source followers. The basic circuit is shown in Fig. 15a. The drain is AC-grounded, and the output signal is taken out across the source resistor, which means it operates with 100% local feedback. The gain of the source follower is:
AV = gm x RS/(1 + gm x RS)."
se
andy_c said:
YES!
I think it is already very clear ( and was clear for many years before this thread started) that any emitter/source/cathode follower uses negative feedback and only people who market "zero feedback" amplifiers using followers are keep objecting
. I will quote on this matter from Ben Duncan's excellent book "High Performance Audio Power Amplifers" :
" 'Zero' really means no overall feedback. It is usually practically nesessary in production, and is also no less advantageous to measured linearity and sonics, to add local feedback where it isn't already used, and to increase the amounts." (page 158)
x-pro
Re: insert suitable quote here!
Charles has already discussed the technical aspect of various
Darlington topologies. One thing I think he missed to include (or
I just missed it in his postings) is the Darlington devices, that is
those Darlintons on a chip sold as Darlinton transistors. One
amplifier using such devices is the now discontinued swedish
power amp Sentec PA-9 which is still one of the recommended
amplifiers by the Swedish audiotechnical society (LTS). Their
comment is that apart from the somewhat restricted power they
have found no shortcomings in it. A friend of mins has those
monoblocks since many years and they are not bad at all. I am
not saying they are the best things ever made or even that they
would qualify has high-end, but they are quite good
amps. That said, my friend has reported improvements from
upgrading the transformers and lytics in his amps, but that might
be because his new speakers really would need more powerful
beasts.
PrimeCase said:
Charles has already discussed the technical aspect of various
Darlington topologies. One thing I think he missed to include (or
I just missed it in his postings) is the Darlington devices, that is
those Darlintons on a chip sold as Darlinton transistors. One
amplifier using such devices is the now discontinued swedish
power amp Sentec PA-9 which is still one of the recommended
amplifiers by the Swedish audiotechnical society (LTS). Their
comment is that apart from the somewhat restricted power they
have found no shortcomings in it. A friend of mins has those
monoblocks since many years and they are not bad at all. I am
not saying they are the best things ever made or even that they
would qualify has high-end, but they are quite good
amps. That said, my friend has reported improvements from
upgrading the transformers and lytics in his amps, but that might
be because his new speakers really would need more powerful
beasts.
Really? How can that be? Isn't there a causality issue that you're missing? The change a of voltage at the emitter is caused by the change in input voltage. There may not be much delay, but surely there's delay.
The whole delay issue is a red herring, else you couldn't drive a car.
Yes, every active circuit in the real world has some feedback, much as those of us who sell (or in my case, used to sell) like to wave our hands around the issue. Even an undegenerated common emitter stage has some feedback due to re.
Power bipolar transistors show a huge B-E capacitance when conducting and a C-B capacitance an order of magnitude lower
When using bipolars as emitter followers, B-E capacitance guarantees a gain of 1 or more and little phase shift at high frequencies. A a capacitive current divider is formed with C-B and B-E capacitances and most of the B-E 'charging' current is delivered to the load
When using bipolars in common emitter configuration, gain may be much less than unity and phase shift is huge at high frequencies. Most of the drive current is used to 'charge' the B-E junction and therefore is never delivered to the load
So emitter follower configuration is allways 'faster' than common emitter since it has higher current gain and lower phase shift specially at high frequencies
In CFPs, the small and fast emitter follower tries to compensate for the slowness of the big and slow common emitter one, but this means the small device must handle huge currents at high frequencies
Also, the small device has nothing to do to speed up turn-off of the power device and this causes heavy cross-conduction spikes at high frequencies and problems of HF oscillations due to very high current rise slopes and very low current fall slopes [and lots of lag before effectively current falls on the power device]
Driving bipolar devices is not so dfferent from driving MOS devices, B-E junction needs high positive currents to be 'charged' and high negative currents to be 'discharged' [of a magnitude of half IC]. Lower currents mean slower charge displacement and slower IC changes. The only difference is that bipolars require a steady-state current while MOS devices doesn't require it
I recomend experimenting with the switching behavior of bipolar devices to understand its HF characteristics. Currently I'm using bipolars on a full bridge to switch 6A@300V DC and trying to optimice switching losses
When using bipolars as emitter followers, B-E capacitance guarantees a gain of 1 or more and little phase shift at high frequencies. A a capacitive current divider is formed with C-B and B-E capacitances and most of the B-E 'charging' current is delivered to the load
When using bipolars in common emitter configuration, gain may be much less than unity and phase shift is huge at high frequencies. Most of the drive current is used to 'charge' the B-E junction and therefore is never delivered to the load
So emitter follower configuration is allways 'faster' than common emitter since it has higher current gain and lower phase shift specially at high frequencies
In CFPs, the small and fast emitter follower tries to compensate for the slowness of the big and slow common emitter one, but this means the small device must handle huge currents at high frequencies
Also, the small device has nothing to do to speed up turn-off of the power device and this causes heavy cross-conduction spikes at high frequencies and problems of HF oscillations due to very high current rise slopes and very low current fall slopes [and lots of lag before effectively current falls on the power device]
Driving bipolar devices is not so dfferent from driving MOS devices, B-E junction needs high positive currents to be 'charged' and high negative currents to be 'discharged' [of a magnitude of half IC]. Lower currents mean slower charge displacement and slower IC changes. The only difference is that bipolars require a steady-state current while MOS devices doesn't require it
I recomend experimenting with the switching behavior of bipolar devices to understand its HF characteristics. Currently I'm using bipolars on a full bridge to switch 6A@300V DC and trying to optimice switching losses
"JFETs as Followers
Just like tubes and bipolar transistors, JFETs can also be operated as followers, more specifically source followers. The basic circuit is shown in Fig. 15a. The drain is AC-grounded, and the output signal is taken out across the source resistor, which means it operates with 100% local feedback.
The gain of the source follower is: AV = gm × RS/(1 + gm × RS).
Two things become obvious from the formula: first, the source follower does not reverse the phase of the signal, and second, if gm x RS >> 1, then the gain becomes approximately unity. In order to make RS large, you can use a constant-current source with high output impedance (Fig. 15b). The linearity is also dependent on RS..."
from "JFETS: THE NEW FRONTIERS, PART 2" by Erno Borbely
JF
Just like tubes and bipolar transistors, JFETs can also be operated as followers, more specifically source followers. The basic circuit is shown in Fig. 15a. The drain is AC-grounded, and the output signal is taken out across the source resistor, which means it operates with 100% local feedback.
The gain of the source follower is: AV = gm × RS/(1 + gm × RS).
Two things become obvious from the formula: first, the source follower does not reverse the phase of the signal, and second, if gm x RS >> 1, then the gain becomes approximately unity. In order to make RS large, you can use a constant-current source with high output impedance (Fig. 15b). The linearity is also dependent on RS..."
from "JFETS: THE NEW FRONTIERS, PART 2" by Erno Borbely
JF
Attachments
I think that what traderbam was saying was that the emitter follower was drawn as a two-terminal device -- base and emitter. So if we insert a different two terminal device (e.g., a resistor), then we can see that one lead corresponds to the base and the other to the emitter. In that case, the resistor value becomes Rl and gm = 1. But maybe I'm just misunderstanding what he was saying.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.