scott wurcer said:
PMA said:
I understand this audio rectification but this is always a dubbel process: first coupling of the interference then rectification on the first non linear element.
Your experiment with a motor is IMHO generating near field magnetic field and the low input impedance of the BJT will couple better than a high input. Then in this case of course rectification at a point where gain is highest.
If you would make an experiment with a plane wave which is mainly electrical field, would you not have stronger coupling to the JFET and then rectification on the first non linear element. Of course the gain there will then be lower.
I understand and it is an interesting point the audio rectification thread of BJT at the input, but from a susceptibility point of view are they more susceptible? I believe it depends on the type of interference.
In high sensitivity inputs, one should always design for the best input immunity whatever the input.
JPV
scott wurcer said:
Which isn't audio as far as I can tell. And who listens to their hifi gear unshielded whilst operating power tools? When it comes to discrete amplifier circuits using bipolar input stages with emitter degeneration and moderate Ic, RFI susceptibility is a non-issue, IMHO.
Which isn't audio as far as I can tell. And who listens to their hifi gear unshielded whilst operating power tools? When it comes to discrete amplifier circuits using bipolar input stages with emitter degeneration and moderate Ic, RFI susceptibility is a non-issue, IMHO.
Well said.
If th e front is designed correctly, it should not overload (i.e. start to operate in its non-linear region) before about 1.5V. As Glen says, this is acheived through degeneration. Anyone not degenerating the bipolar input on a modern design is likely to pick up RFI and distortion problems, to say nothing of the fact that the circuit will probably end up with excessive loop gain
None of the above detracts of course from the JFETS which are excellent . . . . but let us not lumber bipolars as in any way inferior by way of an argument which is not rooted in practical, good design methodology circa 2008.
Well said.
If th e front is designed correctly, it should not overload (i.e. start to operate in its non-linear region) before about 1.5V. As Glen says, this is acheived through degeneration. Anyone not degenerating the bipolar input on a modern design is likely to pick up RFI and distortion problems, to say nothing of the fact that the circuit will probably end up with excessive loop gain
None of the above detracts of course from the JFETS which are excellent . . . . but let us not lumber bipolars as in any way inferior by way of an argument which is not rooted in practical, good design methodology circa 2008.
Bonsai said:
It is not well said at all, and deep ignorance can be seen. Who of you guys, who state that RFI or HF EMI is not an issue for audio, has already measured it? In the world of SMPS, digital signal processing, and industrially polluted mains network, it is one of the main issues! Try your measurements on a whole audio chain, repeat it with different signal cables and grounding schemes, make housing design modifications, and then return and tell me, then I would probably take your opinion into any account.
john curl said:
scott wurcer said:
I am not opening a debate because I have no opinion. I am only trying to understand.
Your input about the handbook is what I need. I will read it, thank you.
The coupling mecanisms is always application specific in EMC and you should know them when you design.
It is obvious to me that replacing a bjt front end component by a fet front end component if you have RFI problems is limited. You should know more and try to fight the ingress at the same time.
If for good reasons I need to use the AD620, I better know what are the critical coupling mechanisms and why.
I still wonder if you pass a drill over your circuit and after that a cell phone with an essential electrical field, is the susceptibility the same? This is a question not an opinion.
Thanks
JPV
It is not well said at all, and deep ignorance can be seen. Who of you guys, who state that RFI or HF EMI is not an issue for audio, has already measured it? In the world of SMPS, digital signal processing, and industrially polluted mains network, it is one of the main issues! Try your measurements on a whole audio chain, repeat it with different signal cables and grounding schemes, make housing design modifications, and then return and tell me, then I would probably take your opinion into any account.
I little disingenuos Pavel.
. . . . most of us take care with shielding, gounding and input filtering and becuase of this very reason, it is controllable and avoidable in most cases.
What do you do?
I little disingenuos Pavel.
. . . . most of us take care with shielding, gounding and input filtering and becuase of this very reason, it is controllable and avoidable in most cases.
What do you do?
PMA said:
I did and it is a non issue for any competent design (bipolar or FET) under normal environment circumstances (including SMPSs but excluding a 2KW AM transmitter in the neighborhood and unshielded loops at the input).
I have found things like ground loops, magnetic effects, etc... much more significant than overloading the input stage with RF.
http://www.amazon.com/Noise-Reduction-Techniques-Electronic-Systems/dp/0471657263
This is a great book - turns the theory you learn at college into really useful working knowledge.
This is a great book - turns the theory you learn at college into really useful working knowledge.
From my experience, not only bipolar/jfet input type that important in EMI susceptibility. Input differential's standing current is another important factor. Jfet input opamps still have different EMI susceptibility. OPA604, which is running high standing current in input differential (due to folded cascode structure) is the most "bullet proof". OPA627, OPA134 and bipolar opamps are not as "bullet proof" as OPA604.
PMA said:
LOL! Strawman alert!
Well I'm definitely not one of those "guys, who state that RFI or HF EMI is not an issue for audio". In fact my opinion is the opposite, but thanks for your comments anyway.
I was specifically referring to the “RFI susceptibility” of discrete “bipolar input stages with emitter degeneration and moderate Ic”.
And yes, I have measured, extensively, and of my bipolar-input designs have any issues with rectifying/detecting RF (even from the 50kW AM broadcast transmitter less than 10km away).
A HiFi system so crappy that enough RF interference has the opportunity to enter and be detected by the input stage (JFET or bipolar) doesn’t rate a mention anyway.
JPV said:
There are some bipolar-input opamps out there of sub-optimal design that are well known for RF detection issues (local AM radio station) on input signals inadequately shielded for RFI pickup.
It is therefore erroneously inferred ad nauseam around here that a using BJT in supposedly any amplifier input stage configuration is necessarily inferior (for all practical purposes) to selecting JFETs – mostly by those of a religious belief that JFETs “sound better”.
Cheers,
Glen
scott wurcer said:
Very good chapter in the handbooK.
Interesting to understand that it is the quadratic non linearity of Fet that is responsible for the less efficient rectification compared to the highly non linear exponential of the bjt.
Reference is made to a chapter on "systems" to have the full derivation. I cannot find this reference.
If you have it on hand could you pass me the information.
Thank you in advance
JPV
By the way, rectification happens on b-e junctions of output devices as well, when emitter followers are used on inputs. Cellphone near speaker cables causes rectified signal to enter input of the amp through a GNFB loop. Also, the signal itself, especially higher obertones, modulate bias of output transistors because of capacitive load (capacitance of cables and reactance of speakers).
all the more reason to implement a Thiele network on the output rather than a simple Zobel.Wavebourn said:
Maybe we should all be thinking about a Pi filter on the output now that many of us have mobile, wireless networks, wireless security & alarms, etc in our homes.
Pi filter = Zobel (R+C) at the output stage, then R//L followed by further Zobel at the speaker terminals.
- Status
- Not open for further replies.