Graham Maynard said:
G’day Graham.
I don’t think that is a fair generalisation.
Practically any solid-state amplifier with global negative feedback, regardless of topology, can be made to operate stably into highly capacitive loads. It all depends on how inelegantly the most difficult ones are strangled in the frequency compensation department.
An amplifier design with a gain crossover frequency in the vicinity of 2MHz, with heaps of high frequency negative feedback and ultra low distortion throughout the audio band may very well need a 0.5uH or higher output inductor in order to ensure stability when driving highly capacitive loads.
I’d say that such designs are far more technically competent than the majority of mediocre designs in a different performance league that don't require an output inductor, or that forgo or compromise many of the performance aspects related to maintaining high negative feedback in order to do away with it.
The Leech amp with it's feedback scheme that removes the output stage from the negative feedback loop at high frequencies is an excellent example.
There are much better ways of building an output-inductor-less amplifier than simply sticking with a generic topology and strangling it with heavy frequency compensation, or using the Leech feedback method for sure, but they still don’t come without compromises that do not deserve to be weighed up in comparison to the cons of using a load-isolating output inductor.
Cheers,
Glen
Hi Bob,
The thing that started this thread was a discussion of the role the output inductor plays in isolating the output stage from high speed external events - such as shorts to ground. And the possibility that such protection was needed when using MOSFETs at high supply voltages.
Could you comment on this please?
Thanks,
Tony
The thing that started this thread was a discussion of the role the output inductor plays in isolating the output stage from high speed external events - such as shorts to ground. And the possibility that such protection was needed when using MOSFETs at high supply voltages.
Could you comment on this please?
Thanks,
Tony
Interesting disussion - I include an output inductor as a knee jerk sort of thing, usually a 10ohm, 2W resistor, close wound with 22AWG magnet wire. This goes back in time to the old Motorola app notes that were my bible when I was first building amps. I haven't tried swapping them in and out, as I usually leave something in place if it doesn't cause any obvious grief. I've heard about audibility caveats over the years, but I'm usually impatient enough by the time I finally get an amp loaded into a case that I avoid doing a lot of messing around if it plays and is stable.
I have a pair of class A tweeter amps w/electronic crossover that are ready for mounting on the heatsink. They currently have the aforementioned inductor. If I remember tomorrow at work, I'll pop one off and measure it, as well as run the impedance plot in PSpice. Maybe it's time for an A/B as well.
The amps are a pretty bonehead design using a mirror-loaded JFET input stage (matched PN4393), curent source loaded voltage amp, and a current source loaded emitter follower output. Output device is a TIP35C connected as a darlington with a 2SC2235, stacked on top of a TIP142 current sink that is driven as an asymmetrical mirror with a thermally coupled TIP110 and an LM317 current source. Bias is stable at 1A with +/- 20V supply rails. The design simulated well, and should deliver enough juice to tickle about 19 ohms worth of tweeters.
I have a pair of class A tweeter amps w/electronic crossover that are ready for mounting on the heatsink. They currently have the aforementioned inductor. If I remember tomorrow at work, I'll pop one off and measure it, as well as run the impedance plot in PSpice. Maybe it's time for an A/B as well.
The amps are a pretty bonehead design using a mirror-loaded JFET input stage (matched PN4393), curent source loaded voltage amp, and a current source loaded emitter follower output. Output device is a TIP35C connected as a darlington with a 2SC2235, stacked on top of a TIP142 current sink that is driven as an asymmetrical mirror with a thermally coupled TIP110 and an LM317 current source. Bias is stable at 1A with +/- 20V supply rails. The design simulated well, and should deliver enough juice to tickle about 19 ohms worth of tweeters.
I agree that an iductor might isolate RF. But keep in mind that two parallel conductors asused in an ordinary cable are not that good an antenna. And if you want to use an air cored coil to block RF make sure that it is shielded well enough since it might be the better antenna than the cable is.
Regards
Charles
Regards
Charles
Hi Bob,
I too agree with Glen's last statement (except for his generalisation in the last sentence - "but they still don’t come without compromises") so I wonder why you are selectively highlighting/tracking comments made to me by others, without actually adding any input of value to the threads yourself.
Hi Wrenchone,
I should be pleased to read if you hear any A/B difference when the choke is bypassed, but unless you are using very low Z ribbon tweeters you may well not hear a difference because you have already filtered the drive such that the series choke is not powering a complex reactive load at all AF frequencies.
Hi Phase_accurate,
A series output choke (open air core or not) is unlikely to have any RF rejection properties due to it being paralleled with a low value resistor; the external closed ferrite ring has worked okay for me when the LS wires effectively became antennas.
Cheers ........ Graham.
I too agree with Glen's last statement (except for his generalisation in the last sentence - "but they still don’t come without compromises") so I wonder why you are selectively highlighting/tracking comments made to me by others, without actually adding any input of value to the threads yourself.
Hi Wrenchone,
I should be pleased to read if you hear any A/B difference when the choke is bypassed, but unless you are using very low Z ribbon tweeters you may well not hear a difference because you have already filtered the drive such that the series choke is not powering a complex reactive load at all AF frequencies.
Hi Phase_accurate,
A series output choke (open air core or not) is unlikely to have any RF rejection properties due to it being paralleled with a low value resistor; the external closed ferrite ring has worked okay for me when the LS wires effectively became antennas.
Cheers ........ Graham.
Cellardoor said:
Output coils of reasonable size (< 2uH) provide no protection from short circuits to either a BJT or MOSFET output stage, in my opinion.
When properly protected, MOSFETs are just as rliable as BJTs at high voltage. However, not all of the same output stage design techniques used for bipolars map well to MOSFETs. They each have their individual needs. MOSFETs can be extremely fast (faster than a fuse, for example), because they will happily try to draw several tens of amps with gate voltage of, say 15V. On the other hand, if you have a boat-load of MOSFETs in parallel, less sophisticated protection approaches may be adequate.
As an aside, I'll be out of pocket for the next few days presenting audiophile workshops at the Home Entenetainment 2007 show in New York City at the Grand Hyatt (www.he2007.com). The workshops are being jointly sponsored by Stereophile and Ray Kimber. I hope some of you in the NY area can attend.
Cheers,
Bob
Today by bob Cordell, from thread "Bob Cordell Interview: BJT vs. MOSFET",
---Also keep in mind that a speaker cable can look like an unterminated transmission line, where very radical and funny things can happen to the impedance seen looking into it at frequencies where there may be reflections, like at the quarter wave length. This depends a lot on the speaker load at the other end. Some speakers look inductive above 20 kHz, and this can give rise to the unterminated line effect at HF. In other cases, this is mitigated by the resistive tweeter pad that is usually there. For these reasons, some advocate the addition of an R-C shunt Zobel network at the speaker end of the speaker cable. This can effectively far-end terminate the speaker cable in its natural impedance at HF.---
This is the conclusion of the Cyril Bateman's recent works I mentionned in post #157. It seems that the R-C Zobel termination has been practised in Japan as well as by some manufacturers of out of price speaker cables which one end has a "magic box" containing such a network buried in black resin.
Before speaking of the audibility of cables or output coils, it could be a good idea to test this idea.
---Also keep in mind that a speaker cable can look like an unterminated transmission line, where very radical and funny things can happen to the impedance seen looking into it at frequencies where there may be reflections, like at the quarter wave length. This depends a lot on the speaker load at the other end. Some speakers look inductive above 20 kHz, and this can give rise to the unterminated line effect at HF. In other cases, this is mitigated by the resistive tweeter pad that is usually there. For these reasons, some advocate the addition of an R-C shunt Zobel network at the speaker end of the speaker cable. This can effectively far-end terminate the speaker cable in its natural impedance at HF.---
This is the conclusion of the Cyril Bateman's recent works I mentionned in post #157. It seems that the R-C Zobel termination has been practised in Japan as well as by some manufacturers of out of price speaker cables which one end has a "magic box" containing such a network buried in black resin.
Before speaking of the audibility of cables or output coils, it could be a good idea to test this idea.
Cheers Bob, though I did not say that your contributions were not worthwhile !
Hi forr,
I am sure there are some (simple) LS where the impedance rises beyond AF, but LS leads are so short that the wavelengths involved would be in the ShortWave band, and the cables themselves have inductances similar to output chokes too.
Is there anything within an amp/LS system that is going to pass the amplifier's output Zobel and then energise at such high frequencies other than the choke.
Hi John,
The voltage developed by the choke always leads current, thus the choke momentarily 'stores' energy as its field builds, and then subsequently releases it in time with waveform change; this voltage storage error and release will reverse at moments when loudspeaker back-EMF causes loudspeaker current to lead amplifier output voltage whereupon the NFB node cannot linearly control LS terminal voltage.
I remember Andy_C doing the test several years ago, but felt that the discrimination needs to run into supersonic frequencies, where standard digital examination might prove inadequate.
Cheers ........ Graham.
Hi forr,
I am sure there are some (simple) LS where the impedance rises beyond AF, but LS leads are so short that the wavelengths involved would be in the ShortWave band, and the cables themselves have inductances similar to output chokes too.
Is there anything within an amp/LS system that is going to pass the amplifier's output Zobel and then energise at such high frequencies other than the choke.
Hi John,
The voltage developed by the choke always leads current, thus the choke momentarily 'stores' energy as its field builds, and then subsequently releases it in time with waveform change; this voltage storage error and release will reverse at moments when loudspeaker back-EMF causes loudspeaker current to lead amplifier output voltage whereupon the NFB node cannot linearly control LS terminal voltage.
I remember Andy_C doing the test several years ago, but felt that the discrimination needs to run into supersonic frequencies, where standard digital examination might prove inadequate.
Cheers ........ Graham.
Guys, guys, lets not get personal... some of us like blondes, some of us like brunettes, but we don't fight over it...normally.
It is sad to see so many personal punches thrown as much as it is seeing people take things personly, which realy were not meant to be...
Now children, play nice, don't fight.
Back on topic.
Is there any significant diffirence in operation for two inductors in series, one being a small aircore inductor... the other a large inductor forced to drive against a magnetic field... I.e. one simply sits around and the other one is working with the energy...
Also... my very basic knowledge of inductors in power supplies are along the lines of they reduce ripple... shurely we don't want to reduce any ripple from our music signals?
If it does provide such filtering... is it wideband and only affecting the amplitude of the ripple , in which case I think our brains can compensate. Or is it over a small band which would cause a non linear response on an audio signal...
I say back on topic as it is about audiability, and not about the effect on stability.
It is sad to see so many personal punches thrown as much as it is seeing people take things personly, which realy were not meant to be...
Now children, play nice, don't fight.
Back on topic.
Is there any significant diffirence in operation for two inductors in series, one being a small aircore inductor... the other a large inductor forced to drive against a magnetic field... I.e. one simply sits around and the other one is working with the energy...
Also... my very basic knowledge of inductors in power supplies are along the lines of they reduce ripple... shurely we don't want to reduce any ripple from our music signals?
If it does provide such filtering... is it wideband and only affecting the amplitude of the ripple , in which case I think our brains can compensate. Or is it over a small band which would cause a non linear response on an audio signal...
I say back on topic as it is about audiability, and not about the effect on stability.
forr said:
G'day
One thing that's worth pointing out here is that a typical output inductor dampened with a small value parallel resistor will offer very little RFI suppression. The inductor may look like a high impedance to the RF, but the resistor wont.
The best output stability network for RFI suppression that I know of is the L+R/C low-pass filter circuit that has a capacitor (~47 to ~150n) in parallel with the speaker output. The capacitor very effectively shunts the RF to ground.
Just about every audio amplifier design (eg SC480, Studio 350) published in the Australian electronics mag Silicon Chip and most that I’ve seen in the now defunct "Electronics Australia" mag successfully used this arrangement.
Cheers,
Glen
Graham Maynard said:
Graham, one can safely say that practically any optimisation of an amplifier circuit to improve it's performance in one area will inevitably compromise the performance in at least one other.
Since you seem to disagree, could you perhaps highlight the exceptions to the rule, with regards to improving amplifiers driving stability into a capacitive load?
Graham Maynard said:
Thats what I just said, but I should have read your post first.
Hi Glen,
Once you have an amplifier circuit capable of driving any load in series with a few feet of cable (say 0.5uH in series) there should be no worry about 'capacitive' loads, as long as the leading current draw remains within amplifier output drive rating.
Personally I do not know of any loudspeaker not having adequate series R or L and which would not not work properly with a few feet of interconnect with such a design anyway - - that is - - as long as capacitive leads are not introduced following salesman's persuasions.
Far too many people think that increasing global NFB 'optimises' amplifier performance, and that reducing it is a compromise. If a designer does not apply so much global NFB in the first place then s/he would not have already introduced compromises within a global sensing loop which slow internal responses or lead to stage saturations due to stabilisation degraded open loop capabilities.
Removing a compromise is not a compromise.
Cheers .......... Graham.
Once you have an amplifier circuit capable of driving any load in series with a few feet of cable (say 0.5uH in series) there should be no worry about 'capacitive' loads, as long as the leading current draw remains within amplifier output drive rating.
Personally I do not know of any loudspeaker not having adequate series R or L and which would not not work properly with a few feet of interconnect with such a design anyway - - that is - - as long as capacitive leads are not introduced following salesman's persuasions.
Far too many people think that increasing global NFB 'optimises' amplifier performance, and that reducing it is a compromise. If a designer does not apply so much global NFB in the first place then s/he would not have already introduced compromises within a global sensing loop which slow internal responses or lead to stage saturations due to stabilisation degraded open loop capabilities.
Removing a compromise is not a compromise.
Cheers .......... Graham.
Graham Maynard said:
Hi Graham.
You are now arguing over wether or not it is worthwhile to design an amplifier to operate stably into a highly capacitive load. That's a different argument.
The point of contention was over the design compromises that may need to be made if an amplifier is to be made to actually do such without a load-isolating output inductor.
Cheers,
Glen
Hi Glen,
I was not aware that I was arguing ?
It seems you are seeking answers which I have already attempted to state in other threads but are not accepted, and thus have led to members of this Forum giving me such a bad time (also elsewhere starting with Douglas Self in WW/EW in 1994), so I am not going to repeat !!!
What I have stated above is that I do not accept the compromises other designers make in order to reduce THD by increasing global NFB. You already know how designers increase global NFB without me needing to say, and this is where investigation is necessary.
Look at the thread by Carlos Mergulhao, (Destroyer X),
http://www.diyaudio.com/forums/showthread.php?s=&threadid=96237&perpage=10&pagenumber=1
he has a simple design which he does not claim to be the best of its kind, but he deliberately chose to NOT include any circuits which he auditioned for degradation of audio, no matter what GBW/NFB/stability exponents might state is otherwise possible.
When I apply reverse testing to Carlos' final design to observe how it might behave in the presence of LS generated back-EMF, I find
(a) he has not compromised the internal open loop performance
(b) he does not use an output choke, and yet
(c) constructors find his design stable in use.
The amount of global feedback which might be applied to any amplifier without compromising the internal stability-component-modified-open-loop-gain is limited by the number and type of stages enclosed by the global loop and the device types used, esp. relating to carrier/charge recovery under momentary high load. Exceed that (device based) limit and simple global NFB cannot remain sufficiently coherent with source waveform.
Cheers .......... Graham.
I was not aware that I was arguing ?
It seems you are seeking answers which I have already attempted to state in other threads but are not accepted, and thus have led to members of this Forum giving me such a bad time (also elsewhere starting with Douglas Self in WW/EW in 1994), so I am not going to repeat !!!
What I have stated above is that I do not accept the compromises other designers make in order to reduce THD by increasing global NFB. You already know how designers increase global NFB without me needing to say, and this is where investigation is necessary.
Look at the thread by Carlos Mergulhao, (Destroyer X),
http://www.diyaudio.com/forums/showthread.php?s=&threadid=96237&perpage=10&pagenumber=1
he has a simple design which he does not claim to be the best of its kind, but he deliberately chose to NOT include any circuits which he auditioned for degradation of audio, no matter what GBW/NFB/stability exponents might state is otherwise possible.
When I apply reverse testing to Carlos' final design to observe how it might behave in the presence of LS generated back-EMF, I find
(a) he has not compromised the internal open loop performance
(b) he does not use an output choke, and yet
(c) constructors find his design stable in use.
The amount of global feedback which might be applied to any amplifier without compromising the internal stability-component-modified-open-loop-gain is limited by the number and type of stages enclosed by the global loop and the device types used, esp. relating to carrier/charge recovery under momentary high load. Exceed that (device based) limit and simple global NFB cannot remain sufficiently coherent with source waveform.
Cheers .......... Graham.
It is a provocation Graham..do not enter this game
The ones could understand you, those ones already did that...others will think about.
Observe the respect John Curl dedicate to you...and he is one of the best ones we have in this forum...and even beeing a famous doctor he respect you...even in doubts he accept your thougths to reflections..this is decence and respect .... and you deserve that.
Relax and be happy dear Graham.
You will much more usefull to mankind, doing your designs than be kicking stones that appear in front of your way.... stones will go rolling till be kicked once more...and you have better things to do than to destroy your shoes.... as they will continue beeing stones and you will have a damaged shoe.
I am sorry to call your attention to that...Hugh was already fished..do not bite that trap too.
There are people that use forum in a sadic way... maybe a very clever way too.
You are famous, has a history to tell folks, many published things made...also Hugh is famous...has a business, an important man..others having nothing special goes challenging you two..this will promote the Kamikazi only (famous after death)..as he has nothing so special to loose... your attention only promote those folks.
Thanks God, they are only few...we cannot count them using our hand fingers...others had good ideas from you to reflections...there are folks that laugh, as your first cicle theories seems ridiculous to their brains...they just could not understand.
This thread has a lot of respectable guys, discussing in a very high level..one is infiltrated.
regards,
Carlos
The ones could understand you, those ones already did that...others will think about.
Observe the respect John Curl dedicate to you...and he is one of the best ones we have in this forum...and even beeing a famous doctor he respect you...even in doubts he accept your thougths to reflections..this is decence and respect .... and you deserve that.
Relax and be happy dear Graham.
You will much more usefull to mankind, doing your designs than be kicking stones that appear in front of your way.... stones will go rolling till be kicked once more...and you have better things to do than to destroy your shoes.... as they will continue beeing stones and you will have a damaged shoe.
I am sorry to call your attention to that...Hugh was already fished..do not bite that trap too.
There are people that use forum in a sadic way... maybe a very clever way too.
You are famous, has a history to tell folks, many published things made...also Hugh is famous...has a business, an important man..others having nothing special goes challenging you two..this will promote the Kamikazi only (famous after death)..as he has nothing so special to loose... your attention only promote those folks.
Thanks God, they are only few...we cannot count them using our hand fingers...others had good ideas from you to reflections...there are folks that laugh, as your first cicle theories seems ridiculous to their brains...they just could not understand.
This thread has a lot of respectable guys, discussing in a very high level..one is infiltrated.
regards,
Carlos
- Status
- Not open for further replies.