| Home |
| Search |
| Today's Posts |
|
#1
|
|||
|
|||
Real Error Correction
Go to ABSE for a basic circuit which provides the error correction
functions in a vacuum tube amp. The main amp consists of a 12AT7 Diff Amp input driving a pair of 2A3's. The 2A3's are intentionally over biased so that they are running well into Class AB & into B. The side amp compares a fraction of the output with the input & then amplifies that to a level sufficient to correct the non-linearities in the output waveform. The correcting waveform in this case is inserted in series with the main output. This circuit is for demonstration only & is not optimized. I was able to recover three of Patricks attempts at error correcting from the usenet-replayer, thanx to Watt Sun's information. I have to report to all that John Byrne's comments reference these circuits is absolutely correct. They are all NFB. Cheers, John Stewart |
|
#2
|
|||
|
|||
|
John Stewart wrote: Go to ABSE for a basic circuit which provides the error correction functions in a vacuum tube amp. The main amp consists of a 12AT7 Diff Amp input driving a pair of 2A3's. The 2A3's are intentionally over biased so that they are running well into Class AB & into B. The side amp compares a fraction of the output with the input & then amplifies that to a level sufficient to correct the non-linearities in the output waveform. The correcting waveform in this case is inserted in series with the main output. This circuit is for demonstration only & is not optimized. I was able to recover three of Patricks attempts at error correcting from the usenet-replayer, thanx to Watt Sun's information. I have to report to all that John Byrne's comments reference these circuits is absolutely correct. They are all NFB. I disagree here, and no need for me to repeat *all* my objections. I cannot see your latest post to ABSE, and it is not yet displayed at http://www.usenet-replayer.com/cgi/content/archive I guess in the days ahead it may appear there. But I have some general comments about error correction and NFB.... In many if not most EC schemes, there is a reference signal which is almost free of distortion and which is generated from the amp input signal. The reference signal may even be the input signal itself. From this ref sig a pair of R in series are taken to the output signal of the amp, which has its phase opposite to the reference signal, and the two R are sized in such a way as to null out most if not all the wanted signal, ie the music or test tone used does not appear at the junction of the two R. Error correction basically nulls +ve phase and -ve phase of the same signal, and extracts the distortion signal contained in one of the phases, ie, the output phase, and this Dn can be amplified and re-injected into the signal path at any convenient point we choose, so that it cancels the open loop distortions. But also associated with any nulled Dn voltage is the compensation signal voltage to control the output resistance of the amp. The Ro compo signal is usually a far smaller signal than that which is fed back in conventional NFB arrangements. EC still requires us to get our relative phases correct where applying amplified Dn signals back into the amp. If wrong, we build a good oscillator. At one particular value of load, the amplified or recovered Dn voltage has no signal voltage with it. But at all other values of load, some signal voltage will appear, and with speakers being all sorts of values at different F, the dn and compo signal will be greater always than the nulled signal when using a sine wave where open loop phase shift is zero, and the load is just right. When there is no wanted signal voltage to amplify and re-apply to the amp, there is no NFB. But nevertherless the distortion gets fed back to the amp at some point, to cancel the open loop distortion, and one could always call it NFB, but there is no wanted signal fraction always associated with normal NFB applications, hence my perception of a difference between normal NFB application and error correction. In most series voltage NFB amps, such as a normal cathode follower stage, all of the output voltage is fed back in series with the grid input voltage, and it includes any thd, which is amplified by the cathode follower's open loop gain and then subtracted from the open loop thd at the output. So a 6SN7 with open loop gain of 15 with a 33k load will have a closed loop gain of 16/15 when used as a CF, and the open loop thd that does appear in the CF is amplified to cancel most, but never all of the distortion that would have appeared had the 33k been used in the anode circuit without any NFB operating. There may well be some way of construcing a second low thd gain amp, and with a pair of R we extract the thd of the CF, and amplify this thd. We can have a series R between the RCA input terminal to to the CF grid. We can then apply the amplified thd signal from our CF via a current source, to the CF grid, via say a pentode anode circuit, so as to not attenuate the wanted signal from the rca input terminal. Thus the CF ends up being a super CF, with much lower than normal thd, and lower Ro. But just remember, all that NFB and error correction could cause oscillations. Its all a bit complex though, and since CF thd is usually low enough at the 1vrms output level, I don't see any point, but its just an example of what might be achieved. A pentode used as a CF and used so it runs as a real pentode, not a trioded pentode will make a pretty fine CF, since if the load was 33k, and the tube a 6AU6, the open loop gain might be 100, so the distortion reduction factor would be 1/101, so if there was 0.1% open loop thd at 1 v, it'd be 0.001% with a CF, so not much need to add any error correction, since noise would ruin any benefits. Patrick Turner. |
|
#3
|
|||
|
|||
|
John Larkin wrote: On Thu, 23 Sep 2004 09:56:04 -0400, John Stewart wrote: Go to ABSE for a basic circuit which provides the error correction functions in a vacuum tube amp. The main amp consists of a 12AT7 Diff Amp input driving a pair of 2A3's. The 2A3's are intentionally over biased so that they are running well into Class AB & into B. The side amp compares a fraction of the output with the input & then amplifies that to a level sufficient to correct the non-linearities in the output waveform. The correcting waveform in this case is inserted in series with the main output. This circuit is for demonstration only & is not optimized. I was able to recover three of Patricks attempts at error correcting from the usenet-replayer, thanx to Watt Sun's information. I have to report to all that John Byrne's comments reference these circuits is absolutely correct. They are all NFB. Cheers, John Stewart Most modern cellphone base-station power amps use feedforward distortion correction vaguely similar to this. A small distortion signal is created that's summed into the final output, canceling the native distortion of the power amp. But at these speeds, subtraction isn't practical, so the distortion is created by nonlinear parts, and tweaked to the right amount, so there's low net distortion without feedback. John Your comment regarding cell phone amps sound like an awful solution to try to get hi-fi performance, but I doubt there is the slightest simularity in what John Stewart's "real error correction" actually does, if it can actually be determined if it does anything at all. After examining the circuit which turned up at the usenet replayer archives site today, I found it very hard to follow because the individual working signal voltages are not indicated, with their relevant phases, which is very important for those even with a keen eye who can fill in most of the missing details that should be there. There is nothing unusual with the PP power amp which has a 12AT7 LTP and pair of 2A3 OPVs transformer coupled to the output. There is no NFB in this amp forming part of the whole circuit, or any attempt at error correction within the loop of the 2A3 amp. No attempt is made to use the available gain in the 2A3 power amp to perform the task of amplifying a phase version of the distortion in such a way to cancel its production, which happens in FB amps, or in those which have error correction within the open loop. We could assume the input voltage to this power amp is 2vrms, and signals at the AT7 anodes to be +/- 32v, then -/+100v at the A3 anodes, and if the OPT is 6k to 6 ohms, we could expect 6.3v at the output for 6.6 watts, class A1, depending on the bias. At the 2A3 OPT, which is labelled with a blue'1', the lower sec leg is grounded, and the upper leg goes to a leg of a secondary of another complete mystery transformer which is powered from a 6EM7. The 6EM7 triode is driven by a non inverting 12AT7 pair. The 2v input voltage is applied to one side of the pair, and a signal from the 100 ohms in the 5k, 9k, 100 ohm series divider. The resultant at the pair signal produces a drive signal to the 6EM7 to then produce what might be a low impedance signal at the 6EM7 OPT, at the top, marked with a blue 10. None of how the error correction is supposed to work is clearly visible on the schematic, but JH's original post at r.a.t. said only " The main amp consists of a 12AT7 Diff Amp input driving a pair of 2A3's. The 2A3's are intentionally over biased so that they are running well into Class AB & into B. The side amp compares a fraction of the output with the input & then amplifies that to a level sufficient to correct the non-linearities in the output waveform. The correcting waveform in this case is inserted in series with the main output." So from what JH says, the output signal to the speaker or RL which is not shown must be between the top leg of the sec of 6EM7 OPT at the top, and the bottom leg of the bottom 2A3 OPT, or 0V. In other words, the output signal to a speaker is taken from across the 5 k resistor at the 5k, 9k 100 ohms divider seen near the 2A3 OPT. he connected CRO is monitoring the low impedance signal taken from the the 6EM7 OPT, ie, it is monitoring the "anti distortion signal" whose current is added to the current from the 2A3 OPT. The load current must flow in the two secs of each OPT, and we are left to guess what voltage reflection interaction will occur in the 6EM7 anode circuit. My guess is that the 6EM7 OPT would have to be a very low secondary output resistance so the load current won't change the signal generated at the 6EM7 anode, which some how is supposed to be a cancelling version of the distortion signal which exists in the 2A3 amp output signal. It would make things a lot clearer to understand if JH were to number each R&C throughout his theoretical amp, and then provide all the working signal voltages, and to provide a little more info rather than leave everyone bemused and befuddled. He sometimes complains about my earnest attempts to explain things which take pages he'd rather not read; he expects me to miraculously explain all in 10 lines, but I don't like to be a bamboozler, and so I try to include all the details, and if I need to dish out 100 lines, then so be it. Perhaps we could all persuade our John, who is our learned and much respected friend, that he too spend a little more time at his keyboard to make it impossible for his inventions to be misunderstood. Circuits like this may have a very valid basis, and deserve some careful presentation. Patrick Turner. |
|
#4
|
|||
|
|||
|
On Fri, 24 Sep 2004 23:25:17 +1000, Patrick Turner
wrote: John Larkin wrote: On Thu, 23 Sep 2004 09:56:04 -0400, John Stewart wrote: Go to ABSE for a basic circuit which provides the error correction functions in a vacuum tube amp. The main amp consists of a 12AT7 Diff Amp input driving a pair of 2A3's. The 2A3's are intentionally over biased so that they are running well into Class AB & into B. The side amp compares a fraction of the output with the input & then amplifies that to a level sufficient to correct the non-linearities in the output waveform. The correcting waveform in this case is inserted in series with the main output. This circuit is for demonstration only & is not optimized. I was able to recover three of Patricks attempts at error correcting from the usenet-replayer, thanx to Watt Sun's information. I have to report to all that John Byrne's comments reference these circuits is absolutely correct. They are all NFB. Cheers, John Stewart Most modern cellphone base-station power amps use feedforward distortion correction vaguely similar to this. A small distortion signal is created that's summed into the final output, canceling the native distortion of the power amp. But at these speeds, subtraction isn't practical, so the distortion is created by nonlinear parts, and tweaked to the right amount, so there's low net distortion without feedback. John Your comment regarding cell phone amps sound like an awful solution to try to get hi-fi performance, No argument there. but I doubt there is the slightest simularity in what John Stewart's "real error correction" actually does, if it can actually be determined if it does anything at all. The idea of "feed-beside" error correction isn't new, but it only makes sense when true nfb isn't practical, which is mainly the case for very fast circuits. It would make things a lot clearer to understand if JH were to number each R&C throughout his theoretical amp, and then provide all the working signal voltages, and to provide a little more info rather than leave everyone bemused and befuddled. Why do I keep getting the impression that audio people seldom actually understand electronics? John |
|
#5
|
|||
|
|||
|
On Fri, 24 Sep 2004 08:16:35 -0700, John Larkin
wrote: [snip] Why do I keep getting the impression that audio people seldom actually understand electronics? John ROTFLMAO! ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice  480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. |
|
#6
|
|||
|
|||
|
I read in alt.binaries.schematics.electronic that Jim Thompson
wrote (in npe8l0ttmcnpk7pbpa63jenpsqdasgf7oh@ 4ax.com) about 'Real Error Correction', on Fri, 24 Sep 2004: On Fri, 24 Sep 2004 08:16:35 -0700, John Larkin jjlarkin@highlandSNIPte chTHISnologyPLEASE.com wrote: [snip] Why do I keep getting the impression that audio people seldom actually understand electronics? John ROTFLMAO! I haven't seen JL's article, but I would say that it's fairly reciprocal. A lot of electronics people don't understand audio. While it may be simpler than some other branches of electronics, it's not by any means trivial and can be very subtle. -- Regards, John Woodgate, OOO - Own Opinions Only. The good news is that nothing is compulsory. The bad news is that everything is prohibited. http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk |
|
#7
|
|||
|
|||
|
On Fri, 24 Sep 2004 17:03:27 +0100, John Woodgate
wrote: I read in alt.binaries.schematics.electronic that Jim Thompson wrote (in npe8l0ttmcnpk7pbpa63jenpsqdasgf7oh@ 4ax.com) about 'Real Error Correction', on Fri, 24 Sep 2004: On Fri, 24 Sep 2004 08:16:35 -0700, John Larkin jjlarkin@highlandSNIPte chTHISnologyPLEASE.com wrote: [snip] Why do I keep getting the impression that audio people seldom actually understand electronics? John ROTFLMAO! I haven't seen JL's article, but I would say that it's fairly reciprocal. A lot of electronics people don't understand audio. While it may be simpler than some other branches of electronics, it's not by any means trivial and can be very subtle. Perhaps we're not expert in acoustics and tuned boxes, but we don't claim to be. Audiphools act as if they know something about electronics when they actually know ZERO... to wit, the on-going "error correction" thread... wildest malarkey I've seen in years... so bad it makes Paul Burridge appear educated ;-) ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. |
|
#8
|
|||
|
|||
|
In article , Patrick Turner
wrote: I have some general comments about error correction and NFB.... In many if not most EC schemes, there is a reference signal which is almost free of distortion and which is generated from the amp input signal. The reference signal may even be the input signal itself. From this ref sig a pair of R in series are taken to the output signal of the amp, which has its phase opposite to the reference signal, and the two R are sized in such a way as to null out most if not all the wanted signal, ie the music or test tone used does not appear at the junction of the two R. Patrick, you seem to be flip-flopping here, but I suppose the "if not all" part of the "null out most if not all the wanted signal" is your out. In your first explanation of "error correction" the main amplifier had a differential input, and the input signal was feed to the non-inverting input, while the signal at the junction of the two resistors you describe above was feed to the inverting input. You made a point at that time that the wanted signal at the junction of the two resistors was not nulled out, but that the resistors were adjusted so that the wanted signal at the inverting input was equal in amplitude, and opposite in phase, to the input signal at the non-inverting input. Your flip-flop on that is really of no consequence, but lets consider the circuit a little further. Implicit in your description above is that the input signal is feed into the amplifier at some unspecified point. Let's call the junction of the two resistors you mention the "summing junction". Would it not be legitimate to feed the input signal through a third resistor, in the appropriate phase, into this same "summing junction at the inverting input? This would seem to be legitimate within your description of the "error correction" topology given above. If it is legitimate then the two resistors from the input to the summing junction are in parallel and we should be able to combine them into a single resistor equal to the parallel equivalent of the two separate resistors. The resulting topology is a standard well known NFB circuit, hence your "error correction" circuit is nothing more than plain old NFB. I would consider feed forward, as used in the circuit proposed by JS to be a true "error correction" topology. Malcom Hawksford's circuit may be another form of "error correction", although I don't fully understand Malcom's circuit which relies on positive feedback for its operation, so I am not positive it qualifies. It is clear that your circuit is nothing more than ordinary NFB disguised with some extra circuit complexity and a lot of fancy words to bamboozle the unwary. Error correction basically nulls +ve phase and -ve phase of the same signal, and extracts the distortion signal contained in one of the phases, ie, the output phase, and this Dn can be amplified and re-injected into the signal path at any convenient point we choose, so that it cancels the open loop distortions. But also associated with any nulled Dn voltage is the compensation signal voltage to control the output resistance of the amp. In what significant way is this any different than ordinary NFB, you have yet to explain the difference? What is the essential difference in the two circuits? It would be instructive if you could contrast your "error correction" circuit with the standard NFB circuit, the feed forward error correction circuit, and Hawksford's circuit, and point out in what ways they are similar, and in what ways they are different, in other than cosmetic ways. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
|
#9
|
|||
|
|||
|
Why do I keep getting the impression that audio people seldom actually understand electronics? John Not all of us are dummies. But its remarkable how few ppl ever try to fiddle and tweak their TV to get a better picture, or mess with their mobile telephone circuits. I know of none who have built TVs, or mobile phones. Audio is simple enough to attract folks who wanna fiddle and tweak, and apart from a CD player, its possible to DIY an amp, speakers and even a TT. Most get infatuated with the idea of DIY, like the dudes trying to build their own house, to save paying the experts, and to enjoy the experience, to feel connected to the universe, not one or two steps removed from it. Some go bush and need to milk a cow each morning to feel whole, and to they know SFA about farming, since they never studied agriculture. ( I have other preferences for early morning tit contact ) I am happy with my city existance and my dependance on farmers and the experts, but I have spent years of study on electronics, enough at least to make a living from audio manufacturing. I used to be a professional builder until crook knees forced me out of the trade, so I learnt a new one which allows me to sit while I work for 1/2 the time. Partrick Turner. |
|
#10
|
|||
|
|||
|
John Byrns wrote: In article , Patrick Turner wrote: I have some general comments about error correction and NFB.... In many if not most EC schemes, there is a reference signal which is almost free of distortion and which is generated from the amp input signal. The reference signal may even be the input signal itself. From this ref sig a pair of R in series are taken to the output signal of the amp, which has its phase opposite to the reference signal, and the two R are sized in such a way as to null out most if not all the wanted signal, ie the music or test tone used does not appear at the junction of the two R. Patrick, you seem to be flip-flopping here, but I suppose the "if not all" part of the "null out most if not all the wanted signal" is your out. In your first explanation of "error correction" the main amplifier had a differential input, and the input signal was feed to the non-inverting input, while the signal at the junction of the two resistors you describe above was feed to the inverting input. You made a point at that time that the wanted signal at the junction of the two resistors was not nulled out, but that the resistors were adjusted so that the wanted signal at the inverting input was equal in amplitude, and opposite in phase, to the input signal at the non-inverting input. The option of using the error amplifier as a phase inverter is there if you want it to. I made all that clear. If all the wanted signal is nulled by the two R network, then its only truly possible at one load value, and one F, due to phase shift. But at all different load values and F there will be some signal at the null point, and this **small** signal will form what is the correction signal along with the distortion signal and applied back into the amp and do its job just like NFB will. Your flip-flop on that is really of no consequence, but lets consider the circuit a little further. Implicit in your description above is that the input signal is feed into the amplifier at some unspecified point. There are several ways of applying the error signal. Let's call the junction of the two resistors you mention the "summing junction". Would it not be legitimate to feed the input signal through a third resistor, in the appropriate phase, into this same "summing junction at the inverting input? One would need more hardware than I have proposed, so why do it your way? Post a schematic, with a full explantion please. This would seem to be legitimate within your description of the "error correction" topology given above. If it is legitimate then the two resistors from the input to the summing junction are in parallel and we should be able to combine them into a single resistor equal to the parallel equivalent of the two separate resistors. The resulting topology is a standard well known NFB circuit, hence your "error correction" circuit is nothing more than plain old NFB. The action of the error correction circuits I have proposed is the **same** as series voltage negative feedback, OK, but I have not fed back a signal nearly as big as input signal to a diff input stage. There is a subtle difference and you just wanna just dumb down what I have proposed. I can't call it negative feedback, because its not the same. There is nothing like it in RDH4. So I have to pick a name, so error correction is the category I choose. There has been a vast array of ideas about error correction between when they wrote RDH4 in the tube era, and the present day. Most have centered around solid state. Arguments have raged about the terminology, and what a lot of boring noise all that was. Everyone wanted to invent something new and different, to be the new kid on the block and resented anyone else for doing so, (like you perhaps). I must forgive silly human nature most days..... I would consider feed forward, as used in the circuit proposed by JS to be a true "error correction" topology. JS's idea isn't feed forward. He extracts the distortion of an amp1, amplifies it in amp2, and applies it in series with the signal from amp1, so you have two amps both with near as possible equal distortion voltages appearing at their outputs, so there is no distortion voltage across the load. no distortion current either. Output resistance is kept to as near zero ohms as possible. Because an error signal isn't re-injected into the gain circuit of amp1 anywhere, its possible to have better stability than with NFB. Malcom Hawksford's circuit may be another form of "error correction", although I don't fully understand Malcom's circuit which relies on positive feedback for its operation, so I am not positive it qualifies. It is clear that your circuit is nothing more than ordinary NFB disguised with some extra circuit complexity and a lot of fancy words to bamboozle the unwary. I have made it extraordinarly clear how my ideas work. Mr Hawksford is another bamboozler of the first order. The reason you can't understand his ideas is that perhaps you are a bit slow, but mainly because Hawksford can't be bothered to spell things out to ordinary mortals; he can only be understood by those on a similar level of intellectual capacity, due to their math ability, and qualifications gained at universities. We all flounder in the presence of such fellows, who have taken electronics far beyond the cutting edge of audio amplifiers as it was in 1955, when a good Williamson was king, and nothing else mattered. But a decent Williamson still sounds well compared to anything Hawksford may have designed. It also measures well enough. Putting more zeros in the THD results does not greatly improve the sound. Hawksford has written on numerous other aspects of audio, both in the analog domain, and even more perplexingly, in the digital domain, and I have no intention of following what this brilliant long haired git has to say because its all too hard for me to decypher it all. He has no interest in explaining it in a way which allows us to apply his ideas easily, and he'd probably think we were ****ing into the wind by experimenting with novel or at least different tube circuit ideas. And his ideas may not be easily applied, even if expresed simply enough. Much modern electronics is like that. I don't give a damn. Error correction basically nulls +ve phase and -ve phase of the same signal, and extracts the distortion signal contained in one of the phases, ie, the output phase, and this Dn can be amplified and re-injected into the signal path at any convenient point we choose, so that it cancels the open loop distortions. But also associated with any nulled Dn voltage is the compensation signal voltage to control the output resistance of the amp. In what significant way is this any different than ordinary NFB, you have yet to explain the difference? What is the essential difference in the two circuits? I won't repeat myself for the 4th time. It would be instructive if you could contrast your "error correction" circuit with the standard NFB circuit, the feed forward error correction circuit, and Hawksford's circuit, and point out in what ways they are similar, and in what ways they are different, in other than cosmetic ways. I have said plenty enough. I don't give a **** about your dismissivisms. Patrick Turner. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
|
#11
|
|||
|
|||
|
Jim Thompson wrote: On Fri, 24 Sep 2004 17:03:27 +0100, John Woodgate wrote: I read in alt.binaries.schematics.electronic that Jim Thompson wrote (in npe8l0ttmcnpk7pbpa63jenpsqdasgf7oh@ 4ax.com) about 'Real Error Correction', on Fri, 24 Sep 2004: On Fri, 24 Sep 2004 08:16:35 -0700, John Larkin jjlarkin@highlandSNIPte chTHISnologyPLEASE.com wrote: [snip] Why do I keep getting the impression that audio people seldom actually understand electronics? John ROTFLMAO! I haven't seen JL's article, but I would say that it's fairly reciprocal. A lot of electronics people don't understand audio. While it may be simpler than some other branches of electronics, it's not by any means trivial and can be very subtle. Perhaps we're not expert in acoustics and tuned boxes, but we don't claim to be. Audiphools act as if they know something about electronics when they actually know ZERO... to wit, the on-going "error correction" thread... wildest malarkey I've seen in years... so bad it makes Paul Burridge appear educated ;-) ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. Thanx to all who took a look at my musings. BTW, what does ROTFLMAO mean? This old guy still needs some Education. Cheers, JLS |
|
#12
|
|||
|
|||
|
"John Stewart" wrote in message ... Jim Thompson wrote: On Fri, 24 Sep 2004 17:03:27 +0100, John Woodgate wrote: I read in alt.binaries.schematics.electronic that Jim Thompson wrote (in npe8l0ttmcnpk7pbpa63jenpsqdasgf7oh@ 4ax.com) about 'Real Error Correction', on Fri, 24 Sep 2004: On Fri, 24 Sep 2004 08:16:35 -0700, John Larkin jjlarkin@highlandSNIPte chTHISnologyPLEASE.com wrote: [snip] Why do I keep getting the impression that audio people seldom actually understand electronics? John ROTFLMAO! I haven't seen JL's article, but I would say that it's fairly reciprocal. A lot of electronics people don't understand audio. While it may be simpler than some other branches of electronics, it's not by any means trivial and can be very subtle. Perhaps we're not expert in acoustics and tuned boxes, but we don't claim to be. Audiphools act as if they know something about electronics when they actually know ZERO... to wit, the on-going "error correction" thread... wildest malarkey I've seen in years... so bad it makes Paul Burridge appear educated ;-) ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. Thanx to all who took a look at my musings. BTW, what does ROTFLMAO mean? This old guy still needs some Education. Cheers, JLS ROTFLMAO :: Rolling on the floor laughing my ass off. Charles |
|
#13
|
|||
|
|||
|
"Charles W. Johson Jr." wrote: "John Stewart" wrote in message ... Jim Thompson wrote: On Fri, 24 Sep 2004 17:03:27 +0100, John Woodgate wrote: I read in alt.binaries.schematics.electronic that Jim Thompson wrote (in npe8l0ttmcnpk7pbpa63jenpsqdasgf7oh@ 4ax.com) about 'Real Error Correction', on Fri, 24 Sep 2004: On Fri, 24 Sep 2004 08:16:35 -0700, John Larkin jjlarkin@highlandSNIPte chTHISnologyPLEASE.com wrote: [snip] Why do I keep getting the impression that audio people seldom actually understand electronics? John ROTFLMAO! I haven't seen JL's article, but I would say that it's fairly reciprocal. A lot of electronics people don't understand audio. While it may be simpler than some other branches of electronics, it's not by any means trivial and can be very subtle. Perhaps we're not expert in acoustics and tuned boxes, but we don't claim to be. Audiphools act as if they know something about electronics when they actually know ZERO... to wit, the on-going "error correction" thread... wildest malarkey I've seen in years... so bad it makes Paul Burridge appear educated ;-) ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. Thanx to all who took a look at my musings. BTW, what does ROTFLMAO mean? This old guy still needs some Education. Cheers, JLS ROTFLMAO :: Rolling on the floor laughing my ass off. Charles Thanx Charles. Looks like a good on to be used here from time to time. Here is another that used to work on some girls I knew-- B 4 I 4 Q R ( U / 18 ) Q T P I To be read phonetically. Perhaps carefully, as well!! Cheers, JLS |
|
#14
|
|||
|
|||
|
Patrick Turner wrote:
I took a few days to examine the graphics of the outside world here. Not too suprisingly, it sure beats what is on the iNet. The bicycle got a real workout, 40 to 50 miles per day. That used to be easy. It isn't anymore!!! Your comment regarding cell phone amps sound like an awful solution to try to get hi-fi performance, but I doubt there is the slightest simularity in what John Stewart's "real error correction" actually does, if it can actually be determined if it does anything at all. After examining the circuit which turned up at the usenet replayer archives site today, I found it very hard to follow because the individual working signal voltages are not indicated, with their relevant phases, which is very important for those even with a keen eye who can fill in most of the missing details that should be there. Settle down now Patrick. I would have thought a guy of your claimed technical caliber could have sorted out that circuit in a millisecond or less. In any case, I believe the simple description I gave in the original post sez it all. Why get complicated? One doesn't need a lot of calcs & figures to get the gist of that circuit. And as I said, it is not optomized, so one would need to do a bit work to get it running. It does fulfill all of the points made in a paper on the subject by Peter J Walker. Only difference is that the correction is applied in series with the main signal rather than parallel at the output. I had not read the Walker paper, otherwise I might have tried the parallel route. There is nothing unusual with the PP power amp which has a 12AT7 LTP and pair of 2A3 OPVs transformer coupled to the output. That is why I used this simple topology in the example I posted. Would the circuit have been anymore understandable had I used 300B's or 304TH's? Cheers to all, John Stewart Hope we are all learning a bit here!!! |
|
#15
|
|||
|
|||
|
I took a few days to examine the graphics of the outside world here. Not too suprisingly, it sure beats what is on the iNet. The bicycle got a real workout, 40 to 50 miles per day. That used to be easy. It isn't anymore!!! Your comment regarding cell phone amps sound like an awful solution to try to get hi-fi performance, but I doubt there is the slightest simularity in what John Stewart's "real error correction" actually does, if it can actually be determined if it does anything at all. After examining the circuit which turned up at the usenet replayer archives site today, I found it very hard to follow because the individual working signal voltages are not indicated, with their relevant phases, which is very important for those even with a keen eye who can fill in most of the missing details that should be there. Settle down now Patrick. I would have thought a guy of your claimed technical caliber could have sorted out that circuit in a millisecond or less. In any case, I believe the simple description I gave in the original post sez it all. Why get complicated? One doesn't need a lot of calcs & figures to get the gist of that circuit. And as I said, it is not optomized, so one would need to do a bit work to get it running. It does fulfill all of the points made in a paper on the subject by Peter J Walker. Only difference is that the correction is applied in series with the main signal rather than parallel at the output. I had not read the Walker paper, otherwise I might have tried the parallel route. There is nothing unusual with the PP power amp which has a 12AT7 LTP and pair of 2A3 OPVs transformer coupled to the output. That is why I used this simple topology in the example I posted. Would the circuit have been anymore understandable had I used 300B's or 304TH's? Cheers to all, John Stewart Hope we are all learning a bit here!!! |
|
#16
|
|||
|
|||
|
I read in alt.binaries.schematics.electronic that Charles W. Johson Jr.
wrote (in ..earthlink.net) about 'Real Error Correction', on Sat, 25 Sep 2004: Jim Thompson wrote, but I didn't get the article: Perhaps we're not expert in acoustics and tuned boxes, but we don't claim to be. Audiphools act as if they know something about electronics when they actually know ZERO... to wit, the on-going "error correction" thread... wildest malarkey I've seen in years Error correction in audio was first proposed, I think, by E C Cherry (if Paul Voight didn't think of it first, as he did for a vast range of things) and has subsequently been studied by Malcolm Hawksford. Neither of them is in any way an audiophool. ... so bad it makes Paul Burridge appear educated ;-) Nothing, but nothing, could do that. ...Jim Thompson -- -- Regards, John Woodgate, OOO - Own Opinions Only. The good news is that nothing is compulsory. The bad news is that everything is prohibited. http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk |
|
#17
|
|||
|
|||
|
On Sat, 25 Sep 2004 19:45:52 +0100, John Woodgate
wrote: I read in alt.binaries.schematics.electronic that Charles W. Johson Jr. wrote (in .earthlink.net) about 'Real Error Correction', on Sat, 25 Sep 2004: Jim Thompson wrote, but I didn't get the article: Perhaps we're not expert in acoustics and tuned boxes, but we don't claim to be. Audiphools act as if they know something about electronics when they actually know ZERO... to wit, the on-going "error correction" thread... wildest malarkey I've seen in years Error correction in audio was first proposed, I think, by E C Cherry (if Paul Voight didn't think of it first, as he did for a vast range of things) and has subsequently been studied by Malcolm Hawksford. Neither of them is in any way an audiophool. ... so bad it makes Paul Burridge appear educated ;-) Nothing, but nothing, could do that. I am sure that Cherry's discussions are at least several orders of magnitude more erudite than the circular reasoning of Patrick turner ;-) ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. |
|
#18
|
|||
|
|||
|
John Larkin wrote:
Most modern cellphone base-station power amps use feedforward distortion correction vaguely similar to this. A small distortion signal is created that's summed into the final output, canceling the native distortion of the power amp. But at these speeds, subtraction isn't practical, so the distortion is created by nonlinear parts, and tweaked to the right amount, so there's low net distortion without feedback. John For fans of RF & MW go to ABSE to see a real MW Predistorter. This one is= new, fairly unique & appears to do the job very well. I apologize to those on wireline as I am. Even after compression the four pages still occupy arou= nd 700K while on the net. This is the world I lived in for the past 20 years. Not the circuits but = the many advanced pieces of TE that are required to develop, install & mainta= in the resulting communication systems that we all seem to be addicted too now. After leaving HP in the spring of 1984 I joined Rohde & Schwarz. The headhunters were as busy then as they are today. R&S was in the cell phon= e test game from the get go. At R&S I saw for the first time ever an advanced (f= or that time) Radio Communication Test Set, their model SMFP2. It was microprocessor controlled & had many built in automated test functions. U= nlike the American made Cushman, it was also of laboratory precision. When hook= ed up through the IEC625/IEEE488 to a controller such as the Commodore 64 a completely automated test was possible. For that reason R&S had written a= complete suite of tests one could pick from a file. The whole thing ran i= n BASIC. Rather than getting tied too Commodore, R&S had their own rack mou= nted computer, the PUC. A souped up version, the SCUD included all the signali= ng required for AMPS & something like six other European standards. The whol= e thing was a Base Station Simulator that could talk straight to a cell pho= ne. Things moved very quickly so that we soon had a slick new Radiocomm Teste= r, the CMT. The CMT could memorize a complete sequence of manually performed tests & = output all results & warnings to a printer. It didn=92t need a separate controll= er so that it was a lot easier for the techs to drag around. Upgrading the firm= ware was relatively easy as well. When I had to I changed those IC=92s in a ho= tel room. Next came the CMS, even smarter & more compact. But by now cell phones we= re getting cheap to build & more difficult to fix. The expertise of techs in= the field seem to go down at that level as well. Another surprise at R&S was their FSA Spectrum Analyzer. After something = like 20 years without a Spec A they were back in a big way. When I could demo = =96150 dbm our customers were amazed. Much better than anything HP had at that t= ime. That line has many products & continues today. In 1994 R&S sales & marketing in North America went to TEK. Many do not k= now that TEK & R&S have over many years had agreements such as this covering various parts of the world. The first salesman I ever met offering R&S pr= oducts was a TEK sales guy, Marvin Crouch. That would be around 1960. Anyway, when sales went to TEK, some of us had no job, a situation famili= ar to many others today. I lost no time at all since I immediately joined the N= avair Division of Field Aviation. Their main product line at the time was IFR o= ut of Wichita, so with the Radiocomm Testers & the Spectrum Analyzers it was a = good fit for me. It became obvious the market was changing so the guys at Wich= ita made it possible to test base stations as well as cell phones. That was a= ll done by an advanced version of the COM-120B. Trouble was it did TDMA, while the trend in this part of the world was CD= MA. Some of us are lucky enough to get old, so I retired in 2000. I=92ve been= back twice to fill in for the guys who didn=92t make in my old job, as recentl= y as the past spring. Working in hitech sales for almost 40 years has been a blast= =2E Cheers, John Stewart |
|
#19
|
|||
|
|||
|
I read in alt.binaries.schematics.electronic that Jim Thompson
wrote (in 0igbl01e6h5ucg8v2a8lulsakd1ks3ha0g@ 4ax.com) about 'Real Error Correction', on Sat, 25 Sep 2004: I am sure that Cherry's discussions are at least several orders of magnitude more erudite than the circular reasoning of Patrick turner ;-) I didn't comment on PT's contributions. Maybe we could get Kevin to translate them into plain English. (;-) -- Regards, John Woodgate, OOO - Own Opinions Only. The good news is that nothing is compulsory. The bad news is that everything is prohibited. http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk |
|
#20
|
|||
|
|||
|
On Sat, 25 Sep 2004 21:07:49 +0100, John Woodgate
wrote: I didn't comment on PT's contributions. Maybe we could get Kevin to translate them into plain English. (;-) --- Playin' English?^) -- John Fields |
|
#21
|
|||
|
|||
|
In article , Patrick Turner
wrote: John Byrns wrote: Patrick, you seem to be flip-flopping here, but I suppose the "if not all" part of the "null out most if not all the wanted signal" is your out. In your first explanation of "error correction" the main amplifier had a differential input, and the input signal was feed to the non-inverting input, while the signal at the junction of the two resistors you describe above was feed to the inverting input. You made a point at that time that the wanted signal at the junction of the two resistors was not nulled out, but that the resistors were adjusted so that the wanted signal at the inverting input was equal in amplitude, and opposite in phase, to the input signal at the non-inverting input. The option of using the error amplifier as a phase inverter is there if you want it to. What "error amplifier"? In your original design there were only two amplifiers, the "main" amplifier with a differential input, and the "auxiliary phase inverter", which is the "error amplifier"? If all the wanted signal is nulled by the two R network, then its only truly possible at one load value, and one F, due to phase shift. But at all different load values and F there will be some signal at the null point, and this **small** signal will form what is the correction signal along with the distortion signal and applied back into the amp and do its job just like NFB will. Indeed, it is "just like NFB", precisely because it is NFB, all dressed up with a fancy new name. Your flip-flop on that is really of no consequence, but lets consider the circuit a little further. Implicit in your description above is that the input signal is feed into the amplifier at some unspecified point. There are several ways of applying the error signal. Let's call the junction of the two resistors you mention the "summing junction". Would it not be legitimate to feed the input signal through a third resistor, in the appropriate phase, into this same "summing junction at the inverting input? One would need more hardware than I have proposed, so why do it your way? Post a schematic, with a full explantion please. How do you figure more hardware is required? It appears to me that there is no significant difference in the amount of hardware required. You can find the schematic for my design at this URL: http://users.rcn.com/jbyrns/stuff/NFBamp.jpg As you can see the same amount of hardware is required either way. I have omitted the details of the V1 stage in my design because I haven't yet been able to find a tube for V1 that will provide the same 1.7 volt input sensitivity as with your design. If a dual 6SN7 stage is used as in your design, then the input sensitivity decreases to about 1.96 volts from 1.7 volts in your amplifier. On the other hand there are several tubes that could be used for V1 which approximately double the sensitivity of my design over yours, but I can't find any tubes that give your original input sensitivity. The action of the error correction circuits I have proposed is the **same** as series voltage negative feedback, OK, but I have not fed back a signal nearly as big as input signal to a diff input stage. You can't have it both ways, it's either the same or it's not. The only difference between the different topologies using a differential input is the amount of common mode signal applied to the input of the differential amplifier stage. There is a subtle difference and you just wanna just dumb down what I have proposed. You still haven't explained the supposed "subtle difference", most likely because there isn't any difference at all, subtle or not. I can't call it negative feedback, because its not the same. Sure you can, just repeat after me, "it's just ordinary negative feedback." Now that wasn't so hard was it? There is nothing like it in RDH4. I suggest that you carefully read chapter 7 of the RDH4, and then look at your circuit again. So I have to pick a name, so error correction is the category I choose. You only "have to pick a name" because you are imagining a difference that isn't there. As my wife would say, "you have cobwebs over your eyes". I would consider feed forward, as used in the circuit proposed by JS to be a true "error correction" topology. JS's idea isn't feed forward. He extracts the distortion of an amp1, amplifies it in amp2, and applies it in series with the signal from amp1, so you have two amps both with near as possible equal distortion voltages appearing at their outputs, so there is no distortion voltage across the load. no distortion current either. Output resistance is kept to as near zero ohms as possible. If that isn't feed forward, then what is? Malcom Hawksford's circuit may be another form of "error correction", although I don't fully understand Malcom's circuit which relies on positive feedback for its operation, so I am not positive it qualifies. It is clear that your circuit is nothing more than ordinary NFB disguised with some extra circuit complexity and a lot of fancy words to bamboozle the unwary. I have made it extraordinarly clear how my ideas work. Mr Hawksford is another bamboozler of the first order. The reason you can't understand his ideas is that perhaps you are a bit slow, but mainly because Hawksford can't be bothered to spell things out to ordinary mortals; he can only be understood by those on a similar level of intellectual capacity, due to their math ability, and qualifications gained at universities. Actually I think I pretty much understand his idea as far as he went in the article, the problem I had is that he didn't discuss the conditions for stability in his circuit, beyond simply mentioning that they exist. We all flounder in the presence of such fellows, who have taken electronics far beyond the cutting edge of audio amplifiers as it was in 1955, when a good Williamson was king, and nothing else mattered. But a decent Williamson still sounds well compared to anything Hawksford may have designed. It also measures well enough. Putting more zeros in the THD results does not greatly improve the sound. Hawksford has written on numerous other aspects of audio, both in the analog domain, and even more perplexingly, in the digital domain, and I have no intention of following what this brilliant long haired git has to say because its all too hard for me to decypher it all. He has no interest in explaining it in a way which allows us to apply his ideas easily, and he'd probably think we were ****ing into the wind by experimenting with novel or at least different tube circuit ideas. And his ideas may not be easily applied, even if expresed simply enough. Much modern electronics is like that. Speak for yourself, I don't think the math has changed significantly since 1955, only the active devices have changed. You are simply putting Malcom down because you can't even begin to grasp what he is saying. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
|
#22
|
|||
|
|||
|
John Stewart wrote: I took a few days to examine the graphics of the outside world here. Not too suprisingly, it sure beats what is on the iNet. The bicycle got a real workout, 40 to 50 miles per day. That used to be easy. It isn't anymore!!! Your comment regarding cell phone amps sound like an awful solution to try to get hi-fi performance, but I doubt there is the slightest simularity in what John Stewart's "real error correction" actually does, if it can actually be determined if it does anything at all. After examining the circuit which turned up at the usenet replayer archives site today, I found it very hard to follow because the individual working signal voltages are not indicated, with their relevant phases, which is very important for those even with a keen eye who can fill in most of the missing details that should be there. Settle down now Patrick. I would have thought a guy of your claimed technical caliber could have sorted out that circuit in a millisecond or less. In any case, I believe the simple description I gave in the original post sez it all. Why get complicated? One doesn't need a lot of calcs & figures to get the gist of that circuit. And as I said, it is not optomized, so one would need to do a bit work to get it running. Sorry John, but your presentation of your idea was quite incompetent, and so could you fill in the missing details as requested. If you don't, don't be surprised if folks all think your idea can't work. But I don't think anyone else tried to see if your idea could work at all. We are not mind readers, and its not up to us to work out all the operating voltages and transformer ratios. Mind you, I was the only soul to give your idea any serious criticism, so let's change that to I am not a mind reader. It does fulfill all of the points made in a paper on the subject by Peter J Walker. Your schematic has no operation principles like Walker proposed. Only difference is that the correction is applied in series with the main signal rather than parallel at the output. I had not read the Walker paper, otherwise I might have tried the parallel route. Rubbish. There is nothing unusual with the PP power amp which has a 12AT7 LTP and pair of 2A3 OPVs transformer coupled to the output. That is why I used this simple topology in the example I posted. Would the circuit have been anymore understandable had I used 300B's or 304TH's? The power amp could have been any darn thing. Could have been a feedback amp, with the error amp tacked on in series. Cheers to all, John Stewart Hope we are all learning a bit here!!! Er, no, not yet. Patrick Turner. |
|
#23
|
|||
|
|||
|
Jim Thompson wrote: On Sat, 25 Sep 2004 19:45:52 +0100, John Woodgate wrote: I read in alt.binaries.schematics.electronic that Charles W. Johson Jr. wrote (in .earthlink.net) about 'Real Error Correction', on Sat, 25 Sep 2004: Jim Thompson wrote, but I didn't get the article: Perhaps we're not expert in acoustics and tuned boxes, but we don't claim to be. Audiphools act as if they know something about electronics when they actually know ZERO... to wit, the on-going "error correction" thread... wildest malarkey I've seen in years Error correction in audio was first proposed, I think, by E C Cherry (if Paul Voight didn't think of it first, as he did for a vast range of things) and has subsequently been studied by Malcolm Hawksford. Neither of them is in any way an audiophool. ... so bad it makes Paul Burridge appear educated ;-) Nothing, but nothing, could do that. I am sure that Cherry's discussions are at least several orders of magnitude more erudite than the circular reasoning of Patrick turner ;-) Cherry ain't a professor for nothing, and he's got the gift of the gab alright, and can bamboozle folks for hours on end with glib talk about nested feedback loops. I've got his disertations, and a couple of schematics of his. I tried his ideas, and got a lot of oscillations at HF. I then went back to rely on my own "easier to get right" designs, and built a 300 watt/ch amp with 0.005% thd at 298 watts, a nice simple class AB mosfet amp which has diminishing thd at the level is reduced, showing that its possible to get the crossover distortion to dissapear into the noise, like all well designed solid state, just relying on nfb, and enough class A to ensure there is output device gain regardless of bias drift. It sounds no better than a tube amp. Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. |
|
#24
|
|||
|
|||
|
In article , Patrick Turner
wrote: John Stewart wrote: Settle down now Patrick. I would have thought a guy of your claimed technical caliber could have sorted out that circuit in a millisecond or less. In any case, I believe the simple description I gave in the original post sez it all. Why get complicated? One doesn't need a lot of calcs & figures to get the gist of that circuit. And as I said, it is not optomized, so one would need to do a bit work to get it running. Sorry John, but your presentation of your idea was quite incompetent, and so could you fill in the missing details as requested. If you don't, don't be surprised if folks all think your idea can't work. But I don't think anyone else tried to see if your idea could work at all. You are wrong, I gave it the once over, it looked reasonable to me, given that it was only presented as a concept so far. The only complains I had with the presentation were that he forgot to explicitly indicate the output terminals, and the purpose of the redundant 5k resistor across the output of the main amplifier was not obvious, other than that it is straight forward enough. We are not mind readers, and its not up to us to work out all the operating voltages and transformer ratios. I'm sure when he has worked out all those details and built one, we will read about it in that magazine he writes for, whose name I forget. Mind you, I was the only soul to give your idea any serious criticism, so let's change that to I am not a mind reader. There is nothing to criticize about the basic idea which is sound, perhaps there will be reason to criticize the actual implementation, only time will tell about that. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
|
#25
|
|||
|
|||
|
On Sun, 26 Sep 2004 12:14:56 +1000, Patrick Turner
wrote: Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John |
|
#26
|
|||
|
|||
|
John Byrns wrote: In article , Patrick Turner wrote: John Byrns wrote: Patrick, you seem to be flip-flopping here, but I suppose the "if not all" part of the "null out most if not all the wanted signal" is your out. In your first explanation of "error correction" the main amplifier had a differential input, and the input signal was feed to the non-inverting input, while the signal at the junction of the two resistors you describe above was feed to the inverting input. You made a point at that time that the wanted signal at the junction of the two resistors was not nulled out, but that the resistors were adjusted so that the wanted signal at the inverting input was equal in amplitude, and opposite in phase, to the input signal at the non-inverting input. The option of using the error amplifier as a phase inverter is there if you want it to. What "error amplifier"? In your original design there were only two amplifiers, the "main" amplifier with a differential input, and the "auxiliary phase inverter", which is the "error amplifier"? The main amp has a diff input. The auxiliary amp is the error amp. It also acts as a phase inverter, depending on values of the R taken from its anode circuit to the output. I have already explained that. Did you forget that you read and understood what I wrote? If all the wanted signal is nulled by the two R network, then its only truly possible at one load value, and one F, due to phase shift. But at all different load values and F there will be some signal at the null point, and this **small** signal will form what is the correction signal along with the distortion signal and applied back into the amp and do its job just like NFB will. Indeed, it is "just like NFB", precisely because it is NFB, all dressed up with a fancy new name. Your flip-flop on that is really of no consequence, but lets consider the circuit a little further. Implicit in your description above is that the input signal is feed into the amplifier at some unspecified point. There are several ways of applying the error signal. Let's call the junction of the two resistors you mention the "summing junction". Would it not be legitimate to feed the input signal through a third resistor, in the appropriate phase, into this same "summing junction at the inverting input? One would need more hardware than I have proposed, so why do it your way? Post a schematic, with a full explantion please. How do you figure more hardware is required? It appears to me that there is no significant difference in the amount of hardware required. You can find the schematic for my design at this URL: http://users.rcn.com/jbyrns/stuff/NFBamp.jpg This schematic above is a variation of my error correction so that it becomes a shunt FB design. It is a genuine NFB amp. My original could be arranged so V1 only supplies an error signal to one of the two diff inputs and so its error correction amp. The outcome from the method I originally proposed and you addaption of my schematic is exactly the same, except that my version provides better sensitivity. As you can see the same amount of hardware is required either way. I have omitted the details of the V1 stage in my design because I haven't yet been able to find a tube for V1 that will provide the same 1.7 volt input sensitivity as with your design. A 12AT7 would be fine for the diff amp, V2 & V3. The AT7 will have about twice the gain of the 6SN7. If a dual 6SN7 stage is used as in your design, then the input sensitivity decreases to about 1.96 volts from 1.7 volts in your amplifier. On the other hand there are several tubes that could be used for V1 which approximately double the sensitivity of my design over yours, but I can't find any tubes that give your original input sensitivity. You could use a variety of tubes for V1. A pentode would be fine, except that the CCS load for V1 is necessary to make the effective R1 of the R1 - R2 nulling shunt FB nework to be as large as possible. The action of the error correction circuits I have proposed is the **same** as series voltage negative feedback, OK, but I have not fed back a signal nearly as big as input signal to a diff input stage. You can't have it both ways, it's either the same or it's not. The only difference between the different topologies using a differential input is the amount of common mode signal applied to the input of the differential amplifier stage. There is a subtle difference and you just wanna just dumb down what I have proposed. You still haven't explained the supposed "subtle difference", most likely because there isn't any difference at all, subtle or not. I can't call it negative feedback, because its not the same. Sure you can, just repeat after me, "it's just ordinary negative feedback." Now that wasn't so hard was it? I don't like beligerant interogations. Your'e like Stalin telling someone in Russia in 1933, "now this is the truth, ok-etsky!!!"... There is nothing like it in RDH4. I suggest that you carefully read chapter 7 of the RDH4, and then look at your circuit again. There is nothing like my circuit in chapter 7 which uses an input tube to null the wanted signal, leaving only the error signal to be applied to one of the diffamp inputs. There is reference on pages 333,334 to shunt NFB. In 1996, I tested the schematic I designed shown at http://www.turneraudio.com.au/htmlwe...basicbshfb.htm There is no circuit in RDH4 which resembles this one either. But its still basic shunt FB, only balanced. The only down side of this circuit is that the OPT secondary isn't included in the loop, so the winding resistance of the OPT is not reduced by the NFB. But otherwise it worked splendidly, with fine HF performance. So I have to pick a name, so error correction is the category I choose. You only "have to pick a name" because you are imagining a difference that isn't there. As my wife would say, "you have cobwebs over your eyes". I know you do as well as she does. I would consider feed forward, as used in the circuit proposed by JS to be a true "error correction" topology. JS's idea isn't feed forward. He extracts the distortion of an amp1, amplifies it in amp2, and applies it in series with the signal from amp1, so you have two amps both with near as possible equal distortion voltages appearing at their outputs, so there is no distortion voltage across the load. no distortion current either. Output resistance is kept to as near zero ohms as possible. If that isn't feed forward, then what is? Feed beside, as one respondent put it. But nothing is fed forward. Malcom Hawksford's circuit may be another form of "error correction", although I don't fully understand Malcom's circuit which relies on positive feedback for its operation, so I am not positive it qualifies. It is clear that your circuit is nothing more than ordinary NFB disguised with some extra circuit complexity and a lot of fancy words to bamboozle the unwary. I have made it extraordinarly clear how my ideas work. Mr Hawksford is another bamboozler of the first order. The reason you can't understand his ideas is that perhaps you are a bit slow, but mainly because Hawksford can't be bothered to spell things out to ordinary mortals; he can only be understood by those on a similar level of intellectual capacity, due to their math ability, and qualifications gained at universities. Actually I think I pretty much understand his idea as far as he went in the article, the problem I had is that he didn't discuss the conditions for stability in his circuit, beyond simply mentioning that they exist. You don't understand the contents of his papers on the EC matters, which are posted on the web. You have trouble with all the simple concepts and schematics I post, even when I provide all the details. Please don't make an idiot out of yourself by insisting you understand Hawksford! We all flounder in the presence of such fellows, who have taken electronics far beyond the cutting edge of audio amplifiers as it was in 1955, when a good Williamson was king, and nothing else mattered. But a decent Williamson still sounds well compared to anything Hawksford may have designed. It also measures well enough. Putting more zeros in the THD results does not greatly improve the sound. Hawksford has written on numerous other aspects of audio, both in the analog domain, and even more perplexingly, in the digital domain, and I have no intention of following what this brilliant long haired git has to say because its all too hard for me to decypher it all. He has no interest in explaining it in a way which allows us to apply his ideas easily, and he'd probably think we were ****ing into the wind by experimenting with novel or at least different tube circuit ideas. And his ideas may not be easily applied, even if expresed simply enough. Much modern electronics is like that. Speak for yourself, I don't think the math has changed significantly since 1955, only the active devices have changed. C'mon, much has changed since 1955!!!!!!!!! Drag yourself out of the fog of the past, please! Try living in the present, and realise that there has been a lot of progress. You are simply putting Malcom down because you can't even begin to grasp what he is saying. I am cranky with the likes of these boffins with their theories, not because they are wrong, ( I cannot prove they are wrong, ok,) because they keep their secrets locked up amoungst their own clan of electronical high priests, by using terminology which few of us, most certainly including yourself, cannot understand. If we try out their ideas in circuits without full understanding, we will never achieve what the wanted outcome would be, ie, a near distortionless amplifier. Let me know when you have cobbled up a solid state amp with Hawksford's ideas, and got it working properly. I ain't gonna hold my breath waiting. To my mind, distortionless amps are not worth trying to build, since so much else tugs at fidelity. An SE amp with 0.05% thd with 15 dB of NFB at 5 watts is a fine thing to listen to music through. And often sounds better than any SS amp even if it measures at 0.0005%. The objectevists will condemn me for saying that, but enough people line up at my front door to tell me how it is, and they scream for tube amps, not SS amps, or else I'd be building them. Patrick Turner. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
|
#27
|
|||
|
|||
|
John Larkin wrote: On Sun, 26 Sep 2004 12:14:56 +1000, Patrick Turner wrote: Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John The Halcro amps use mosfets in their output stages. They claim to achieve 0.0001% thd at 200 watts at any F up to 20 kHz. All other parameters are claimed to be excellent. Don't ask me how they do it, and they try desperately to keep it a secret, lest the clone merchants move in. One always has the choice to not use a single bjt in any audio amps. There will always be the pro bjt brigade, I raise the hat as they pass me by... Patrick Turner. |
|
#28
|
|||
|
|||
|
On Sat, 25 Sep 2004 19:59:30 -0700, John Larkin
wrote: On Sun, 26 Sep 2004 12:14:56 +1000, Patrick Turner wrote: Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John John, You are incorrect ;-) ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. |
|
#29
|
|||
|
|||
|
In article , Patrick Turner
wrote: John Byrns wrote: In article , Patrick Turner wrote: John Byrns wrote: The option of using the error amplifier as a phase inverter is there if you want it to. What "error amplifier"? In your original design there were only two amplifiers, the "main" amplifier with a differential input, and the "auxiliary phase inverter", which is the "error amplifier"? The main amp has a diff input. Yes, that is true. The auxiliary amp is the error amp. The "auxiliary amp" doesn't amplify the "error" signal, only the input signal is amplified, the only thing the "auxiliary amp" has to do with the "error" signal" is that the "error" signal" is created in the plate circuit of "auxiliary amp" by summing the feedback signal from the output of the main amplifier with the output of the "auxiliary amp" in the plate circuit. That hardly makes the "auxiliary amp" an "error amp", an "error summer" maybe. It also acts as a phase inverter, depending on values of the R taken from its anode circuit to the output. Definitely a phase inverter, independent of what the R value from the output to its anode is. I have already explained that. Did you forget that you read and understood what I wrote? Clearly not. One would need more hardware than I have proposed, so why do it your way? Post a schematic, with a full explantion please. How do you figure more hardware is required? It appears to me that there is no significant difference in the amount of hardware required. You can find the schematic for my design at this URL: http://users.rcn.com/jbyrns/stuff/NFBamp.jpg This schematic above is a variation of my error correction so that it becomes a shunt FB design. It is a genuine NFB amp. My original could be arranged so V1 only supplies an error signal to one of the two diff inputs and so its error correction amp. The outcome from the method I originally proposed and you addaption of my schematic is exactly the same, except that my version provides better sensitivity. Just a minute here, if I am reading your statement above correctly, you are now admitting that your "error correction" amplifier circuit is nothing more than a plain old NFB amplifier circuit! You state clearly that my adaptation of your circuit "is a genuine NFB amp" and then you go on to state that the "method I (you) originally proposed and you (my) addaption of my (your) schematic is exactly the same, except that my (your) version provides better sensitivity." Isn't that the same thing as saying that your circuit is also just a plain old "NFB amp" with slightly greater sensitivity, i.e. 1.7 v vs. 1.96 v? As you can see the same amount of hardware is required either way. I have omitted the details of the V1 stage in my design because I haven't yet been able to find a tube for V1 that will provide the same 1.7 volt input sensitivity as with your design. A 12AT7 would be fine for the diff amp, V2 & V3. The AT7 will have about twice the gain of the 6SN7. I wanted to keep the main amplifier circuit the same as you had so that the amount of NFB remained unchanged. V1 is outside the feedback loop and so is fair game for change as long as Ro remains 47k. If a dual 6SN7 stage is used as in your design, then the input sensitivity decreases to about 1.96 volts from 1.7 volts in your amplifier. On the other hand there are several tubes that could be used for V1 which approximately double the sensitivity of my design over yours, but I can't find any tubes that give your original input sensitivity. You could use a variety of tubes for V1. Yes, but I haven't been able to find a suitable tube that doesn't increase the gain considerably, only a small increase in gain is called for, and the gain can't be decreased by increasing the cathode degeneration because that would also change Ro. Perhaps a combination of cathode feedback and "shunt" feedback around V1 would do the trick with a higher gain tube. There is nothing like it in RDH4. I suggest that you carefully read chapter 7 of the RDH4, and then look at your circuit again. There is nothing like my circuit in chapter 7 which uses an input tube to null the wanted signal, leaving only the error signal to be applied to one of the diffamp inputs. Changing the precise details of how the summing is done does not change the fact that it is still nothing more than a NFB amplifier. I would consider feed forward, as used in the circuit proposed by JS to be a true "error correction" topology. JS's idea isn't feed forward. He extracts the distortion of an amp1, amplifies it in amp2, and applies it in series with the signal from amp1, so you have two amps both with near as possible equal distortion voltages appearing at their outputs, so there is no distortion voltage across the load. no distortion current either. Output resistance is kept to as near zero ohms as possible. If that isn't feed forward, then what is? Feed beside, as one respondent put it. But nothing is fed forward. That respondent was either making a joke about how the schematic was drawn, or he doesn't have a clue about what feed forward is. In a feed forward amplifier all signal flows are in the down stream direction, or down stream direction, there is no upstream signal flow as in an ordinary feedback amplifier. JS's amplifier meets this criteria if you ignore the local feedback loop JS used around his error amplifier. If you draw JS's amplifier in block diagram form, with the two amplifiers, including any local feedback, shown as blocks one following the other, then you can clearly see that the signal flow is only in the forward or downstream direction. What is your definition of feed forward? Actually I think I pretty much understand his idea as far as he went in the article, the problem I had is that he didn't discuss the conditions for stability in his circuit, beyond simply mentioning that they exist. You don't understand the contents of his papers on the EC matters, which are posted on the web. It doesn't really matter one way of the other, Hawksford's EC circuit is not relevant to this discussion, or are you now claiming that your amplifier uses Hawksford's topology? You have trouble with all the simple concepts and schematics I post, even when I provide all the details. I understand your schematic perfectly well, the only problem I have is with your claim that it is somehow fundamentally different than an ordinary NFB amplifier. Please don't make an idiot out of yourself by insisting you understand Hawksford! OK, as I said Hawksford's circuit isn't relevant to this discussion anyway. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
|
#30
|
|||
|
|||
|
John Byrns wrote: In article , Patrick Turner wrote: John Byrns wrote: In article , Patrick Turner wrote: John Byrns wrote: The option of using the error amplifier as a phase inverter is there if you want it to. What "error amplifier"? In your original design there were only two amplifiers, the "main" amplifier with a differential input, and the "auxiliary phase inverter", which is the "error amplifier"? The main amp has a diff input. Yes, that is true. The auxiliary amp is the error amp. The "auxiliary amp" doesn't amplify the "error" signal, only the input signal is amplified, the only thing the "auxiliary amp" has to do with the "error" signal" is that the "error" signal" is created in the plate circuit of "auxiliary amp" by summing the feedback signal from the output of the main amplifier with the output of the "auxiliary amp" in the plate circuit. That hardly makes the "auxiliary amp" an "error amp", an "error summer" maybe. It also acts as a phase inverter, depending on values of the R taken from its anode circuit to the output. Definitely a phase inverter, independent of what the R value from the output to its anode is. I have already explained that. Did you forget that you read and understood what I wrote? Clearly not. One would need more hardware than I have proposed, so why do it your way? Post a schematic, with a full explantion please. How do you figure more hardware is required? It appears to me that there is no significant difference in the amount of hardware required. You can find the schematic for my design at this URL: http://users.rcn.com/jbyrns/stuff/NFBamp.jpg This schematic above is a variation of my error correction so that it becomes a shunt FB design. It is a genuine NFB amp. My original could be arranged so V1 only supplies an error signal to one of the two diff inputs and so its error correction amp. The outcome from the method I originally proposed and you addaption of my schematic is exactly the same, except that my version provides better sensitivity. Just a minute here, if I am reading your statement above correctly, you are now admitting that your "error correction" amplifier circuit is nothing more than a plain old NFB amplifier circuit! No. YOU have placed a revised circuit of mine in that URL of yours, and that is YOUR idea to do so, and its a shunt FB amp. My original had the V1 acting as the R1 arm of a two R signal nulling network. The use of a pentode instead of a triode at V1 and with a CCS load allows for virtually all the distortion signal to appear at the V1 anode, from where it can be applied to one of the two diff inputs. V1 is acting as a current amplifier, not a voltage amplifier, although in my original circuit I had it doing the duty of a phase inverter, which confused you rather badly. Where you have a current amp which opposes a flow of current, it is still an amplifier, even though no voltage gain is seen. You state clearly that my adaptation of your circuit "is a genuine NFB amp" and then you go on to state that the "method I (you) originally proposed and you (my) addaption of my (your) schematic is exactly the same, except that my (your) version provides better sensitivity." Isn't that the same thing as saying that your circuit is also just a plain old "NFB amp" with slightly greater sensitivity, i.e. 1.7 v vs. 1.96 v? Look, I will repeat again that where *all* the input signal goes to one port of the diff pair, and the nulled distortion only goes to the other port, then there is no fed back signal, just the distortion. If greater error correction is wanted, the distortion can be amplified to whatever level is required, stability permitting, and then applied to the distortion port. Smells like error correction to me, quacks like EC, feathers like EC, but the effect is similar or the same as NFB. As you can see the same amount of hardware is required either way. I have omitted the details of the V1 stage in my design because I haven't yet been able to find a tube for V1 that will provide the same 1.7 volt input sensitivity as with your design. A 12AT7 would be fine for the diff amp, V2 & V3. The AT7 will have about twice the gain of the 6SN7. I wanted to keep the main amplifier circuit the same as you had so that the amount of NFB remained unchanged. V1 is outside the feedback loop and so is fair game for change as long as Ro remains 47k. Indeed V1 is outside the loop in my design. But in your version of my design, the V1 is in the loop more, but the fed back signal current via the 22k R opposes the current produced in V1, which is shunt negative voltage FB. Normally, the NFB signal is applied to the cathode of V1 to make it act as a diff amp, and this is series negative FB. Shunt FB is rarely if ever used in tube power amps. In order to make it work best, V1 should provide linear current amplification, hence the Rk of V1 needs careful choice to reduce the V1 gain. So with such low gain, the HF performance will be good, and miller effect on the input signal will be low. If a dual 6SN7 stage is used as in your design, then the input sensitivity decreases to about 1.96 volts from 1.7 volts in your amplifier. On the other hand there are several tubes that could be used for V1 which approximately double the sensitivity of my design over yours, but I can't find any tubes that give your original input sensitivity. You could use a variety of tubes for V1. Yes, but I haven't been able to find a suitable tube that doesn't increase the gain considerably, only a small increase in gain is called for, and the gain can't be decreased by increasing the cathode degeneration because that would also change Ro. Perhaps a combination of cathode feedback and "shunt" feedback around V1 would do the trick with a higher gain tube. I have arranged it so the distortion reduction factor is still 1 / ( 1 + [ A x B ] , where A is the open loop gain between diff input to V2/3 and the output, and B = Dn' / Dn where Dn is the distortion at the output and Dn' is the applied distortion at the anode of V1. There is nothing like it in RDH4. I suggest that you carefully read chapter 7 of the RDH4, and then look at your circuit again. There is nothing like my circuit in chapter 7 which uses an input tube to null the wanted signal, leaving only the error signal to be applied to one of the diffamp inputs. Changing the precise details of how the summing is done does not change the fact that it is still nothing more than a NFB amplifier. I would consider feed forward, as used in the circuit proposed by JS to be a true "error correction" topology. JS's idea isn't feed forward. He extracts the distortion of an amp1, amplifies it in amp2, and applies it in series with the signal from amp1, so you have two amps both with near as possible equal distortion voltages appearing at their outputs, so there is no distortion voltage across the load. no distortion current either. Output resistance is kept to as near zero ohms as possible. If that isn't feed forward, then what is? Feed beside, as one respondent put it. But nothing is fed forward. That respondent was either making a joke about how the schematic was drawn, or he doesn't have a clue about what feed forward is. In a feed forward amplifier all signal flows are in the down stream direction, or down stream direction, there is no upstream signal flow as in an ordinary feedback amplifier. JS's amplifier meets this criteria if you ignore the local feedback loop JS used around his error amplifier. If you draw JS's amplifier in block diagram form, with the two amplifiers, including any local feedback, shown as blocks one following the other, then you can clearly see that the signal flow is only in the forward or downstream direction. What is your definition of feed forward? Its where you feed a correction signal directly into the input flow. Depending on the impedances involved, its quite easy to arrange. Nobody understands it, and it all requires more glassware, so it lies ignored, as a useless technique. Actually I think I pretty much understand his idea as far as he went in the article, the problem I had is that he didn't discuss the conditions for stability in his circuit, beyond simply mentioning that they exist. You don't understand the contents of his papers on the EC matters, which are posted on the web. It doesn't really matter one way of the other, Hawksford's EC circuit is not relevant to this discussion, or are you now claiming that your amplifier uses Hawksford's topology? No. You have trouble with all the simple concepts and schematics I post, even when I provide all the details. I understand your schematic perfectly well, the only problem I have is with your claim that it is somehow fundamentally different than an ordinary NFB amplifier. The original circuit of mine *is* different to a NFB amp, but the outcome is similar. Isn't that a simple explanation? If you are always going to nit pick on the words used to describe something, and never see the operation of a circuit as outside the familiar square you are used to, you will always have trouble with anything different to the status quo, and end up always in a turgid flatus flow discussion. Please don't make an idiot out of yourself by insisting you understand Hawksford! OK, as I said Hawksford's circuit isn't relevant to this discussion anyway. Which exact Hawksford circuit are you refering to? There are rather a lot, and rather difficult to follow. Patrick Turner. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
|
#31
|
|||
|
|||
|
On Sun, 26 Sep 2004 08:02:30 -0700, Jim Thompson
wrote: On Sat, 25 Sep 2004 19:59:30 -0700, John Larkin wrote: On Sun, 26 Sep 2004 12:14:56 +1000, Patrick Turner wrote: Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John John, You are incorrect ;-) ...Jim Thompson About what the audiofolks prefer, or about the tekkie stuff? John |
|
#32
|
|||
|
|||
|
On Sun, 26 Sep 2004 15:17:00 -0700, John Larkin
wrote: On Sun, 26 Sep 2004 08:02:30 -0700, Jim Thompson wrote: On Sat, 25 Sep 2004 19:59:30 -0700, John Larkin wrote: On Sun, 26 Sep 2004 12:14:56 +1000, Patrick Turner wrote: Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John John, You are incorrect ;-) ...Jim Thompson About what the audiofolks prefer, or about the tekkie stuff? John Audio is analog. Fets are NOT easier to bias, and are not more rugged, nor better at slew-rate. Where did you get that BIAS? ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. |
|
#33
|
|||
|
|||
|
On Sun, 26 Sep 2004 16:14:46 -0700, Jim Thompson
wrote: On Sun, 26 Sep 2004 15:17:00 -0700, John Larkin wrote: On Sun, 26 Sep 2004 08:02:30 -0700, Jim Thompson wrote: On Sat, 25 Sep 2004 19:59:30 -0700, John Larkin m wrote: On Sun, 26 Sep 2004 12:14:56 +1000, Patrick Turner wrote: Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John John, You are incorrect ;-) ...Jim Thompson About what the audiofolks prefer, or about the tekkie stuff? John Audio is analog. Fets are NOT easier to bias, and are not more rugged, nor better at slew-rate. Where did you get that BIAS? ...Jim Thompson No second breakdown. No thermal runaway. No Ft falloff (no Ft at all, actually). No big peak base currents (no base at all, actually). No nasty nonlinear b-e junction. No pig-slow minority carriers. No storage. I will try to get over my CURRENT annoyance at your BASE insults. John |
|
#34
|
|||
|
|||
|
Audio is analog. Fets are NOT easier to bias, and are not more rugged, nor better at slew-rate. Where did you get that BIAS? ...Jim Thompson No second breakdown. No thermal runaway. No Ft falloff (no Ft at all, actually). No big peak base currents (no base at all, actually). No nasty nonlinear b-e junction. No pig-slow minority carriers. No storage. I will try to get over my CURRENT annoyance at your BASE insults. John You need to go to rec.audio.solid.state to argue all such boring arguments we have all been bored with before. The guys at Halcro must be really worried, Jim has condemned mosfets and worldwide sales of this premier brand of solid state amp are now rapidly declining after what he said. Patrick Turner. |
|
#35
|
|||
|
|||
|
On Mon, 27 Sep 2004 12:01:16 +1000, Patrick Turner
wrote: Audio is analog. Fets are NOT easier to bias, and are not more rugged, nor better at slew-rate. Where did you get that BIAS? ...Jim Thompson No second breakdown. No thermal runaway. No Ft falloff (no Ft at all, actually). No big peak base currents (no base at all, actually). No nasty nonlinear b-e junction. No pig-slow minority carriers. No storage. I will try to get over my CURRENT annoyance at your BASE insults. John You need to go to rec.audio.solid.state to argue all such boring arguments we have all been bored with before. The guys at Halcro must be really worried, Jim has condemned mosfets and worldwide sales of this premier brand of solid state amp are now rapidly declining after what he said. Patrick Turner. Audio is indeed boring. The most I expect here is a decent pun thread. John |
|
#36
|
|||
|
|||
|
"John Larkin" wrote in message ... On Sun, 26 Sep 2004 16:14:46 -0700, Jim Thompson wrote: On Sun, 26 Sep 2004 15:17:00 -0700, John Larkin wrote: On Sun, 26 Sep 2004 08:02:30 -0700, Jim Thompson wrote: On Sat, 25 Sep 2004 19:59:30 -0700, John Larkin m wrote: On Sun, 26 Sep 2004 12:14:56 +1000, Patrick Turner wrote: Some say it sounds worse, but at least its a shirtload better than many SS amps using bjts that I have tried. Patrick Turner. I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John John, You are incorrect ;-) ...Jim Thompson About what the audiofolks prefer, or about the tekkie stuff? John Audio is analog. Fets are NOT easier to bias, and are not more rugged, nor better at slew-rate. Where did you get that BIAS? ...Jim Thompson No second breakdown. No thermal runaway. No Ft falloff (no Ft at all, actually). No big peak base currents (no base at all, actually). No nasty nonlinear b-e junction. No pig-slow minority carriers. No storage. But looking at it from the days of germaniums, the silicon BJTs are mile ahead of the old timers. And going back further, well, like someone else said, you can't do things like complementary output with toobs. And then there was A.G. Bell, who tried to make something that would handle analog speech by using a buzzer or interruptor. BJTs look positively amazing compared to that. So I can live with those small quirks that you point out above. And if you can't live with that, then keep away from the linear region, and make a Class D amp. I will try to get over my CURRENT annoyance at your BASE insults. John |
|
#37
|
|||
|
|||
|
On Sun, 26 Sep 2004 20:29:06 -0700, "Watson A.Name - \"Watt Sun, the
Dark Remover\"" wrote: No second breakdown. No thermal runaway. No Ft falloff (no Ft at all, actually). No big peak base currents (no base at all, actually). No nasty nonlinear b-e junction. No pig-slow minority carriers. No storage. But looking at it from the days of germaniums, the silicon BJTs are mile ahead of the old timers. And going back further, well, like someone else said, you can't do things like complementary output with toobs. And then there was A.G. Bell, who tried to make something that would handle analog speech by using a buzzer or interruptor. BJTs look positively amazing compared to that. So I can live with those small quirks that you point out above. You can, but why? John |
|
#38
|
|||
|
|||
|
Go to ABSE for more information.
The attachments show a simulation of one of Patrick T's amplifiers with a table of test results. The top two rows in the table show the results for the amp connected as in Patrick's original circuit. When the 22K resistor which looks like NFB is returned to gnd the gain increases by about 6 db (doubles). The lower two rows of results show the amplifier connected as suggested by John B. Again the amp gain doubles when the 22K is returned to gnd. I have removed all of the circuitry which would have no bearing on the final analysis. That includes parasitic suppression resistors & the bias balance network of the output section. I've also taken the liberty of using triode connected 6550's in the output. The output transformer used has a 20:1 turns ratio so the impedance reflected back to the 6550's is 4K, as recommended in the data sheets. I realize that most of Pat's circuits also include cathode FB in the output after the Quad circuit but I didn't try to include that here. The results do not show me anything I could call error correction other than ordinary NFB. Both error correction & NFB depend on a comparison of the input with the output of the amplifier (or any other servo system). If used for NFB issues of phase shift create problems at the band limits. If instead the error correction information is used to make the necessary adjustments at the output then we can avoid the problems associated with phase shift. In my mind that is the fundamental difference between these pair of schemes. Cheers, John Stewart |
|
#39
|
|||
|
|||
|
"John Larkin" wrote in message ... On Sun, 26 Sep 2004 20:29:06 -0700, "Watson A.Name - \"Watt Sun, the Dark Remover\"" wrote: No second breakdown. No thermal runaway. No Ft falloff (no Ft at all, actually). No big peak base currents (no base at all, actually). No nasty nonlinear b-e junction. No pig-slow minority carriers. No storage. But looking at it from the days of germaniums, the silicon BJTs are mile ahead of the old timers. And going back further, well, like someone else said, you can't do things like complementary output with toobs. And then there was A.G. Bell, who tried to make something that would handle analog speech by using a buzzer or interruptor. BJTs look positively amazing compared to that. So I can live with those small quirks that you point out above. You can, but why? John There are millions of BJT power amps out there, and there aren't that many FET power amps. |
|
#40
|
|||
|
|||
|
On Sun, 26 Sep 2004 17:17:36 -0700, John Larkin
wrote: On Sun, 26 Sep 2004 16:14:46 -0700, Jim Thompson wrote: On Sun, 26 Sep 2004 15:17:00 -0700, John Larkin wrote: [snip] I get the impression that many audio folks still prefer bipolar transistors in output stages, instead of fets. I don't understand this, as fets are easier to bias, are more rugged, and are less likely (with reasonable design) to have slew limit problems. John John, You are incorrect ;-) ...Jim Thompson About what the audiofolks prefer, or about the tekkie stuff? John Audio is analog. Fets are NOT easier to bias, and are not more rugged, nor better at slew-rate. Where did you get that BIAS? ...Jim Thompson No second breakdown. I suppose you allude to BVCEO? Why are you operating an AUDIO amplifier there? No thermal runaway. No such thing in the hands of a good designer. No Ft falloff (no Ft at all, actually). Sure there is. it's just expressed as a figure of merit: gm/C No big peak base currents (no base at all, actually). Ever turned a big-assed FET on while observing gate current ?:-) I can recall AMPS of gate current is a satellite spin-up motor controller I worked on. No nasty nonlinear b-e junction. Isn't the cut-off-to-ON transition a non-linearity? No pig-slow minority carriers. No storage. What are you doing saturating a transistor in an audio amplifier? As switches, FETs excel; but they are a real pain-in-the-ass to use as audio outputs. I will try to get over my CURRENT annoyance at your BASE insults. John Sno-o-o-ort! ROTFLMAO! ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice 480)460-2350 | || E-mail Address at Website Fax 480)460-2142 | Brass Rat || http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food. |
| Reply |
|
| Thread Tools | |
| Display Modes | |
|
|
Similar Threads
|
||||
| Thread | Forum | |||
| Real Stereophonic Recordings | Pro Audio | |||
| Record Real Media Stream | Pro Audio | |||
| Digidesign to release new HD Plugin - REAL OXIDE | Pro Audio | |||
| real audio settings | General | |||
| Real Audio Stream to Files !! | General | |||
Linear Mode
