Interesting....I tried recording some LP's onto CD a few weeks ago with poor results. Sure it sounded "like LP" in as much as it was more rounded than CD. It was in fact, at least in terms of frequency balance, not much different to the original. In terms of depth, transparancy and space around instruments however, it was a travesty of the original.....however this was probably due to the quality of the op amps and ADC in the recorder...or was it? The debate rages on 
A more relavent (which this post is anything but at this rate
) is a comparison between digital and 1/2 track 15ips open reel. I have tried recording the same live feed from quality condensor mikes onto both and can report that open reel won...by a large margin. I would not call 1/2 track at 15ips treble shy but yet it still sounds "analogue" (well ya know what I mean). The subjective V the objective still confounds us all, that's what makes audio engineering so interesting and at the same time frustrating IMHO 
Atenuation of RF at the output inductor may be in most cases virtually non existant as it is usual practice to put a resistor of one to ten Ohms in parallel with it to damp out ringing in said inductor.....and as Janneman said, the output impedance is likely to be quite high (and unpredictable to some extent) at RF frequencies anyway.
Inductors /LPF in the feedback path is likely to make achieving loop stability fun.....not!
There could well be something important going on in all this though. Low NFB = low loop gain=less opportunity for the RF nasties to do any damage...maybe? It's never boring this hifi lark 'eh
A more relavent (which this post is anything but at this rate
) is a comparison between digital and 1/2 track 15ips open reel. I have tried recording the same live feed from quality condensor mikes onto both and can report that open reel won...by a large margin. I would not call 1/2 track at 15ips treble shy but yet it still sounds "analogue" (well ya know what I mean). The subjective V the objective still confounds us all, that's what makes audio engineering so interesting and at the same time frustrating IMHO Atenuation of RF at the output inductor may be in most cases virtually non existant as it is usual practice to put a resistor of one to ten Ohms in parallel with it to damp out ringing in said inductor.....and as Janneman said, the output impedance is likely to be quite high (and unpredictable to some extent) at RF frequencies anyway.
Inductors /LPF in the feedback path is likely to make achieving loop stability fun.....not!
There could well be something important going on in all this though. Low NFB = low loop gain=less opportunity for the RF nasties to do any damage...maybe? It's never boring this hifi lark 'eh
In one of the Cherry's amps, the feedback point is taken at the node between the resistor and the grounded capacitor of the ouput RC Zobel network.
DARKFENRIZ
To me, Ottala's TID has something related to saturation of an internal stage of an amp. So how can it be said it's a distorsion ? During a saturation, there is no information passing at all.
Jung's SID is a real distorsion mechanism, related to an increasing demand of current feeding a compensation capacitor when frequency increases. It was well known before Ottala. I do not think amplifiers sound very differently before and after the publication of Ottala's papers. However, it is true we may feel more confortable having faster amps and op-amps which apparition was not accompanied by bad side-effects.
JANNEMAN
---Possibly we blamed the CD for the wrong things.---
Possibly high feedback is also blamed for the wrong things.
DARKFENRIZ
To me, Ottala's TID has something related to saturation of an internal stage of an amp. So how can it be said it's a distorsion ? During a saturation, there is no information passing at all.
Jung's SID is a real distorsion mechanism, related to an increasing demand of current feeding a compensation capacitor when frequency increases. It was well known before Ottala. I do not think amplifiers sound very differently before and after the publication of Ottala's papers. However, it is true we may feel more confortable having faster amps and op-amps which apparition was not accompanied by bad side-effects.
JANNEMAN
---Possibly we blamed the CD for the wrong things.---
Possibly high feedback is also blamed for the wrong things.
forr said:
This modification of Thiel's network, it would appear, (and Cherry contends) is necessary with the later's nested loop design.
The worthy, however, recommends that said modification be adopted universally.
The temptation to take this ''idea'' at face value must, in this context, be avoided at all costs, otherwise the RC network introduces a conspicuous pole in the major feedback loop that virtually guarantees instability.
forr said:
I think this pretty much says it all... When I first read this interview I thought it was posted as a joke; Mikeks messin with the congregation. This stuff is 25+ years old and the current comments could have been drawn from a tech article in Stereo Review. But then the discussion began. Mix in a couple of links to past threads where one notes that this has been beat to death already and I walk away shaking my head.
Since this is the internet, and this is a forum, I'll add a few interesting observations based on my experience.
Feedback is not the source of all evil. It is actually good and as Mr. B has pointed out that it's great at nailing down the circuits operating points. It is also very good at pointing out what you don't understand about your circuit.
In all of these discussions about feedback, stability, compensation poles, etc there is never any mention of the circuit layout, various impedance levels, biasing loops versus signal loops and returns, which all have a dramatic effect on not only the stability but also the low level detail and the quality of the sound.
In all of the discussion of distortions there is no attempt to understand what it actually is and why is it happening. Consider a single gain stage with no feedback, why is it creating the various orders of distortion. Where is this energy coming from and why does it chose the form that it takes? The formulas are only useful when they help one visualize what is happening.
Considering the progress that has been made over the past 30 years, sonically (that's a joke, son), I'd say building yet another amp, only this time using titanium/nichrome wire is pretty much a dead end.
My take on audio design (tough to call it engineering; that's what we do in our day jobs) is that it doesn't learn from it's experiences. It's like Bart Simpson repeatedly sticking his finger in the outlet and saying ouch each time. The quote above says it all.
This site gives off very confusing signals, one minute highly technical discussions that leave the truly interested learners wallowing in there own stupidity and the next finds it endlessly repeating old wives tales and cockeyed theories as fact. I can't say I haven't learned alot, because I have, but more of that has happened because I disagree with more of what is posted than I agree with (gets me thinking). Unfortunately there is little real discussion going on.
But that's OK (once a person dials it in) there's gems to be had in the scraps. And then there's the entertainment value.
One other observation: It's a great time to be a cutting edge designer because not only is the profit astronomical, it almost guarantees that no one will be able to critique your work (if they do you blame it on a mains power issue or a negative interaction with the other companies design). It's great work if you can get it. Life is good.
And yes I do believe there are interesting things being done it's just that I have no idea where anyones discussing it.
Regards, Mike.
RFI Ingress
I've often wondered about RFI ingress into the guts of an amplifier via the antenna effect of the speaker cable, and I do take measures in my designs to mitigate it. I must honestly say, however, that I have never tried to measure it.
For these and related reasons, I do advocate the use of the R-L network in series with the output of the amplifier to isolate the output stage and the feedback loop from RFI and HF effects at the speaker terminals. Of course, it should also have an R-C Zobel on one or both sides of it.
I believe that the omission of the L-R network by some is a misguided pursuit of sonic improvement that leaves their amplifiers vulnerable to RFI effects and possibly prone to misbehavior. Its a cruel world out there beyond the speaker terminals. Some of the possible misbehavior might not appear on the testbench.
If we are worried about RFI coming in from the speaker cables and making its way back to the input stage via the feedback loop, we should be equally concerned about possible RFI intermodulation effects directly at the output of the output stage, it would seem. Suppose we have 10 mV p-p of RFI that has made its way back to the input stage. The feeback network has typically attenuated this by a factor of about 20X (unless the designer has included a lead compensation capacitor across the feedback resistor - not a good idea in the context of this discussion). We then must have about 200 mV p-p of RFI being impressed right at the junction of the emitters (or sources) of the output stage. Given the low impedance we expect to see there, this implies some amount of RFI current. Might this not intermodulate with the Class AB crossover nonlinearity of the output stage? Just some food for thought... This would be an effect that would happen even in amplifiers with no negative feedback.
Nevertheless, getting back to the need by some to eliminate the inductor, in fairness even a conventional air core inductor can exhibit a bit of nonlinearity if it is in proximity to the wrong material. Also, amperes can be flowing in this inductor, possibly creating a magnetic field that is not welcome by other amplifier circuits. Usually any such effects can be minimized by good physical design. However, just for good measure, I do prefer to use an air core torroidal output inductor to at least try to keep the output inductor's magnetic field somewhat contained. I also like to flank the R-L network with Zobels on both sides to maintain plenty of HF damping and RFI absorption.
Bob Cordell
I've often wondered about RFI ingress into the guts of an amplifier via the antenna effect of the speaker cable, and I do take measures in my designs to mitigate it. I must honestly say, however, that I have never tried to measure it.
For these and related reasons, I do advocate the use of the R-L network in series with the output of the amplifier to isolate the output stage and the feedback loop from RFI and HF effects at the speaker terminals. Of course, it should also have an R-C Zobel on one or both sides of it.
I believe that the omission of the L-R network by some is a misguided pursuit of sonic improvement that leaves their amplifiers vulnerable to RFI effects and possibly prone to misbehavior. Its a cruel world out there beyond the speaker terminals. Some of the possible misbehavior might not appear on the testbench.
If we are worried about RFI coming in from the speaker cables and making its way back to the input stage via the feedback loop, we should be equally concerned about possible RFI intermodulation effects directly at the output of the output stage, it would seem. Suppose we have 10 mV p-p of RFI that has made its way back to the input stage. The feeback network has typically attenuated this by a factor of about 20X (unless the designer has included a lead compensation capacitor across the feedback resistor - not a good idea in the context of this discussion). We then must have about 200 mV p-p of RFI being impressed right at the junction of the emitters (or sources) of the output stage. Given the low impedance we expect to see there, this implies some amount of RFI current. Might this not intermodulate with the Class AB crossover nonlinearity of the output stage? Just some food for thought... This would be an effect that would happen even in amplifiers with no negative feedback.
Nevertheless, getting back to the need by some to eliminate the inductor, in fairness even a conventional air core inductor can exhibit a bit of nonlinearity if it is in proximity to the wrong material. Also, amperes can be flowing in this inductor, possibly creating a magnetic field that is not welcome by other amplifier circuits. Usually any such effects can be minimized by good physical design. However, just for good measure, I do prefer to use an air core torroidal output inductor to at least try to keep the output inductor's magnetic field somewhat contained. I also like to flank the R-L network with Zobels on both sides to maintain plenty of HF damping and RFI absorption.
Bob Cordell
Hi Bob,
Was your published amplifier ever commercially produced?
When your article was published, I was going to try and make one, but the circuit board work foor DIY back in those days would have been rather daunting.
Do you have any updates to it with modern semis, or a circuit board available for it? If not, how about a file that a CB could be made from?
Was your published amplifier ever commercially produced?
When your article was published, I was going to try and make one, but the circuit board work foor DIY back in those days would have been rather daunting.
Do you have any updates to it with modern semis, or a circuit board available for it? If not, how about a file that a CB could be made from?
your imagination is very plausible
Lumanauw, yes, your imagination is not just imagination, but is quite plausible.
My only caveat, and this goes for me as well, is that we often need to plug in the numbers and do the math, and take some real data, to see the degree to which our postulated effects are likely to come into play. This doesn't mean that we throw out the hypothesis, but it helps us judge the likelihood. As I mentioned in my post, I've never actually measured the amount of RF crud at the terminals of an amplifier arising from the speaker cable antenna effect, and yet I am guilty of spending a little extra money in my designs to guard against it.
Bob
Lumanauw, yes, your imagination is not just imagination, but is quite plausible.
My only caveat, and this goes for me as well, is that we often need to plug in the numbers and do the math, and take some real data, to see the degree to which our postulated effects are likely to come into play. This doesn't mean that we throw out the hypothesis, but it helps us judge the likelihood. As I mentioned in my post, I've never actually measured the amount of RF crud at the terminals of an amplifier arising from the speaker cable antenna effect, and yet I am guilty of spending a little extra money in my designs to guard against it.
Bob
MOSFET power amplifier with error correction
Pooge, the Meyer Sound MS1000 was largely based on that error correction circuit design, but was a much higher-power amplifier and employed lateral MOSFETs in the output.
More recently, some or all of the Halcro amplifiers appear to use a variation of that design, and do use vertical power MOSFETs as I did.
I'm currently working on some new variants of that design, but don't yet know what will become of it.
Bob
Pooge, the Meyer Sound MS1000 was largely based on that error correction circuit design, but was a much higher-power amplifier and employed lateral MOSFETs in the output.
More recently, some or all of the Halcro amplifiers appear to use a variation of that design, and do use vertical power MOSFETs as I did.
I'm currently working on some new variants of that design, but don't yet know what will become of it.
Bob
Re: RFI Ingress
http://www.diyaudio.com/forums/showthread.php?postid=585564#post585564
Bob Cordell said:
http://www.diyaudio.com/forums/showthread.php?postid=585564#post585564
One more Cordell-like design (hallo Bob! ) by Pavel Dudek.http://margo.student.utwente.nl/el/ampz/sch330.html
Got wonderful reviews in polish magazines.
Got wonderful reviews in polish magazines.
I suspect one reason (other than fashion) why some commercial amplifiers don't use an output inductor these days is that as more bells and whistles are squeezed into the enclosure it becomes harder to keep a suitable distance between the inductor and other elements. I learned about this the hard way in my first cookbook amplifier. Close proximity to each other can result in audible cross talk, close proximity to an input signal can result in oscillations.
Presumably the more powerful the amp the more distance is required to avoid the above. Shielding is an alternative but that take money and space as well.
Presumably the more powerful the amp the more distance is required to avoid the above. Shielding is an alternative but that take money and space as well.
darkfenriz said:
Hi Darkfenriz,
I have some difficulty reading the Pavel circuit you posted, but knowing some previous versions of this amp, this is NOT error correction as in the "Cordell amp". It is a kind of local feedback; error corection as used by Bob (and based on work from Malcolm Hawksford) uses a very specific return ratio and very specific method to feed the error signal back into the amp.
That is not to say that Pavel's amp is not a good amp; just it's not 'one more Cordell-like design'.
Jan Didden
Output inductor usually placed there because of 1 reason : to prevent instability with capacitive load.
It appears that the same thing in the same place there has other "merit" that is not intended
Power amps have LPF in the input section, trying to eliminate RF entering the non-inverting input. It's like a man going to a battle using steel armour in the chest, hoping the enemy will be attacking this chest armour from the front, actually the enemy stab from behind.
Inductor and capacitor has their non-linearities. Some designers are trying to eliminate both of these things inside power amp.
But one dumb thinking : even if the power amp is free from capacitors and inductors anywhere in the cct, as soon as it is in use, it is connected via speaker cable (which is inherent RLC) then entering passive x-over network which is full of RLC, then entering speaker coil/planar membrane which is another L/C
It appears that the same thing in the same place there has other "merit" that is not intended
Power amps have LPF in the input section, trying to eliminate RF entering the non-inverting input. It's like a man going to a battle using steel armour in the chest, hoping the enemy will be attacking this chest armour from the front, actually the enemy stab from behind.
Inductor and capacitor has their non-linearities. Some designers are trying to eliminate both of these things inside power amp.
But one dumb thinking : even if the power amp is free from capacitors and inductors anywhere in the cct, as soon as it is in use, it is connected via speaker cable (which is inherent RLC) then entering passive x-over network which is full of RLC, then entering speaker coil/planar membrane which is another L/C
lumanauw said:Output inductor usually placed there because of 1 reason : to prevent instability with capacitive load.
Bingo, The speaker wires are not acting like an antenna... Think about the logistics of receiving a radio signal when your trying to receive it. Unless your right on top of the source with a high-Z circuit node the effect is nil.
The output of the amp is about the worst set of conditions for this, the signal is in the microvolt range and must pass through a 20:1 resistive divider back to the gain stages and it's all solidly referenced to ground. It isn't gonna happen. RFi can enter an amp through its high impedance nodes and uncontrolled loops acting as antennas in the signal path. Although this is not as much of a problem as it's built up to be especially if the circuit is inteliigently laidout.
Regards, Mike.
Hi, Mike,
I'm confused. Isn't that these 2 are conflicting at each other? Proper PCB layout is a proove that one pcb track (small copper line) is prone to disturbance.
I've seen a pcb track, about 10-15cm, can generate voltage (spikes/ripple) between the 2 ends, seen on osciloscope.
Speaker cables are meters long. They should be affected more than a pcb track?
lumanauw said:
True but the speaker cable is looking into the low impedance of the output stage and the actual feedback entry into the gain stages is isolated (not the best term here) by the 20:1 drop of the FB network. Think about the conditions present to be able to see the spikes that you experienced. there wasn't a low impedance load or a set of power transistors in parallel back to the power supply. Probably a scope with a 10X probe.
As a kid I started building crystal radios and the antenna was probably 50' long when it was attached to the reciever. the signal levels where at a whisper and it took being rectified and filtered to get that from the headphones, and that was when the length of the antenna matched (1/4 wavelength) to the carrier frequency.
There are many ways that interference can be a problem but only if the reciever conditions are met.
Regards, Mike.
There has to be