This is why I use an open loop output stage, call me crazy, but even if one cannot entirely prove the back emf theory it is also another step similar to degeneration in the voltage gain stage. It did and still makes too much sense to me to break the loop before the speaker 🙂
Vynuhl, see also DF96 last remark. In addition: you need the output stage in the loop to lower the output impedance of you amplifier. If you don't, impedance variations in the load will lead to linear distortion. Back EMF is one of the mechanisms that create such impedance variations.
Therefore, it is completely opposite to what you argue.
This is not to say that the effect of increasing the output impedance of your amp may not be enjoyable (for a while). It will add some warmth to your typical loudspeaker, which will have a resonant peak on the low end. It may also crash the voice coil of your bass driver into the back plate.
Therefore, it is completely opposite to what you argue.
This is not to say that the effect of increasing the output impedance of your amp may not be enjoyable (for a while). It will add some warmth to your typical loudspeaker, which will have a resonant peak on the low end. It may also crash the voice coil of your bass driver into the back plate.
It is no problem to use for various measurements complex load, simulating real loudspeaker load properties. Properly designed amplifier will have no problems , and you will not find here nothing special, nor "magical". And each serious, competent designer have to do it so, it is nothing special. Finally, audio amplifier is intended to drive loudspeakers.
Cordell and I agreed a while back on the efficacy of a fairly high input Z for a power amp, to make the system less sensitive to source Z, especially things like electrolytic coupling caps. Also, Self has a discussion of distortions arising at inputs, I think in his Small Signal Audio Design.
I made one power amp with a LSK170 bootstrapped-drain source follower and a megohm gate R. It also had load compensation using a negative R generator. The only problem was that the sales guy hooked up to it with unshielded clip leads and it quickly smoked an output Zobel network 😱 That prompted a shutdown system, although for many applications it would never be activated.
Brad
I've settled on 50 ohms output Z on my pre's (although I have one with 22 Ohms) and10k with 1k and 220 pF in parallel with that for the input filter on the PA. The input z of the input devices 30 to 50 k in parallel with that again.
I was reading a review a short while ago on a Vaccuum State preamp in Hi-Fi+ where the oust Z was quoted as '100 k Ohms'. If used with a solid state PA that will almost certainly be affected sound wise - even with a more typical 22 k to 47 k. For those types of pre's you really do need high Z inputs.
Personally, my preference is for about 10 k - good compromise all round IMV.
I was reading a review a short while ago on a Vaccuum State preamp in Hi-Fi+ where the oust Z was quoted as '100 k Ohms'. If used with a solid state PA that will almost certainly be affected sound wise - even with a more typical 22 k to 47 k. For those types of pre's you really do need high Z inputs.
Personally, my preference is for about 10 k - good compromise all round IMV.
Self wrote about arising distortion at the input in his 6th Audio Power Amp Design book. If input pars are complementary as in most cases for CFA then there is no distortion arising at the input with higher source impedance.
Damir
I separate: idealized global feedback (with extremely high open loop bandwidth), from usual global feedback, and local feedback. In my opinion, idealized global feedback 'might' be OK, most of the time, but usual global feedback is problematic. I have rarely found a problem with local feedback, except in removing cathode bypass caps in an open loop circuit, and then only with live music. Still, it is noting the subtle differences that makes for real success or just so so.
Having just picked up non switched amplifiers again (after having to live with designing class D for while) I took this approach:
I have had a great deal of success using simulators to demonstrate the capability of different circuits, and even had the joy of building the first prototype of my current project and not have the thing blow up in my face upon first power up.
For me it is all about getting an amp accepted by the golden ears, and then trying to somehow match their complaints to something measurable... at task not always easy - then iterate to next project.
\\\Jens
- Topology chosen based on requirements (Cost, audio spec....)
- No sub circuit should be allowed to saturate - ever or as close to as possible
- All circuits should be made as linear as possible, using local FB
- Global feedback will be applied to tame DC offset, output impedance and to get THD to an acceptable level (Depending on requirements)
I have had a great deal of success using simulators to demonstrate the capability of different circuits, and even had the joy of building the first prototype of my current project and not have the thing blow up in my face upon first power up.
For me it is all about getting an amp accepted by the golden ears, and then trying to somehow match their complaints to something measurable... at task not always easy - then iterate to next project.
\\\Jens
Member
Joined 2009
Paid Member
I think this is the approach I take (except when I'm lazy, just playing around or want something with a 'sound') - I suspect it is universal?
Well there's a thought to push as much of the gain into the outer loop, as *theoretically* that should bring about the greatest error correction*.
*That said, that must be balanced against local, egregious nonlinearities that are best stabilized by local feedback/degeneration (see: the TIM battles of the 70's). That said^2, with solid-state it shouldn't be *too* hard to get distortion products to inaudible levels while still keeping large-signal nonlinearities in check. For a decent/good designer, of which I make no claims to being. 🙂
*That said, that must be balanced against local, egregious nonlinearities that are best stabilized by local feedback/degeneration (see: the TIM battles of the 70's). That said^2, with solid-state it shouldn't be *too* hard to get distortion products to inaudible levels while still keeping large-signal nonlinearities in check. For a decent/good designer, of which I make no claims to being. 🙂
Member
Joined 2009
Paid Member
Providing that you keep feedback loops stable you only need one, outer loop, to do the job don't you ?
Of course, smaller inner-loops can be easier to make stable with some toplogies, parts and real-world pcb layouts.
Of course, smaller inner-loops can be easier to make stable with some toplogies, parts and real-world pcb layouts.
Yeah, I mean if one looks at degeneration as feedback--then really there are always multiple loops rather than usually the LF: around-everything, HF: let the OPS run free.
I also made the most linear amp that I could and it can work without global feedback. Adding just 6 to 9dB of global feedback loop (CFB) helps and it sounds more clear on treble, I can hear it, but raising amount of global feedback more than 9-10dB -s (even tried 40dB of GNFB) is just "compressing" the sound, sounds worse and one can hear it clearly... So, Bruno`s >30dB of GNFB always improve the sound, does not work in my situation. (Simulation shows lower harmonics with higher GNFB and there is no hi order harmonic raising with low GNFB as Bruno said in his F-word article). So....
You can design an open loop stage with appropriate current for the devices linear region and paralleled devices to get low output impedence without the gnfb loop. I get an output impedence around 100mohms maximum, my Esl are nominal 5 ohms lowering to 1ohm at 20k, so the lowest DF is 10 at 20k and minimum of 50 @1k and lower, in the sub 1k impedence will spike up to 18r, then drop again , so for 8ohms which all amps are measured it has a DF of 80 min which is highly acceptable.
Originally this choice of isolation had nothing to do with back emf, it was done to make for a much more stable amplifier into any load where stability of the gain stage is not affected by the speaker, that's all 🙂
I'm just some guy, y'know?
Audio is only a hobby for me - I'm employed as an engineer in the medical imaging industry and quite happy about it too given the inflated egos that abound in the audio area.
Don't really see why I should need to offer any credentials for expressing my opinions (with what I hope is a reasonable degree of humility) here though.I wholeheartedly agree that Bonsai should be appreciated for his contributions and for making all his hard work in his designs freely available and have previously said so. He lost most of my respect when he started making up / perpetuating bad stories about MF and refused to accept he might have been in the wrong though. And by the way you're obviously confused about who was mudslinging 😱
At the moment, I'm using an Oppo BDP-103 as a CD/WAV file player (superb bargain it is, just love it) with my Silicon Chip Classic DAC (modified a little), my own design preamp(s) and my Leach amp or SC350 monobloks driving my restored B&W801s. It's a pretty satisfying system but I would love to buy some nicer gear if the budget and SWMBO would allow it...
I would love to share my preamp designs here and have thought about it, but figured I'd only get bombarded with a bunch of smart-**** "objectivist" critique based on simulation results or an endless string of arm-chair tweak suggestions and I really don't have the time to keep reading here regularly to keep up with questions. I guess if anyone's interested let me know and I'll give it a try.
I remain puzzled by how important the humble line preamp is to the overall sound of a system - this obviously goes against logical analysis. NJR brand NE5534 (with a current source loading the output to +ve rail as suggested by Self) is my favourite opamp for this if opamps must be used.
As for source material, that'll have to wait for another essay 🙂
I agree entirely. But if you look at the input impedance of most power amps, yes it decreases at HF, mainly due to LPF at the input, but not enough to really explain the SQ difference. For example, say a 20k input resistance power amp has an input Z of 5k at 10kHz, any reasonable preamp output opamp should still drive this with no problems and with still plenty of margin for any complex impedance.
Glad to see you changing your tune (no pun intended) 😛
I think it comes down to the art of recording too - how many good recordings of piano are out there? Bugger all if you ask me. I guess mixing desk designers are only giving the recording engineers what they demand, if you know what I mean... obviously opamps and the hard line "objectivist" stance taken by Self make this task economically achievable, as does pretending it's pure science with no room for art. If the designer had to listen carefully to each stage and rather than just measuring THD/SNR...
Owdeo,
The scientists don't trust their ears, or if they start to something must be wrong, poor buggers 🙁. The Audio industry is in competition to hold on to what is left of a customer base, even the firmly embedded ones more so than ever need to have an angle to sell to make money, the people who truly care about audio reproduction are a diminishing breed unfortunately and the economy does not allow the indulgent spending it used to.
Colin
The scientists don't trust their ears, or if they start to something must be wrong, poor buggers 🙁. The Audio industry is in competition to hold on to what is left of a customer base, even the firmly embedded ones more so than ever need to have an angle to sell to make money, the people who truly care about audio reproduction are a diminishing breed unfortunately and the economy does not allow the indulgent spending it used to.
Colin
Anyone who doesn't look at degeneration as feedback is not seeing straight - it is feedback! However, it senses a proxy for output current rather than the output voltage so it gives a different result - raises output impedance, for example.Derfnofred said: