janneman: " ... the whole thing with a closed loop gain of say 10x which is 20dB. You'll see that the higher the closed loop gain, the less your stability problems. That's logical: less closed loop gain is less [out of phase] feedback. ... "
Pardon my edit above, but this is not really "logical" or intuitive. At first blush one would assume that increasing feedback would or should exaserbate the problem ... unless one were totally familiar with all of the factors of feedback (gain, phase shift, distortion, slew rates, etc). It does explain why a unity gain op-amp with, say, 10k input resistors and 10k (-) feedback resistor is more stable than one with no resistors in the circuit ... ?? Not ??
Needed: an extended book list of revised op-amp cookbooks ala Walt Jung, Don Lancaster, Bob Pease, Forest Mims ... ( http://www.amazon.com/Op-Amp-Cookbook-3rd-Walter-Jung/dp/0138896011 [1986], http://www.amazon.com/Active-Filter-Cookbook-Second-LANCASTER/dp/075062986X [1996], http://www.amazon.com/Timer-Amp-Optoelectronic-Circuits-Projects/dp/0945053290 [2000], http://www.amazon.com/Troubleshooting-Analog-Circuits-Design-Engineers/dp/0750694998 [1991] ) ... plenty here, but cookbooks targeted strictly for analog / audio would be nice. Got links? Got a favorite ??
Pardon my edit above, but this is not really "logical" or intuitive. At first blush one would assume that increasing feedback would or should exaserbate the problem ... unless one were totally familiar with all of the factors of feedback (gain, phase shift, distortion, slew rates, etc). It does explain why a unity gain op-amp with, say, 10k input resistors and 10k (-) feedback resistor is more stable than one with no resistors in the circuit ... ?? Not ??
Needed: an extended book list of revised op-amp cookbooks ala Walt Jung, Don Lancaster, Bob Pease, Forest Mims ... ( http://www.amazon.com/Op-Amp-Cookbook-3rd-Walter-Jung/dp/0138896011 [1986], http://www.amazon.com/Active-Filter-Cookbook-Second-LANCASTER/dp/075062986X [1996], http://www.amazon.com/Timer-Amp-Optoelectronic-Circuits-Projects/dp/0945053290 [2000], http://www.amazon.com/Troubleshooting-Analog-Circuits-Design-Engineers/dp/0750694998 [1991] ) ... plenty here, but cookbooks targeted strictly for analog / audio would be nice. Got links? Got a favorite ??
You really need to read The Art Of Electronics
Basically an opamp has an open loop gain which is V(out) / ( V(in+) - V(in-) ), being a complex number with magnitude and phase.
When you define the closed loop gain with a feedback network, you use part of this OL gain as your closed loop gain, and the rest as fedback factor.
Suppose you have an OL gain of 60dB at 1 kHz ; you want closed loop gain of 10 (20dB), then you'll have 40 dB of feedback.
The more OL gain you use as closed loop gain, the less you have available for feedback.
More intuitive ?
Basically an opamp has an open loop gain which is V(out) / ( V(in+) - V(in-) ), being a complex number with magnitude and phase.
When you define the closed loop gain with a feedback network, you use part of this OL gain as your closed loop gain, and the rest as fedback factor.
Suppose you have an OL gain of 60dB at 1 kHz ; you want closed loop gain of 10 (20dB), then you'll have 40 dB of feedback.
The more OL gain you use as closed loop gain, the less you have available for feedback.
More intuitive ?
Back to Op-Amp Experiences and How To
" ... Basically an opamp has an open loop gain which is V(out) / ( V(in+) - V(in-) ..." thus the term "differential amplifier", and the difference between V(in+) and V(in-) being the "gain factor".
" ... you use part of this OL gain as your closed loop gain ... you want closed loop gain of 10 (20dB), then you'll have 40 dB of feedback. ... The more OL gain you use as closed loop gain, the less you have available for feedback. ... More intuitive ? ..." ... Not.
We may have a "lost in translation" question here as well. I think I understand what you are trying to say ... but the " V(out) / ( V(in+) - V(in-)" part is easy to understand and is "intuitive".
Now ... getting back to the more detailed questions ... the devil is in the details like phase shift reduction, roll off approaching unity gain bandwidth, improving slew rate, reducing output noise, tweakin tips (plastic caps on the power pins, etc.) ... theories are nice, but lab bench experiences are what we are looking for here, no?
" ... Basically an opamp has an open loop gain which is V(out) / ( V(in+) - V(in-) ..." thus the term "differential amplifier", and the difference between V(in+) and V(in-) being the "gain factor".
" ... you use part of this OL gain as your closed loop gain ... you want closed loop gain of 10 (20dB), then you'll have 40 dB of feedback. ... The more OL gain you use as closed loop gain, the less you have available for feedback. ... More intuitive ? ..." ... Not.
We may have a "lost in translation" question here as well. I think I understand what you are trying to say ... but the " V(out) / ( V(in+) - V(in-)" part is easy to understand and is "intuitive".
Now ... getting back to the more detailed questions ... the devil is in the details like phase shift reduction, roll off approaching unity gain bandwidth, improving slew rate, reducing output noise, tweakin tips (plastic caps on the power pins, etc.) ... theories are nice, but lab bench experiences are what we are looking for here, no?
tawn10 said:
analog_sa said:
Are you sure those components actually influence the sound? Are you sure it’s not just the big, squishy grey lump between your ears making the components sound better? When I ask are you sure, I mean facts, graphs, curves, measurements and not anecdotal fluff-words like open, airy, tight, spatial, etc. Not to throw a wrench into the warm-and-fuzzies of DIY audio, but for new builders, would be beneficial to learn some solid science and methodologies along with electronics.
Do some Google searches on op-amps, op-amp design, op-amp applications, audio op-amps, etc. and you will uncover a wealth of information from OEMs, other DIY’ers, and data sheets. Plus, there no better experience than getting a Radio Shack breadboard and a handful of components and building, experimenting, and troubleshooting circuits to get intimately familiar with circuits.
To the newcomers, don’t get frustrated with this site too quickly. It probably has all of the information you’re asking for, but just not all summed up and presented in a nice sequential fashion as we’d all love to see and offer. Part of the fun with this site is to search, read, explore, and learn. Before you know it, you’ll figure out the BS stuff, the good stuff, and probably find yourself thinking about new projects and new design ideas. There’s a wealth of information and ultra-educated posters on this forum, but there’s also the fluff, junk, and woowoo (audiophool’s) garbage as well.
Re: Re: Re: OP AMP EXPERIENCES and HOW TO
Thermal drift and thermal coupling between output and input are certainly related but temperature induced drift which is a function of macroscopic environmental effects does not necessarily correlate with thermal coupling from output to input due to temperature gradients across the chip. An increase in distortion below 1kHz would be an indication of thermal coupling, and this does seem to be quite low in modern op amps. Certainly thermal drift is much higher in discrete circuits but thermal coupling as seen in op amps is absent due to the much larger distance between input and output transistors. An output buffer eliminates thermal coupling in op amps as Walt Jung describes.
Walt's later articles on this and other op amp limitations are from 1998 and are available here . Still well worth reading.
janneman said:
Thermal drift and thermal coupling between output and input are certainly related but temperature induced drift which is a function of macroscopic environmental effects does not necessarily correlate with thermal coupling from output to input due to temperature gradients across the chip. An increase in distortion below 1kHz would be an indication of thermal coupling, and this does seem to be quite low in modern op amps. Certainly thermal drift is much higher in discrete circuits but thermal coupling as seen in op amps is absent due to the much larger distance between input and output transistors. An output buffer eliminates thermal coupling in op amps as Walt Jung describes.
Walt's later articles on this and other op amp limitations are from 1998 and are available here . Still well worth reading.
FastEddy said:janneman: " ... the whole thing with a closed loop gain of say 10x which is 20dB. You'll see that the higher the closed loop gain, the less your stability problems. That's logical: less closed loop gain is less [out of phase] feedback. ... "
Pardon my edit above, but this is not really "logical" or intuitive. At first blush one would assume that increasing feedback would or should exaserbate the problem ... unless one were totally familiar with all of the factors of feedback (gain, phase shift, distortion, slew rates, etc). It does explain why a unity gain op-amp with, say, 10k input resistors and 10k (-) feedback resistor is more stable than one with no resistors in the circuit ... ?? Not ??
Needed: an extended book list of revised op-amp cookbooks ala Walt Jung, Don Lancaster, Bob Pease, Forest Mims ... ( http://www.amazon.com/Op-Amp-Cookbook-3rd-Walter-Jung/dp/0138896011 [1986], http://www.amazon.com/Active-Filter-Cookbook-Second-LANCASTER/dp/075062986X [1996], http://www.amazon.com/Timer-Amp-Optoelectronic-Circuits-Projects/dp/0945053290 [2000], http://www.amazon.com/Troubleshooting-Analog-Circuits-Design-Engineers/dp/0750694998 [1991] ) ... plenty here, but cookbooks targeted strictly for analog / audio would be nice. Got links? Got a favorite ??
Well, I still think that it is intuitive that if you have less feedback you have less problems with stability. I must assume SOME basic knowledge, like that oscillations cannot occur if there is no feedback.
The other factors you mention certainly are important, but are secondary etc effects. I was giving a very simple but correct example as was requested.
Jan Didden
Re: Re: Re: Re: OP AMP EXPERIENCES and HOW TO
Indeed, and thanks for the links. But then again, Walt mentions that modern opamps have 1uV thermal feedback effects from output to input IIRC. I don't think I will worry too much about that. Also, these effects would show up in THD measurements below 1kHz anyway. I agree these are important issues, but would not be a reason to badmouth IC's and switch to discretes for these reasons.
Jan Didden
nuvistor said:
Indeed, and thanks for the links. But then again, Walt mentions that modern opamps have 1uV thermal feedback effects from output to input IIRC. I don't think I will worry too much about that. Also, these effects would show up in THD measurements below 1kHz anyway. I agree these are important issues, but would not be a reason to badmouth IC's and switch to discretes for these reasons.
Jan Didden
FastEddy said:" ... I still think that it is intuitive that if you have less feedback you have less problems with stability...."
... and your favorite op-amp cookbook??
My first one was Walts opamp cookbook from the late 70's IIRC. But I have moved beyond cookbooks after many years of studying on and working with opamps. I have a few books on feedback systems and theory I use to look-up subjects. I am still not too strong in complex calculations with s and j, but I am at a point that I have to master it to go further. One does what one can
Jan Didden
Care to elaborate on how an op-amp circuit (as it relates to this post) can be objectively improved for audio with wire selection?
Off-topic, I’d love to see additional posts on how to definitively test and measure the differences between a $0.01 1% metal-film resistor and a $4.00 1% resistor? Maybe suggest a listening or measurement test to show the difference between 1uF film caps from Auricap and Cornell Dubilier? I’d love to see what 99.999% of the scientific and electronics world doesn’t know.
Off-topic, I’d love to see additional posts on how to definitively test and measure the differences between a $0.01 1% metal-film resistor and a $4.00 1% resistor? Maybe suggest a listening or measurement test to show the difference between 1uF film caps from Auricap and Cornell Dubilier? I’d love to see what 99.999% of the scientific and electronics world doesn’t know.
So who does ABX testing of components? Of course any self-respecting component connoisseur knows that ABX testing does not represent real listening conditions and so cannot be depended on. Fortunately this forum doesn't require ABX testing to back up subjective impressions, unlike hydrogenaudio.org and its TOS #8.
Well, you can start with conductors ... resistivity
http://en.wikipedia.org/wiki/Copper = 16.78 n?·m
http://en.wikipedia.org/wiki/Gold = 22.14 n??·m
http://en.wikipedia.org/wiki/Silver = 15.87 n??·m
So there are differences in the way electrons flow down these, Silver being better, least resistive. Accordingly the impedence (combination of capacitance, resistance, inductance, etc.) of the connections and wiries will be different, Silver being better, lower impedence. Same, same for thermal conductivity. Silver & Hafnium are the new material of choice for Intel's fastest processors, partly because of the lower resistance, keeping up with Moore's Law re: speed, size.
Teflon ( http://en.wikipedia.org/wiki/Teflon ) seems to be about the best compromise for insulators = tough, flexible, chemical resistant, etc., = "... excellent dielectric properties. ... Its extremely high bulk resistivity makes it an ideal material for fabricating long life electrets, useful devices that are the electrostatic analogues of magnets. ..."
I use Teflon insulated, fine stranded silver wire where I can, and the results on a 'scope or meter indicate as much as I could convey, subjectively. The better quality electronics kits, laboratory equipment, etc. use it as well.
http://en.wikipedia.org/wiki/Copper = 16.78 n?·m
http://en.wikipedia.org/wiki/Gold = 22.14 n??·m
http://en.wikipedia.org/wiki/Silver = 15.87 n??·m
So there are differences in the way electrons flow down these, Silver being better, least resistive. Accordingly the impedence (combination of capacitance, resistance, inductance, etc.) of the connections and wiries will be different, Silver being better, lower impedence. Same, same for thermal conductivity. Silver & Hafnium are the new material of choice for Intel's fastest processors, partly because of the lower resistance, keeping up with Moore's Law re: speed, size.
Teflon ( http://en.wikipedia.org/wiki/Teflon ) seems to be about the best compromise for insulators = tough, flexible, chemical resistant, etc., = "... excellent dielectric properties. ... Its extremely high bulk resistivity makes it an ideal material for fabricating long life electrets, useful devices that are the electrostatic analogues of magnets. ..."
I use Teflon insulated, fine stranded silver wire where I can, and the results on a 'scope or meter indicate as much as I could convey, subjectively. The better quality electronics kits, laboratory equipment, etc. use it as well.
"He uses no global feedback. Rather, feedback is employed within each of the four separate gain stages to ensure that all forms of distortion are held as low as possible. RIAA compensation is achieved using passive high pass networks, including the proper time constants, located between the independent gain stages of the amplifier chain, again to achieve low distortion. This also aids in contributing to very high dynamic range and aids in the eradication of transient intermodulation distortion, a common byproduct of phono stage configurations where the throughput gain and RIAA compensation are realized by using loop feedback."
http://www.positive-feedback.com/Issue16/dsa.htm
http://www.positive-feedback.com/Issue16/dsa.htm
KBK: Thanks for your enlightened research, although I don't see any indication of the use of special conductors or insulation in the reference. That phono pre-amp does use multiple stages of op-amps to acheive the desired result, flying in the face of my own experience that this might otherwise degrade rather than enhance the results ... but I've been wrong before ... Maybe positive feedback amps can acomplish this with more, rather than fewer gain stages ...
Hello tawn10:
I didn’t read the posts, short on time. The following may help you out.
No matter how good the op-amp is, and many are fine devices, there is one main issue that determines the sound of any of them.
The grounding technique is what is so very important.
Find the Burr Brown and Analog Devices tutorials, they are loaded with basic info.
Back to grounding: with rare exception an op-amp is referenced to the negative supply rail. This means that whatever noise is on the negative rail bounces the output accordingly.
The instant remedy for power supply noise in general is to strap a 1000pf ceramic (preferably an NPO or class II) across the positive and negative supply terminals of the I.C. underneath the board.
Also, forget the “star-ground” concept, it is useless theory. Use large solid conductors (18-gauge is fine) as busses for supply rails. Put a tantalum and a ceramic across them about every inch. Tin the busses. They are often referred to as “clotheslines.”
Also, put a 10 ohm resistor on each supply pin and use a capacitor to make a low-pass filter. 7-10 ohm plus .1 uf is OK. Don’t go over 10 ohms or it will modulate the supply terminals. Don’t use class-III ceramics in audio. Panasonic stacked metal films are also excellent.
The PSRR (power supply rejection ratio) given in spec sheets is for low freq, about 100Hz. At about 20KHz the PSRR is near nothing, this is why the filters help out.
There are many other important design factors, but the grounding and supply filtering is the most important.
As for practical audio op-amps, the NE5532 and 5534 have always been excellent choices. When I learned about grounding everything started working right. Have Fun, Mark
I didn’t read the posts, short on time. The following may help you out.
No matter how good the op-amp is, and many are fine devices, there is one main issue that determines the sound of any of them.
The grounding technique is what is so very important.
Find the Burr Brown and Analog Devices tutorials, they are loaded with basic info.
Back to grounding: with rare exception an op-amp is referenced to the negative supply rail. This means that whatever noise is on the negative rail bounces the output accordingly.
The instant remedy for power supply noise in general is to strap a 1000pf ceramic (preferably an NPO or class II) across the positive and negative supply terminals of the I.C. underneath the board.
Also, forget the “star-ground” concept, it is useless theory. Use large solid conductors (18-gauge is fine) as busses for supply rails. Put a tantalum and a ceramic across them about every inch. Tin the busses. They are often referred to as “clotheslines.”
Also, put a 10 ohm resistor on each supply pin and use a capacitor to make a low-pass filter. 7-10 ohm plus .1 uf is OK. Don’t go over 10 ohms or it will modulate the supply terminals. Don’t use class-III ceramics in audio. Panasonic stacked metal films are also excellent.
The PSRR (power supply rejection ratio) given in spec sheets is for low freq, about 100Hz. At about 20KHz the PSRR is near nothing, this is why the filters help out.
There are many other important design factors, but the grounding and supply filtering is the most important.
As for practical audio op-amps, the NE5532 and 5534 have always been excellent choices. When I learned about grounding everything started working right. Have Fun, Mark
hailteflon: " ... with rare exception an op-amp is referenced to the negative supply rail. This means that whatever noise is on the negative rail bounces the output accordingly. ..."
Because that is where the electrons are ... Special care and attention to the negative rail means making sure the star ground or bus ground is well filtered. If you have quality capacitance on the positive rail, but not the negative, well ... Same, same for power supply filtering inductors, on all rails, +, - and not to forget the ground rail.
An example of this "double" capacitance and inductance applied to both rails. Example: AC and DC power line filters in laboratory grade equipment = both hot and neutral (hot and common) power lines each have inductors and capacitors, "coming and going", hot and neutral, black and white wires ...
" ... forget the “star-ground” concept ... Use large solid conductors ... as busses for supply rails ... [capacitors] across them about every inch. Tin the busses. ..."
I would not discount the "star-ground" topography so readily, but your point is well taken with regard to a ground bus topography with, as you note, multiple capacitors along the ground bus, whether large, solid conductor or fat PCB traces. My suggestions would include a combination or daisy chain of star-grounds ,,, The Improtant criteria being that there be One and One Only ground bus ... with staggered position / spacing of the capacitors, rather than 1", evenly spaced capacitors ... and of differing dielectric material, electrolytic at the power source, plastic (poly-xxx type) and / or ceramic very close to the op-amp / amp stages' power pins. "Close coupled" plastic or ceramic caps directly connected to op-amp power pins cover a world of power supply sins. Both TI / Burr Brown and Analog Devices as well as National tutorials' mention this. And many manufacturers' analog ICs circuit diagrams show these 0.01 uF to 0.001 uF "snubbing" caps ... the "speed secret" being the cap material = plastic or ceramic.
" ... put a 10 ohm resistor on each supply pin [rail] ..." Actually, I would suggest an inductor (coil) or a wire wound resistor for this = resistance and added inductance = greater filtration at even lower frequencies.
IMOP:
* Snubbing Caps: Plastic (polystyrene, polypropylene, MKT type) being nominally better than ceramic and significantly better than electrlytic, a combination of both or all three types being best.
* Ground Bus: Centrally located, single, fat PCB ground traces are better than attempts at building ground planes and, as you note, generously tinned, in parallel with the power rails. Ground bus size being ~~ twice a large as either + or - rail.
* Cap Locations: Staggered spacing of ground bus filter caps, especially close to the active elements (op-amps) = as close as possible = 1 inch being too much at the power pins. (Interestingly, a logarithmic spacing scenario works quite well (base e or base 10) = use a slide rule as a measuring device: http://upload.wikimedia.org/wikipedia/commons/a/a0/Pocket_slide_rule.jpg .)
* Hail Teflon ... indeed.
Because that is where the electrons are ... Special care and attention to the negative rail means making sure the star ground or bus ground is well filtered. If you have quality capacitance on the positive rail, but not the negative, well ... Same, same for power supply filtering inductors, on all rails, +, - and not to forget the ground rail.
An example of this "double" capacitance and inductance applied to both rails. Example: AC and DC power line filters in laboratory grade equipment = both hot and neutral (hot and common) power lines each have inductors and capacitors, "coming and going", hot and neutral, black and white wires ...
" ... forget the “star-ground” concept ... Use large solid conductors ... as busses for supply rails ... [capacitors] across them about every inch. Tin the busses. ..."
I would not discount the "star-ground" topography so readily, but your point is well taken with regard to a ground bus topography with, as you note, multiple capacitors along the ground bus, whether large, solid conductor or fat PCB traces. My suggestions would include a combination or daisy chain of star-grounds ,,, The Improtant criteria being that there be One and One Only ground bus ... with staggered position / spacing of the capacitors, rather than 1", evenly spaced capacitors ... and of differing dielectric material, electrolytic at the power source, plastic (poly-xxx type) and / or ceramic very close to the op-amp / amp stages' power pins. "Close coupled" plastic or ceramic caps directly connected to op-amp power pins cover a world of power supply sins. Both TI / Burr Brown and Analog Devices as well as National tutorials' mention this. And many manufacturers' analog ICs circuit diagrams show these 0.01 uF to 0.001 uF "snubbing" caps ... the "speed secret" being the cap material = plastic or ceramic.
" ... put a 10 ohm resistor on each supply pin [rail] ..." Actually, I would suggest an inductor (coil) or a wire wound resistor for this = resistance and added inductance = greater filtration at even lower frequencies.
IMOP:
* Snubbing Caps: Plastic (polystyrene, polypropylene, MKT type) being nominally better than ceramic and significantly better than electrlytic, a combination of both or all three types being best.
* Ground Bus: Centrally located, single, fat PCB ground traces are better than attempts at building ground planes and, as you note, generously tinned, in parallel with the power rails. Ground bus size being ~~ twice a large as either + or - rail.
* Cap Locations: Staggered spacing of ground bus filter caps, especially close to the active elements (op-amps) = as close as possible = 1 inch being too much at the power pins. (Interestingly, a logarithmic spacing scenario works quite well (base e or base 10) = use a slide rule as a measuring device: http://upload.wikimedia.org/wikipedia/commons/a/a0/Pocket_slide_rule.jpg .)
* Hail Teflon ... indeed.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.