The OL loop bandwidth has no relationship to the IM performance of an opamp.
I reckon the LME OL bandwidth is between 10 and 100Hz.
The only way to test your assertion that the Borbely sounds better I'd through a double blind test. Trust your ears, but not your processor on this one.
I reckon the LME OL bandwidth is between 10 and 100Hz.
The only way to test your assertion that the Borbely sounds better I'd through a double blind test. Trust your ears, but not your processor on this one.
Hi Jacco:
Thanks for the link to Walt's article!
I learned that the OL bandwidth limitation causes a phase shift around the corner of the bandwidth.
For many op-amps, this corner is in the audio range.
Walt also says that BJT transistors are less linear than JFETS.
Thanks!
Dan
Thanks for the link to Walt's article!
I learned that the OL bandwidth limitation causes a phase shift around the corner of the bandwidth.
For many op-amps, this corner is in the audio range.
Walt also says that BJT transistors are less linear than JFETS.
Thanks!
Dan
IMO it's an interesting article and worth understanding, but it's also a case of setting up a straw man argument. Almost any opamp you choose will have the first pole in the audio region. Even those that aren't unity gain stable will be that way. It's a big "so what" unless you allow out-of-band signals to come blasting into your system and IMO competent designs will have provisions to prevent that. It's a well known issue and the usual solution is obvious compared to the solutions presented. Also, JFETs are nice for many things, but BJTs will win on noise in low source impedance applications. JFETs will win at higher impedances. Courses for horses. (I admit not even knowing what process the LME amp uses.)
discrete component amp stages are made with love.
thats why they are better.
...just kidding, but true, they get special attention to matching and quality.
also voltage swing potential: build a +32 dbu circuit but electrically allow +24 dbu to pass and set clip light detection at this level (standard Neve design principle).
thats why they are better.
...just kidding, but true, they get special attention to matching and quality.
also voltage swing potential: build a +32 dbu circuit but electrically allow +24 dbu to pass and set clip light detection at this level (standard Neve design principle).
i would suggest that the lme49990 is your choice over the plasticky sounding 49710 if you are after truly state of the art new national chips and its quite simply the best chip i have ever not heard. opa827 and opa1641 are also great jfet input opamps. the 49990 is very much like the ad797, but has taken it to another level imo, 797 was my go to chip for several years for both audio and power supply in jung style regulators, but the 49990 also adds something, dont know how to describe it.
i think you should go over and check out the Synaethesia site, lhere is a site that takes the best of both discrete and ic basewd designs and uses both for what they do best. warning, these are pretty extreme designs and if you have an aversion to surface mount components best avoided. the designer Syn08 was a member, but i think got sick of the in fighting of this place and probably an aversion to his best of both worlds techniques, so he cancelled his account. because here people tend to bat for one side or the other. myself i'll use whatever is best for the job.
al design decisions and the progression through each iteration are covered in detail
i think you should go over and check out the Synaethesia site, lhere is a site that takes the best of both discrete and ic basewd designs and uses both for what they do best. warning, these are pretty extreme designs and if you have an aversion to surface mount components best avoided. the designer Syn08 was a member, but i think got sick of the in fighting of this place and probably an aversion to his best of both worlds techniques, so he cancelled his account. because here people tend to bat for one side or the other. myself i'll use whatever is best for the job.
al design decisions and the progression through each iteration are covered in detail
Last edited:
Conrad:
I understand your point. I wonder if OL BW becomes a non-issue once some feedback is applied (i.e. the corner moves out of the audio pass-band)? I do not have anything out of band unless it is on an LP record. I see the full power bandwidth of the LM4562 is 10Mhz. That is excellent! I still do not see an open loop bandwidth number. I guess we'd never use these thinks in open loop anyway! 🙂
Dave:
I understand the theoretical ability to match components better, but for a cascoded JFET input stage, I read that it can handle smaller voltages, especially in common mode. For me, I do not need +32dbu. I am operating at the standard +14 to +20dbu peak. So for me, headroom is not the issue, it is linearity and lack of distortion.
Qusp:
I'll look at that part as well. Maybe you can try one and let us know how it sounds-and more importantly how the sound is defined in measurements-that is my holy grail! 🙂
I have read the synatheasia site. Very nice designs. He did a good job.
My interest here is to have a cordial discussion where I and others learn. I know there are tid-bits in other threads tht I am slowly discovering. I have been drifting a little from the discrete JFET definition of sound to the newer op-amps, so I have muddied things a little. Sorry for having a wide net here.
With all this I am narrowing down what circuits I want to work with. So far, I my dream is to attempt to modify the Borbely discrete preamp to add cascoding and newer JFETs (maybe parallel two sets), and clean the PS a little. I am on the shelf about doing a balanced amp here, due to the amount of re-working. This may happen of I decide to build a new discrete preamp.
My next idea is to build a Hagerman Bugle Pro, but use a newer LME and put a discrete JFET front end to it. This is all intended for moving magnet RIAA and 78 RPM. An idea for down the road will also be to soup-up my MCI JH-110 reel to reel NAB playback preamps.
The main thought is for me to see and understand the major topologies so that I do not miss a major advancement that you know about.
This discussion has been very good!
Thanks!
Dan
I understand your point. I wonder if OL BW becomes a non-issue once some feedback is applied (i.e. the corner moves out of the audio pass-band)? I do not have anything out of band unless it is on an LP record. I see the full power bandwidth of the LM4562 is 10Mhz. That is excellent! I still do not see an open loop bandwidth number. I guess we'd never use these thinks in open loop anyway! 🙂
Dave:
I understand the theoretical ability to match components better, but for a cascoded JFET input stage, I read that it can handle smaller voltages, especially in common mode. For me, I do not need +32dbu. I am operating at the standard +14 to +20dbu peak. So for me, headroom is not the issue, it is linearity and lack of distortion.
Qusp:
I'll look at that part as well. Maybe you can try one and let us know how it sounds-and more importantly how the sound is defined in measurements-that is my holy grail! 🙂
I have read the synatheasia site. Very nice designs. He did a good job.
My interest here is to have a cordial discussion where I and others learn. I know there are tid-bits in other threads tht I am slowly discovering. I have been drifting a little from the discrete JFET definition of sound to the newer op-amps, so I have muddied things a little. Sorry for having a wide net here.
With all this I am narrowing down what circuits I want to work with. So far, I my dream is to attempt to modify the Borbely discrete preamp to add cascoding and newer JFETs (maybe parallel two sets), and clean the PS a little. I am on the shelf about doing a balanced amp here, due to the amount of re-working. This may happen of I decide to build a new discrete preamp.
My next idea is to build a Hagerman Bugle Pro, but use a newer LME and put a discrete JFET front end to it. This is all intended for moving magnet RIAA and 78 RPM. An idea for down the road will also be to soup-up my MCI JH-110 reel to reel NAB playback preamps.
The main thought is for me to see and understand the major topologies so that I do not miss a major advancement that you know about.
This discussion has been very good!
Thanks!
Dan
Oddly enough, Dan, some think that one should pay as much attention to the power supply design as to the main circus... it seems to have some sort of effect upon what is heard... of course, and as always, ymmv.
I suggest that you will be scratching your head if and when you hear differences in a circuit where you plug & swap in a few of these super-ultra-low distortion opamps?
Then too, what are your amps and speakers?? The rest of the signal path plays a role in what you are able or going to perceive.
Welcome to the rabbit hole.
_-_-bear
PS. the answer to all this is simply: Chrome. You need a lot of CHROME and shiny metal if the things are going to be worth a darn! You'll see... Chrome. 😀
I suggest that you will be scratching your head if and when you hear differences in a circuit where you plug & swap in a few of these super-ultra-low distortion opamps?
Then too, what are your amps and speakers?? The rest of the signal path plays a role in what you are able or going to perceive.
Welcome to the rabbit hole.
_-_-bear
PS. the answer to all this is simply: Chrome. You need a lot of CHROME and shiny metal if the things are going to be worth a darn! You'll see... Chrome. 😀
The open loop bandwidth (is there even a legitimate definition of this?) is the first pole of the opamp, which will be the very low frequency corner on the plot of the 4562 data sheet. I don't really see much of a corner there, the line just drops at 6 dB/Octave, so it would be about 10 Hz or less. IMO, totally irrelevant- go by the various distortion curves, the GBP and other more useful design parameters!
I did a lot of power supply testing on my RIAA preamp, even adding a switch so I could flip different filters and such in and out. My conclusion was that a) I probably can't hear most of the stuff people describe, and b) if the amp PCBs are properly bypassed it doesn't make any significant difference. Properly = very short lead lengths to HF capacitors like film or ceramic + larger electrolytics somewhere on each PCB (I prefer Oscons or high performance HF switching type caps).
It would be interesting to standardize your PCBs such that many different designs can be dropped into the same chassis. It would be even more interesting if you could switch between pairs of them, but that might add more complications.
Balanced amps are an interesting topic. You can cancel out external junk that affects the halves the same way, but a balanced design will almost always be noisier than a single ended design because you have two input circuits. S/N is established at the input and the noise doesn't cancel. Square root of 2 statistics apply, I think. Don't trust me on math! You can do very well, but not quite as well as a single ended design. "Balanced believers" can't accept this. IMO, balanced designs are highly desirable in pro sound applications, but I can't see the value in a home setup.
I did a lot of power supply testing on my RIAA preamp, even adding a switch so I could flip different filters and such in and out. My conclusion was that a) I probably can't hear most of the stuff people describe, and b) if the amp PCBs are properly bypassed it doesn't make any significant difference. Properly = very short lead lengths to HF capacitors like film or ceramic + larger electrolytics somewhere on each PCB (I prefer Oscons or high performance HF switching type caps).
It would be interesting to standardize your PCBs such that many different designs can be dropped into the same chassis. It would be even more interesting if you could switch between pairs of them, but that might add more complications.
Balanced amps are an interesting topic. You can cancel out external junk that affects the halves the same way, but a balanced design will almost always be noisier than a single ended design because you have two input circuits. S/N is established at the input and the noise doesn't cancel. Square root of 2 statistics apply, I think. Don't trust me on math! You can do very well, but not quite as well as a single ended design. "Balanced believers" can't accept this. IMO, balanced designs are highly desirable in pro sound applications, but I can't see the value in a home setup.
At this level of performance, and properly applied, I simply can't hear a difference. If you can, great, enjoy it. At 50+ years old it will probably be a memory, assuming it's not an illusion right now. 😀
Last edited:
there is something that people don't know/overlook, the class of operation of the op amp stages. on most, you could put 10k to 8k feed resistor from +Vcc to the output to inject more bias current on the internal feedback circuit resulting in class A operation.
and a more tuby warm sound
and a more tuby warm sound
ive just never liked them (49710), perhaps they just didnt suit the circuit. love them in power supply (performance, not sound) soulless in audio imo.
i see, so thats why all high performance instrumentation amps, measurement amps and any modern dac or adc where noise, or error matters are single ended? i understand your point, but its but one isolated source of noise, differential circuits have far more ways of dealing with noise than creating it imo
Last edited:
Conrad, you are in good company with Doug Sef re the balanced input being more noisy than single ended inputs. They are more noisy.
Some time back I had to do a commercial design where balanced input was a requirement. I tried to convince myself I was wrong about the noise issue and had a couple emails with Doug. Thus our agreement. 'Course we could both be wrong!
Concerning opamp "rolling", if I hear a difference, I go looking for the cause. The term "measurement" poorly describes the process because people tend to think in terms of a few traditional steady state measurements, but that's really what it is. In general, if something sounds different, the cause isn't that difficult to find. Differential comparisons will often show it up. I hate to say it, but when errors and problems are removed from circuits, and beyond a certain level of performance, they all tend to have the same soulless sound- adding or subtracting nothing. When I hear something that stands out from the pack, that's a warning flag that something is wrong!
Concerning opamp "rolling", if I hear a difference, I go looking for the cause. The term "measurement" poorly describes the process because people tend to think in terms of a few traditional steady state measurements, but that's really what it is. In general, if something sounds different, the cause isn't that difficult to find. Differential comparisons will often show it up. I hate to say it, but when errors and problems are removed from circuits, and beyond a certain level of performance, they all tend to have the same soulless sound- adding or subtracting nothing. When I hear something that stands out from the pack, that's a warning flag that something is wrong!
I started a thread:
http://www.diyaudio.com/forums/everything-else/191488-source-problem-souless-sound-new-post.html
after reading Conrad's comment, which is not unique to him, but well phrased none-the-less.
_-_-bear
http://www.diyaudio.com/forums/everything-else/191488-source-problem-souless-sound-new-post.html
after reading Conrad's comment, which is not unique to him, but well phrased none-the-less.
_-_-bear
Interesting, I feel that might be the case when you are in a design process, and don't have the time to listen extensively to every iteration. Or in a quick comparison between several components, where the most catchy of them is making illusion of a better sound. But, in my humble experience, long term and careful listening almost always reveal tiny imperfections, to the point of being unbearable when noticed for good.
Well statisticaly you have the pack, then a small number of bad performers, and a small number of good performers.
What is the 'problem' here. BALANCED IN, if done ONLY with devices, whether tube, bipolar or jfet, is 3 dB noisier than single ended in, with the same devices. However, IF make balanced in with a TRANSFORMER, you can reduce this to almost no significant increase in noise, IF the transformer is virtually perfect. (very expensive)
For example, in order to make my latest phono design to match the Vendetta Research phono input noise (10 ohms or 0.4nV/rt.Hz) I have to use 4 times as many devices on the input stage. Yes, instead of 4 jfets in parallel, I have to use 16 jfets in
series-parallel. That's what it takes.
For example, in order to make my latest phono design to match the Vendetta Research phono input noise (10 ohms or 0.4nV/rt.Hz) I have to use 4 times as many devices on the input stage. Yes, instead of 4 jfets in parallel, I have to use 16 jfets in
series-parallel. That's what it takes.
There 's no problem at all. It's just 3db noisier and that's a fact. Does that mean balanced signal chains are no good. No. They have their place in audio and are great at dealing with common mode noise.
- Status
- Not open for further replies.