Thanks Scott, that one set off my b/s detector. I recon if discrete was so much better, we would all still be using it.
I've been doing a bunch of preamp design lately and I think you can get some really excellent performance from op amps but they're not all the same. My personal take is that a 'coloured' result is probably not what to shoot for with op amps as it might be with tubes. (save maybe a transformer/opamp combo which can work out aesthetically if you get it right). But a very clean articulate result is possible, and very nice to listen to, but not as easy as it may seem. Transient distortion can be significant - i.e. ringing and overshoot problems. Very fast current feedback op amps can sort this out but then input / output impedance issues can become difficult. And having dealt with all that, voltage gain in combo with current gain seems to loose detail in practice for no reason I fully understand yet, so a high gain stage is well advised to run into a high quality buffer stage, whatever the numbers on paper seem to say... So I'm saying for my experience they can be great. INA103 plus AD844 can be top rate, though the details can be surprisingly heavyweight to work through. But I say aim for total purity with that and add colour from some other process if and as you want it. Certainly they can sound sh**t if you get it wrong! 🙂
I guess Mr. Jung could comment on the benefits of high bandwidth, especially with his first versions of the superregulator 😛
Could be interesting to do some experimenting with. I'm imagining maybe one opamp as an input stage, transformer follows that wound for a few times voltage gain, which runs into another opamp. With the first opamp to drive the transformer, it could be sensibly small cos it wouldn't need to have a huge primary inductance and you can control the impedance running into the transformer and the amount of drive you give it. Wound for a few times voltage gain should make sure its a bit 'fussy' and need a bit of tweaking with a zobel network or whatever to control ringing. The known input impedance of the next opamp means when you've got that set then its set. Between all that should mean theres controllable colouration from the transformer - degree of harmonic distortion from how much gain you apply at the first opamp. Alterable compromises with how you treat the ringing behaviour would affect masking / phase distortion / resonant gain. Try some different cheep-ee opamps in that second socket and do some adjusting and listening? Or am I crazy? Lol. Could be very revealing about what factors contribute what to the sound... 🙂
Not so sure about that. Attached is a lumped equivalent model of a transformer. X(m) is the primary inductance. If too small, LF signal current will tend to shunt through that inductance to ground rather than make it over to the secondary side. Are you thinking of some circuit topology where that wouldn't be an issue?
Attachments
Sure, I'm just thinking the first opamp behaves like it has a very low output impedance. Pushed too hard it'll reach the limits of the current it can supply but generally its negative feedback makes it act like it has near enough zero output impedance. The frequency that rolling off the LF happens at is dependent on the size of the inductance and of the impedance running into it (also gotta add in the DC resistance of the coil!) So reduce the inductance and the LF knee gets higher, but reduce the impedance feeding it and it gets lower again. The trade-off is that low frequencies are now drawing more current and so working the core harder causing more non-linearity. I've just finished winding a small transformer with a primary inductance of about 0.15H to use to balance the output of a microphone with an internal opamp buffer. I kept it small so I could minimise the parasitic capacitance and get good top end. Just checked it and the LF does go all the way down and the current draw isnt much to worry about really either. For the experiment idea I'm thinking a bit of saturation is part of the fun anyway! 🙂
Assuming this is Samuel Groner's work, looks like it, then that is a measurement artifact mentioned in the discussion in his pdf. ie. "ignore".
Last edited:
The AD-825 is the closest fit to an 'ideal' op amp that I have found for line level operation. Note the open loop bandwidth!
monkeybear,
Okay, understood. What are you using for a core? Most core geometries other than torroids are not especially optimal for saturating, since the ratio of magnetic path length to winding window area tends to limit amp-turns/meter (H-field).
Also, for fun, might be interesting to compare core materials with with wider or narrower B-H curves to see what more or less hysteresis does sound-effect-wise. Seems like saturation may not be the only effect that causes some people to like the sound of a little iron in the signal path.
Okay, understood. What are you using for a core? Most core geometries other than torroids are not especially optimal for saturating, since the ratio of magnetic path length to winding window area tends to limit amp-turns/meter (H-field).
Also, for fun, might be interesting to compare core materials with with wider or narrower B-H curves to see what more or less hysteresis does sound-effect-wise. Seems like saturation may not be the only effect that causes some people to like the sound of a little iron in the signal path.
150mH for primary shunt inductance strikes me as fairly low. I get higher values from small (25mm) ferrite cores. Some even smaller steel cores (14mm) measure 1.2H, with 600turns.
The 'sound of iron' also seems to apply to ferrite so I doubt we're dealing with sound quality properties of the core materials, rather the in-system benefits of improved rejection of common-mode noise brought about by galvanic isolation.
The 'sound of iron' also seems to apply to ferrite so I doubt we're dealing with sound quality properties of the core materials, rather the in-system benefits of improved rejection of common-mode noise brought about by galvanic isolation.
Guys please read and understand all his article, the test circuits, the component choices and exactly what they are measuring. Please don't post pictures that simply look different and make random conclusions from them.
We could make one with a low open-loop BW just for you, it wouldn't sound any different not that anyone cares.
Do you use it open-loop? Of course not, so why does it matter?
It's a bit noisy compared to some, as well, and not incredibly cheap. Seems like there are at least a dozen better op-amps available for line level.
Last edited:
SOIC schmoik. After cutting my teeth with the VSON headphone amplifier from TI, anything larger now is looking sooo outdated and way below my soldering skills. I'm only doing bumped wafer these days, BTW.
Actually, iron seems to be classified as a ferrite, at least by some authors. However, the term "sound of iron" seems to be a colloquial one that refers to ferrites, including iron.
In any case, transformer core material selection and other core design choices can affect sound quality. Agree that transformers can improve sound quality in some cases by helping to reduce the effects of various sources of electrical noise. But they can also be used as fairly subtle effects in order to create a tad bit of euphonic distortion. Same for ferrite core inductors, as may sometimes used in "vintage" audio filter circuits. Small amounts of intentional ferrite-based transformer distortion are often designed into modern versions of other types of "vintage" audio products, such as mic preamps, line amplifiers, compressors, etc., many of which are used in recording studios to recreate the old sounds of rock and roll, and so forth, which some of us grew up listening to and learned to like.
- Status
- Not open for further replies.