I think they are bang on the same GBWP of 55MHz. Slew rates might split them a little, but not so as it matters, I shouldn't think.
But I've just explained that's not the case if your bandwidth is narrow enough. Perhaps you'd like to do the sub 1kHz or 2kHz calculations and see if it really does come out better? I doubt it does, though, as I said, I haven't done the calculations definitively and just suspect it's not going to cut it. This flicker noise is starting nearly two decades above where it normally begins.
I think the 1656 1/f noise is too high and the knee is also way too high at 2 kHz for comfort. That I believe is related to its use of mosfet input and amplifying devices on the chip. The OPA1641/2 uses JFET inputs so you get much lower 1/f, good thermal noise voltage specs with negligible noise current. I’ve gone over to the 1641/2 as my go to general purpose opamp for audio work. It really is a superb opamp in its updated guise. The LM4562 and derivatives seem to have been plagued with noise issues - lot of complaints about it. I’ve ever only used them in line level applications or where I was not relying too much on their noise performance. That said, I have never had a noise problem with them. They make awesome low current (5-10 mA) regulators because their PSRR is superb even at 100 kHz.
I was extremely pleased with the audio performance of a composite amplifier, made of an LT1122 and an LM7171. It's here on the Forums, pcb named Milpitas (link). Like other composite amplifiers, Milpitas has a single feedback loop enclosing both chips, so any distortion or noise introduced by the video speed output opamp, is error-corrected away by the low distortion front end chip.
Aha! I first saw this about 10 hours ago and it's taken until now for me to twig what you're suggesting. LOL. Sorry to be so dim. I even spent a part of those hours thinking "What's he bringing up one of his old projects for?". 🤣 🤔 🤣 This is an extremely good idea - Don't get rid of the op amp you like, just make it better. I have to say it's a bit bloody expensive but I like the Fet inputs and the LT being good with 40V on the rails. (I wonder how much it will go over 40V?) There's a lot of mileage here, once I get over paying a tenner for a 13MHz op amp.
I don't suppose it sounded very much like anything, but would you say your pre- ended up sounding more like the 7171 or did the LT impose the most character? I haven't done much of one op amp controlling another, or a chip amp, in real life, but I've played around with it endlessly in simulation. In general I'd have thought the controller amp would be the one imposing its own sound (like a 4562 around a 3886) but for some reason I seem to feel that in this instance the LM7171 (which I imagine is indistinguishable from a 6171) might prevail.
Incidentally, why did you not bandwidth limit any of this? I'm not in front of your schematic but I don't remember any capacitors in the amplifying part. There's a helluva lot of noise due to just that huge b/w that you could snip off with a capacitor in the right place. And maybe defining the b/w of the system might help it always behave - he says, slightly hopefully. 🙂
hi please excuse me if i jump in But at least from sim the lt1122 seems quite good even by itself
I like the old LF353 , this old freind is unconditionally stable ,fast and delivers lower then 0.01 % thd at gains under 20. It's output is also much stronger then tl072 op amps.
4562 runs hot opa1642 is surface mount only.
4562 runs hot opa1642 is surface mount only.
I like the LF353 too, but it doesn't quite meet the definition of the latest and greatest in cutting edge tech. Meridian used it in the 101 preamp. It's quite a complicated application in general (why use a push pull o/p on a pre with its own Vbe multiplier) and sadly has about half the musical info of a Naim12S or 32S. Yes, the power supply is a lot weedier, but is that the reason why? Interestingly, neither company has changed their general preferences for transistors with Meridian still featuring 2N440x and Naim sticking to the perhaps over-zingy ZTXs. Unfortunately I can no longer look into this in depth as my 101/105 combo was stolen and sold.
Is it not a bit too noisy for you? I think we're in a new paradigm when it comes to noise and we don't have to put up with those very mediocre figures and narrow bandwidths. Acknowledging that there are some problems with the 1/f noise on the OPA16xx fet i/p amps, we're mostly in a whole different class and getting for 50p what we used to really have to shell out for. Is anyone seriously interested in a 20MHz op amp with 10nV/rt Hz noise these days? I thought not.
Hi thank you for your very kind and helpful advice I understand i have to study more before posting I apologize
I am trying to sim something and i have realized that sim with opamps can be more demanding that with discrete designs
I have to start easy Opamps will follow I am not ready yet to face the challenge
Thanks a lot again
I am trying to sim something and i have realized that sim with opamps can be more demanding that with discrete designs
I have to start easy Opamps will follow I am not ready yet to face the challenge
Thanks a lot again
I recommend calculating the signal-to-noise ratio of a linestage (signal level 1.0VRMS) using an opamp whose input noise voltage density is "X nanovolts per root Hertz" and which is operated at unity gain.
Then plug in various values of X and see how the signal-to-noise ratio is affected. Traditionally, SNR calculations use an audio bandwidth of 22 kHz , half the sampling rate of 16-bit, 44 kHz, digital compact discs.
Then plug in various values of X and see how the signal-to-noise ratio is affected. Traditionally, SNR calculations use an audio bandwidth of 22 kHz , half the sampling rate of 16-bit, 44 kHz, digital compact discs.