Might we say TI has done enough on audio opamp development and products to slow down on development. Lots of choices in the opa portfolio. Pretty hard to tell the difference with your ears when parts are such high performance. Crap I liked the sound of a ne5534a back in the late 70’s. It’s almost like we should have a switch on the preamp to select what opamp we want use, your perception might change with the phase of the moon. Maybe let AI determine which one I am in the mood for 
OPA 1656s have been used successfully as headphone amplifiers with load impedances as low as 100Ω. So they won't drive an 8Ω speaker, but how low of a load do you need? OPA 1656s are still King AFAIK; their only drawbacks are the cost of a PDIP-to-SOIC adapter if needed, and the need to observe careful layout and bypass capacitors to avoid oscillation.
NE5534s are still EXCELLENT opamps and are about 1/3rd the price of an LM4562.
NE5534s are still EXCELLENT opamps and are about 1/3rd the price of an LM4562.
It's surprising how good the JRC 4558 is. I replaced some 4558s in the post DAC filter section of a Pioneer CD player - the one that Max Townshend used to modify and sell for £X thousand and managed to ruin it. IIRC I put OPA2604s in, even though it is an op amp I don't tend to like. It seemed like a nice conservative choice and was the "go to" op amp for Naim at the time (maybe still is) but in this instance it drained the life out of the CD player (which is what I think it does in the cheaper Naim CD players too). For me the 2604 is as dull as ditchwater. I don't know the difference between the TLO82 and the 72s but as a fun, peppy, if a little ragged, amp, I love the TLO72. If I were doing something cheap that I wanted people to really enjoy, it's what I'd use. I expect these days I could probably do an implementation that even added a little bit of refinement. I'd be very interested to know what you thought of the LM833Ns as I've got a couple of dozen originals lying around. Also, what you thought of the 4562, if you can remember.
I've gotten excellent results from OPA2134 in DIP-8 packages. They're pretty spendy, however.
As a hobbiest, its easy to over spend on a device, but as a designer, or one selling a audio product, one has to wonder if the cost of the opamp is so important?
esp. when folks say "TL072 & OPA2134 are very nice operational amplifier" and these parts have sonic signatures.
Looking at the datasheets, you can't really differentiate specds from sound signature.
It does depend on where the parts are used of course, like in a MM phono, knowing that you want low current and voltage noise. A NE5534A wins vs TL071
TI trys to have a no obsolescence policy, even if that means your LM3886 can have very long lead times.
Compare using a fet opamp, opa1656 vs a opa2134, I see little reason why one would choose a opa2134 for a new design, maybe that's why TI asks so muchbecause they want you to use the new tech, but it's probably the cost of the mfg process for that tech vs the new processes.
Each part has its specific specs tailored for the application. Enjoy your opamp rolling
esp. when folks say "TL072 & OPA2134 are very nice operational amplifier" and these parts have sonic signatures.
Looking at the datasheets, you can't really differentiate specds from sound signature.
It does depend on where the parts are used of course, like in a MM phono, knowing that you want low current and voltage noise. A NE5534A wins vs TL071
TI trys to have a no obsolescence policy, even if that means your LM3886 can have very long lead times.
Compare using a fet opamp, opa1656 vs a opa2134, I see little reason why one would choose a opa2134 for a new design, maybe that's why TI asks so much
because they want you to use the new tech, but it's probably the cost of the mfg process for that tech vs the new processes.Each part has its specific specs tailored for the application. Enjoy your opamp rolling
Definitely but then why do op-amp rolling? Why use op-amp like AD797 which is designed for submarine sonar or such state of the art stuff? For audio frequencies venerable 553X or opa213X are just fine & can do for most of the part. I mean there is a difference in sonic apart from datasheet spec & maybe that's why some people still enjoy filaments instead of solid state amplifiers.
Over here in the UK my favourite op amp, the LM6172, has gone from under £2 to over £5 in the last 10 years, most of it recently. If they're not discontinuing all the best parts (the BC560 is apparently now no longer manufactured! Uh? It was the sweet spot of the range.) they're hiking the prices to the heavens.
I've never tried the OPA2134 because Roy George from Naim once told me that it was one of the op amps he didn't like. Naim aren't right about everything, of course, but they are usually on the same side of the audio divide as I am - ie. being in it for enjoying the music rather than analysing it.
re. opa1656 vs a opa2134, I see little reason why one would choose a opa2134 for a new design,
I'm not so impressed with the look of the OPA1656s noise curve (I think it was that one I was looking at - one of the 165x anyway). Its 1/f corner is above 1kHz. I haven't done the calculation on it specifically but I did do some calcs on some newer Fet op amps a few weeks ago. Though 1/f noise isn't usually significant, if you were doing a Low Pass crossover filter at 2kHz then that would probably be making up the majority of the noise. This rather defeats the point of using a Fet, which was to be able to use larger values of resistors (and smaller, cheaper capacitors) without much penalty. I was also concerned with what some of them might do in a phono stage, with them encouraging you to have your High Pass higher than you otherwise might.
I'd have to do the calculations to be sure I'm not on red herring here - and if someone wants me to compare two op amps then I probably have time - but it is something that was beginning to concern me. And one reason to be concerned is mfrs are shifting around where they place the scale so that the noise curves look OK when in fact they aren't.
I'm not so impressed with the look of the OPA1656s noise curve (I think it was that one I was looking at - one of the 165x anyway). Its 1/f corner is above 1kHz. I haven't done the calculation on it specifically but I did do some calcs on some newer Fet op amps a few weeks ago. Though 1/f noise isn't usually significant, if you were doing a Low Pass crossover filter at 2kHz then that would probably be making up the majority of the noise. This rather defeats the point of using a Fet, which was to be able to use larger values of resistors (and smaller, cheaper capacitors) without much penalty. I was also concerned with what some of them might do in a phono stage, with them encouraging you to have your High Pass higher than you otherwise might.
I'd have to do the calculations to be sure I'm not on red herring here - and if someone wants me to compare two op amps then I probably have time - but it is something that was beginning to concern me. And one reason to be concerned is mfrs are shifting around where they place the scale so that the noise curves look OK when in fact they aren't.
That TLE072 is an interesting option---MUCH faster than the original 072. And a bit quieter, though still too high for any high-gain applications. Also has rather poor CMRR and THD specs.
A LOT of people like the sound 0f 2134s but, as they are ~$6 each now, they ain't worth it if ya gotta have bunches if 'em.
A LOT of people like the sound 0f 2134s but, as they are ~$6 each now, they ain't worth it if ya gotta have bunches if 'em.
When Texas Instrumenst says "FET input" what they really mean is "MOSFET input" and these transistors are infamously noisy.
True JFET input amplifiers are always marketed as such and the modern ones have impressively low 1/f noise - see ADA4625 for one of the best, or OPA164x for a more affordable option which still almost matches old precision bipolar devices like OP07.
In many practical applications the LF noise of the OPA1656 is hardly a problem. 10nV/rtHz at 100Hz is equivalent to the thermal noise of about 7 kOhms so something like a line-level low-pass Sallen-Key filter with small precision caps (NPO ceramic) and large resistors (like >20K source resistance as seen by the chip) is not compromised... and distortion performance will be stellar in this application, compared to many JFET input opamps.
I like LM6172 a lot. I like how very little of its own character it adds to the sound. It worked nicely as a buffer stage, with AD811 as I/V.
The OPA2134 is probably my worst-liked OPAmp... so muddy, soft, mellow... it could be ideal if the rest of the electronics (power supply mostly...)/rest of the system is bad.
Another one I used a lot to replace typical gain stages in many devices was AD8066. But this one had to be used together with lowering the Vcc/Vee to max 12V (which was a good thing because it meant upgrading the overall quality of voltage rails' regulation as well...)
No sane person designs products with THT parts these days (read: the last two decades), that's why they are phased out. SMT variant BC850/860 is readily available from several manufacturers and will be for many years to come.
You mean THT semiconductors/passives. Many sane reasons to use THT for mechanical robustness for connectors, switches, heavy parts etc. However there's a strong economic reason to avoid THT on a PCB if possible as it will be significantly cheaper without another expensive THT manufacturing step. However there's a mid-ground, pin-and-paste (aka through-hole-reflow), where parts have steel pin(s) that fits a through hole (often not protruding completely though), but which are processed by the standard SMT pick&place method. This can give much of the mechanical robustness of through hole without a separate manufacturing step - this is commonly used for micro-USB sockets and similar components with high mechanical stresses.
For most semiconductors THT has no advantage and was becoming obsolete long ago (1980's onwards) in most electronics. In fact SMT chips with thermal pads on the underside can often perform better thermally than larger THT packages with heatsinks - and these days there are aluminium cored PCBs which means SMT is way better thermally (and which prohibit any through hole components).
For most semiconductors THT has no advantage and was becoming obsolete long ago (1980's onwards) in most electronics. In fact SMT chips with thermal pads on the underside can often perform better thermally than larger THT packages with heatsinks - and these days there are aluminium cored PCBs which means SMT is way better thermally (and which prohibit any through hole components).
Funnily enough this came through as I was looking at BC850/860s on Farnell and Mouser. You say "readily available" but out of the 18 that Farnell comes up with for the 860 only 5 are in stock and, of those, only one allows you to buy in 100s. You can't get the On Semi ones in less than 12,000 so you have to go for the Nexperia ones. I love Philips components in general (and wish their MKP caps were still available, especially because of the baby blue colour
) but it's always a leap in the dark trying a new component, especially when the noise figure is blanked for Typ. It's not as bad here as one of the other datasheets I saw this morning where both Min. and Typ. were blanked and the Max. NF was 10dB! It's a similar story on the 850 but I don't have that page to hand. And same story, mostly, at Mouser - though they now seem to be doing a better job than the neverending decline at Farnell.Whether it's sane or not, we should be objecting to the phasing out of through hole components because it's not just an attack on DIYers but is a deliberate barrier to entry for small companies. Most prototyping is done on Veroboard or in dead bug form and, if you want to change components in a simple A/B test, it is far easier to replace a TH transistor than it is an SMD. Getting a PCB drawn up and having a dozen made with matched components might be quick in a large company but it's still a helluva lot more work than switching your soldering iron on spending 10 minutes changing a few parts. And thru hole is also much less prone to mistakes and bad joints. Seriously, we should view this as big business bullying. I wonder how many modern day JLHes there are out there who are just put off from even bothering to try out their ideas because it's all a bit too tiresome and they don't want to make the transition. Also, as a final point and example of big business favouring their own interests, a lot of useful parts get EOLed in the transition. Where can we find low noise matched transistor arrays these days? Big business can put these onto one of their SoCs but we can't do the equivalent and are forced either to buy their solution or to go without.