Going by Samuel Groner's measurements, OPA(2)604 is best for use at the higher-than-usual supply voltages it'll withstand (note thermals may be problematic in the dual), but no great shakes at a more standard +/-15 V, let alone even lower voltages. Not a particularly good load driver. These used to be a recommendation for DAC I/V stages, so I guess immunity to RF hash may be better than average - as you might expect for a reasonably speedy JFET input part.
I am fairly certain that there has to be a better 1:1 replacement for LF353. LT1057 maybe? With a pinch of Class A bias if need be?
If OPA627 sounds bad I would wager a guess that sloppy layout is to blame, as the part is quite speedy (e.g. 55 V/µs slew rate). The bootstrapped cascode input gives phenomenally low input impedance distortion, especially for a FET input part, plus good common-mode linearity, and output loading immunity still beats a 5532 at least. Not a bad part to use after a 50-100k volume pot. Granted, for what these cost you might as well bootstrap the supplies of a nominally lesser part...
I am fairly certain that there has to be a better 1:1 replacement for LF353. LT1057 maybe? With a pinch of Class A bias if need be?
If OPA627 sounds bad I would wager a guess that sloppy layout is to blame, as the part is quite speedy (e.g. 55 V/µs slew rate). The bootstrapped cascode input gives phenomenally low input impedance distortion, especially for a FET input part, plus good common-mode linearity, and output loading immunity still beats a 5532 at least. Not a bad part to use after a 50-100k volume pot. Granted, for what these cost you might as well bootstrap the supplies of a nominally lesser part...
It's not quite as simple as all that.
The 47k is the -load- resistance and this a small part of the equation.
If you look here Magnetic Phono Pickup Cartridges
there is an approximation of the self impedance of a MM cartridge.
Essentially we are looking at approx 220mH in series with 1k2.
As such the total noise will be opamps voltage noise + current noise reacting
with cartridges self impedance which will be rising with frequency.
Using a low noise jfet IP opamp will probably net the best results.
Then there is the whole subjective side of the equation
T.
You are right, it's a simplified explanation but the purpose was to explain why current noise is so important in certain cases. I mentioned MM as a source example. In this I followed Douglas Self calculations where he used 47k buffer resistor with a "theoretical typical" MM cartridge. The point was to demonstrate the above rather than complicate things by adding actual source implications. In real world when designing a preamp for a given source (eg MM, MC or a microphone etc) one has to include source characteristics in calcs as well.
I did my calcs for a number of different op-amps in non-inverting configuration with a buffer resistor and gain of 10 and and then I looked at balanced inputs as well. Exercise was basically voltage noise, current noise, gain and source load and then adding input signal because in the end what matters is Signal to Noise ratio. Unfortunately S/N calcs are in practice more complicated than Noise Figure.
NS Application Note 104 (Noise Specs Confusing?), which explains noise issues, gives also a much more detailed description of noise issues with a MM load (frequency dependent) than ESP.
cheers,
Bob was a friend of mine industrially speaking. We exchange ideas and designs for a while. I was impressed by him as he was me. People that actually know something and not just a bunch of simulations on computers.
https://www.youtube.com/watch?v=9WCoIzBHDVg
Maybe not that great?
https://www.youtube.com/watch?v=9WCoIzBHDVg
Maybe not that great?
Going by Samuel Groner's measurements, the OPAx604 clearly likes its supplies a bit higher than usual, supported by Sijosae noting oscillation when attempting to drive 33 ohms on +9V. (A tendency towards instability at low supplies has also been noted for the OPAx134.) So I'd use it when highish levels at highish impedance are demanded, like in a mixer or something - then the decently high slew rate also finds some use. (Which I think is why DAC appnotes of the early '90s like to feature OPAx604 as I/V where immunity to ultrasonic crap counts - it must have been among the fastest-slewing audio parts at the time, plus the better gm nonlinearity of a FET input.) Too bad the dual is getting rather toasty at this point - a supply of DIP heatsinks could come in handy.
As to where the instability might be coming from, my hunch would be input capacitance. We know how it is with FETs, higher Vdg tends to reduce Cob a lot (and associated nonlinearity along the way). Why input impedance distortion would essentially vanish when going from +/-15 to +/-24 V though, I don't know. It's almost like some cascoding kicks in, though I don't think this part has a cascoded input stage. (This feature could make it useful in an oldschool preamp or integrated amp, like after the volume control. Noise isn't too low, but if you split gain before/after it may work out.)
You can probably tell that I don't consider the part a good "general purpose" opamp. I'm pretty sure that a fair bit of the commonly reported reported "character" is due to latent instability, limited output voltage swing (soft limiting seemingly setting in at around 15 Vpp on +/-15 V rails is not exactly too impressive) and/or output loading in common circuits. Quirky little bugger.
I wouldn't knock the "simulator people" too much. Everyone's got their own skill sets. I admire classic "hands-on" engineers, and at the end of the day there's no replacement for the real deal when stuff is supposed to work in real life (especially since the kind of models available to us mere mortals tend to be approximate at best) - but there still is plenty to be learned by other approaches, each having its specific set of advantages and limitations. Yours truly is much better at reading and writing on a computer than most anything else and tends to swear by a well-trained gut feeling (in lieu of being able to treat everything at high level) and a well-stocked collection of documents and bookmarks, so that's what I use. Besides, being good at what you do is one thing, being able to explain that stuff well is quite another matter. The above quote on CMRR seems well-meaning but ultimately not an awful lot clearer than mud to me.
As to where the instability might be coming from, my hunch would be input capacitance. We know how it is with FETs, higher Vdg tends to reduce Cob a lot (and associated nonlinearity along the way). Why input impedance distortion would essentially vanish when going from +/-15 to +/-24 V though, I don't know. It's almost like some cascoding kicks in, though I don't think this part has a cascoded input stage. (This feature could make it useful in an oldschool preamp or integrated amp, like after the volume control. Noise isn't too low, but if you split gain before/after it may work out.)
You can probably tell that I don't consider the part a good "general purpose" opamp. I'm pretty sure that a fair bit of the commonly reported reported "character" is due to latent instability, limited output voltage swing (soft limiting seemingly setting in at around 15 Vpp on +/-15 V rails is not exactly too impressive) and/or output loading in common circuits. Quirky little bugger.
I wouldn't knock the "simulator people" too much. Everyone's got their own skill sets. I admire classic "hands-on" engineers, and at the end of the day there's no replacement for the real deal when stuff is supposed to work in real life (especially since the kind of models available to us mere mortals tend to be approximate at best) - but there still is plenty to be learned by other approaches, each having its specific set of advantages and limitations. Yours truly is much better at reading and writing on a computer than most anything else and tends to swear by a well-trained gut feeling (in lieu of being able to treat everything at high level) and a well-stocked collection of documents and bookmarks, so that's what I use. Besides, being good at what you do is one thing, being able to explain that stuff well is quite another matter. The above quote on CMRR seems well-meaning but ultimately not an awful lot clearer than mud to me.
Last edited:
I'm curious if anyone here has tried the ceramic bodied versions of the Philips NE5532, namely the NE5532FE and NE5532AFE?
Sony used to glue ceramic slabs to the DAC chips on some of their top machines and I wonder what benefits this had? Vibration reduction? Temperature control?
You can see examples of this on the PCM64 DACs used on the 557esd and the TDA1541 DAC used on the 555esd.
SONY CDP-557ESD
sony CDP-555esd
I've read some favourable reports about the sound quality of OP270FZ which also has a ceramic body.
Sony used to glue ceramic slabs to the DAC chips on some of their top machines and I wonder what benefits this had? Vibration reduction? Temperature control?
You can see examples of this on the PCM64 DACs used on the 557esd and the TDA1541 DAC used on the 555esd.
SONY CDP-557ESD
sony CDP-555esd
I've read some favourable reports about the sound quality of OP270FZ which also has a ceramic body.
It is given in the data sheet as either PSRR or kSVR which is a ratio of how much the input offset voltage will change in relation to supply voltage changes.
The 5532 is quoted as 80db minimum and typically 100db.
It should be a non issue because any opamp circuitry should be running on supplies that are clean.
CMRR (common mode rejection ratio) test set ups can be used to derive PSRR but I've never seen it done.
As mentioned above, the supplies should be so clean that the effects of deteriorating CMRR and PSRR become a non issue.
The 5532 is quoted as 80db minimum and typically 100db.
It should be a non issue because any opamp circuitry should be running on supplies that are clean.
CMRR (common mode rejection ratio) test set ups can be used to derive PSRR but I've never seen it done.
As mentioned above, the supplies should be so clean that the effects of deteriorating CMRR and PSRR become a non issue.
Newer opamps like opa2134, which were developed specifically and qualified for premium audio applications, certainly do have noise density and noise V frequency graphs as part of their datasheet info. and I assume that's what you expect to find. Unfortunately there is something like a 20 year gap in technology and engineering there. Don't expect to see any retrospective data published on the 5532 any time soon though. They may still be excellent, cheap and reliable devices but manufacturers are much more likely to spend their time and money on new and more lucrative products.
edit: Thought about the LM4562? - should be data available for that, for the time being at least.
Last edited:
This might give you some ideas, however even here, the formulas relate to static DC levels.
http://www.analog.com/media/en/training-seminars/tutorials/MT-043.pdf
http://www.analog.com/media/en/training-seminars/tutorials/MT-043.pdf
Guys, before putting in the last nails in the coffin of the venerable 5532, please make sure that you've actually got the real thing:
Real vs Fake NE5532 OpAmp ICs – Fake Version is “Bare minimum to operate as opamp” << Adafruit Industries – Makers, hackers, artists, designers and engineers!
Also, I am quite sure that somewhere in the pre-TI monopolistic era there were NE5532 schematic versions with PNP input pairs, but I'll have to trawl back in time to find that. And that is a long way!
Anyway yes, I do think that the 5532's days of glory have probably finally been and gone. In the Rotels I have modified, I found that the RIAA circuit 5532's (up to) 800uA input bias current and/or 30kOhm Rin to be an unhealthy DC load on a DC coupled phono cartridge - given its full output voltage range of 2.5mV.
I guess that thirty years of audio IC experience really does count - I have found that the OPA2134 is a significant improvement over the NE5532 on all fronts.
BTW, I also tried the metal can version of the LM4562 in its place, with +/-18V supplies, and it simply gets too hot to touch? Or have I got fake 4562's ?
Real vs Fake NE5532 OpAmp ICs – Fake Version is “Bare minimum to operate as opamp” << Adafruit Industries – Makers, hackers, artists, designers and engineers!
Also, I am quite sure that somewhere in the pre-TI monopolistic era there were NE5532 schematic versions with PNP input pairs, but I'll have to trawl back in time to find that. And that is a long way!
Anyway yes, I do think that the 5532's days of glory have probably finally been and gone. In the Rotels I have modified, I found that the RIAA circuit 5532's (up to) 800uA input bias current and/or 30kOhm Rin to be an unhealthy DC load on a DC coupled phono cartridge - given its full output voltage range of 2.5mV.
I guess that thirty years of audio IC experience really does count - I have found that the OPA2134 is a significant improvement over the NE5532 on all fronts.
BTW, I also tried the metal can version of the LM4562 in its place, with +/-18V supplies, and it simply gets too hot to touch? Or have I got fake 4562's ?
Well, that is not the point. If you load any small ac generator (here, the phono cartridge) with a standing DC load of (up to) 33% of its full range signal output - the opamp input current noise is the very least of your worries.
You might as well remove any anti-skating compensation efforts - the cartridge distortion will be severely affected. Also, loading the poor thing with 30k//47k (ie. effectively down to 18k) will definitely change the cartridge frequency behaviour.
The NE5532 should only be used with DC capacitor circuit blocks both on the input and on its output, with the colouring/distortion that entails.
Whereas the OPA2134 works just fine in the original DC coupled circuit configuration.
the first PDF schematic (post #5) under
Head Pre for Denon DL 103
shows an interesting approach from Tim de Paravicini with the NE5534 for a RIAA head amp.
Maybe usable also for other applications like line gain stages and active crossover networks.
Head Pre for Denon DL 103
shows an interesting approach from Tim de Paravicini with the NE5534 for a RIAA head amp.
Maybe usable also for other applications like line gain stages and active crossover networks.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.