LME49720 Vs LM4562....

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The amplifiers you mention have lower input voltage noise, but when their input current noise is taken into account, when working into typical moving magnet cartridge source impedances, the total noise of the phono amp circuit is not necessarily lower.

Any amplifier has a combination of input voltage noise and input current noise, and both affect the total noise that the circuit will produce. The typical tradeoff with a bipolar input stage design is that to get less input voltage noise, you need to increase the standing current of the input pair, which increases the amplifier's input current noise.

So, while the 5534 has higher voltage noise, it also has lower input current noise, and the RMS sum of the voltage noise of the input stage, the input current noise interacting with the source impedance of the phono cartridge, and the Johnson noise of the source resistances is what's important. The input voltage noise performance alone is not the whole story - you can't just ignore the resulting high input current noise.

It is the combination of these three noise sources that matter, and an amplifier with a higher input voltage noise can very well result in less total noise than an amplifier having much less input voltage noise and higher input current noise, because that input current noise interacts with the decidedly nonzero source impedance of the phono cartridge.
I would agree completely here if a microphone preamp was to be designed - go to
https://www.analog.com/media/en/tec...-Microphone-Design-Considerations-MS-2472.pdf
and
http://emerald.tufts.edu/programs/mma/mrap/ES65/ES65_Lecture2_091318.pdf

But because the record surface noise has significantly higher values compared to operational amplifier, even under favorable conditions concerning kind of stylus and the record itself - go to
Reducing Record Surface Noise - I want to know all Approaches
should a lowest possible signal-to-noise ratio from the RIAA preamplifier itself only be an unimportant feature in order to reach the best sonic quality while record playing (a 709 or 741 isn't a good choice in order to get good values for SNR, but in real live this parameter is unaudible while record playing).

Otherwise there is a risk that excellent signal-to-noise ratio values of the electronic RIAA devices will be at the expense of the sound quality at whole.

To make long story short, extremly low noise design and Op-Amp parts for RIAA preamp applications is unimportant.
Only the sound quality should be always the most important thing.

For phono stages NE5534 is the first choise. No other op amp has lesser noise.
OK, Cool. My first choise for Phono are however AD797 or LME49990.
In order to the sound quality the AD797 outperform the NE5534 clearly (tested in a one stage preamp with RIAA network in the NFB loope).
I am looking for a dual version of AD797 for replace the NE5532. (The AD797 uses intern only one voltage gain stage).
Thank you for advices.
 
Last edited:
I would agree completely here if a microphone preamp was to be designed - go to
https://www.analog.com/media/en/tec...-Microphone-Design-Considerations-MS-2472.pdf
and
http://emerald.tufts.edu/programs/mma/mrap/ES65/ES65_Lecture2_091318.pdf

But because the record surface noise has significantly higher values compared to operational amplifier, even under favorable conditions concerning kind of stylus and the record itself - go to
Reducing Record Surface Noise - I want to know all Approaches
should a lowest possible signal-to-noise ratio from the RIAA preamplifier itself only be an unimportant feature in order to reach the best sonic quality while record playing (a 709 or 741 isn't a good choice in order to get good values for SNR, but in real live this parameter is unaudible while record playing).

Otherwise there is a risk that excellent signal-to-noise ratio values of the electronic RIAA devices will be at the expense of the sound quality at whole.

To make long story short, extremly low noise design and Op-Amp parts for RIAA preamp applications is unimportant.
Only the sound quality should be always the most important thing.



In order to the sound quality the AD797 outperform the NE5534 clearly (tested in a one stage preamp with RIAA network in the NFB loope).
I am looking for a dual version of AD797 for replace the NE5532. (The AD797 uses intern only one voltage gain stage).
Thank you for advices.

Hmmm, must agree about surface noise..and all the other noise contributed by turntable and record..at the end that will probably be what you hear. With this point of view, anything around 60dB or so is enough, better specs considered useless. But, also can’t help to think that with that kind of approach…any kind of dick over two inches would be considered useless also…;););)
 
My feeling is that it takes a lot of effort to decide just how poor you can tolerate your amplifiers, and it's sometimes difficult to 'pessimize' them optimally to achieve the right amount of 'bad'. Why not have nice amplifiers with low noise? These days, we have noise reduction algorithms, and all noise is not the same. So, if you're transcribing a disc to be processed later, who's to say that the preamp noise is the same as the record surface noise?

Again, you might not be able to hear the noise or whatever metric you care about, but around here, we're not racing to the bottom with $5 MP3 players either. You're free to take whatever tradeoff you like, but who's to say what the surface noise of a record is and whether it's important when discussing an amplifier design. Isn't that apples and oranges with a lot of conversion factors between the two worlds?

Amplifier noise relative to a specific source impedance is a nice clean topic that's rather simply engineered, and there's no need to complicate it with record surface noise as transduced by a phonograph cartridge and then applied to some form of audibility decisions. Let's just answer the simple question of which amplifiers are suitable for various source impedances and then move along.
 
Yes, that is true. I compared the above mentioned products with a Shure V-15 phono cartridge (3.5mV nominal output, 1.35KΩ source impedance). All had lower noise than an NE5534 with that source, counting both voltage AND current noise. Current noise generally becomes dominant in source impedances ABOVE ~3KΩ; below that voltage noise is more of a contributor. That is precisely why high impedance sources (guitar pickup) will be quieter with a FET-type opamp than the bi-polar ones mentioned.

The 1K35Ω source impedance of a V15 cartridge is actually its source resistance, not source impedance. So, this is the right value to use to calculate the Johnson noise of a source, since only the real component contributes to Johnson noise. However, it is not appropriate to use when evaluating the noise developed by the amplifier's input current noise. In that case, you need to look at the voltage developed across the entire source _impedance_ because of the input noise current. Because the source impedance includes an inductance (500 millihenries), the total impedance will be higher than the DCR of the cartridge. Similarly, the cartridge's load capacitance will reduce the effective source impedance, so it's not a cut and dry issue - the resulting source impedance will be a function of frequency.

I ran a very simple LTspice simulation using a 500mH inductor and a 1K35Ω resistor, and the series combination has an impedance of 3K42Ω at 1kHz, and 31.44kΩ at 10kHz, so the V15 cartridge's inductance is a significant contributor to its effective source impedance, well into the >3kΩ region that you mention above.
 
The 1K35Ω source impedance of a V15 cartridge is actually its source resistance, not source impedance. So, this is the right value to use to calculate the Johnson noise of a source, since only the real component contributes to Johnson noise.
There is another snag: As source impedance rises towards the high frequencies, it "gives way" to input resistance thermal noise. And unlike what you may expect, higher input resistances gives lower noise - so essentially the input R is sending noise current through the source impedance.
(One of the EEVBlog microphone tutorial videos taught me that. They have a very similar problem in condenser microphones, which are a capacitive source impedance. Hence why they are using bias resistors in the Gigohm range.)

So there would be a meaningful improvement in input noise to be had up there... I got almost 6 dB @ 10 kHz in simulation when going from 47k || 150p to 220k || 50p (source Z: 430R + 440 mH, values for an old AT71). Mind you, the i_n of a 47k resistor is only 0.58 pA/sqrt(Hz) or about NE5534 territory, and for a 220k it's 0.27 pA/sqrt(Hz), so you'd need a low-noise FET input or at least a 5534A to exploit this, plus I don't know how relevant this actually is given the nature of RIAA EQ and it being down 20 dB at 10 kHz. I did see ~17 nV/sqrt(Hz) effective input noise at 10 kHz with 47k||150p, up from 2.7 at <1 kHz, so that would be close to white noise still.

You could be pushing the envelope there a bit with an in-plinth phonopre and possibly some custom tonearm wiring to allow for cable capacitance bootstrapping (assuming you can find out how to make this work... I was messing around in sim for a while, and while it seemed easy enough to bootstrap actual amplifier input capacitance, doing the same for cable capacitance seems much harder).
 
Member
Joined 2012
Paid Member
lme49720 metal can for sale anyone?

Re: asuslover
I've got 7 metal can LME49720's that is for sale if your'e still interested.


Re: NE5532:
On the discussion about NE5532/4 as a RIAA phono opamp, you definitely need to consider the circuit topology.

I upgrade Rotels (see "Improve a Rotel amp THD by 20dB!") and Stan Curtis' original phono amp circuit design is DC coupled to the cartridge. If you use a NE5532 with an input resistance anywhere down to 30kOhm, you will load the cartridge by 18kOhm - not anywhere near the nominal 47kOhm.

Even worse, the BJT input offset and bias currents in the NE5532 can be up into the uA range. I have for a good many years worked with very low output transducers and that sort of standing current loading is not acceptable. You may as well run the turntable without any anti-skating!
I always replace the NE5532 in a Rotel's phono circuit with a FET OPA2134 or OPA2604.

Arguing for faster opamps in a phono circuit is an overkill.

The fastest slew rate a needle in a vinyl LP track can provide is in the region of 2-3 V/us, I seem to remember an old IEEE recommendation of 6 V/us as plenty for vinyl audio hi-fi use.

Finally, noise. I as others have argued, I simply haven't come across any opamps that would come anywhere near the total surface noise of a vinyl record, be they FET, BJT or counterfeit.

I fully agree with sgrossklass' comments and the suggestion of putting the preamp as close to the tone arm and cartridge as possible. Actually, I implemented an old Analog Devices' circuit idea using AD624 Instrumentation amps in my Beogram 3000 and that really did work extremely well.

(I can't find the link to AD's application note, but I think the design stems from W. Jung: "Topology considerations for RIAA phono preamplifiers", 67th Audio Engineering Society Convention, Nov. 1982.)

Maybe someone else has the link?
 
Again the urban legend (fake news?) surfaces. This has been dispelled by people employed by NS and TI -- back then and currently. The LM4562 and LME49720 are IDENTICAL except for the ordering number and marking. The LM4562 was introduced first. Then National Semi, for marketing reasons, decided to create a new part number. My suspicion is that they needed more 'space' in the part number universe for the family of audio-oriented parts and came up with the LMExxxxx numbering scheme. The most difference you are likely to find is lot-to-lot variations, and I expect that is not what's going on here.
I certainly hope so. I just checked prices at Mouser---the LM4562 is $1.64; the LME49720 is $2.68. I wonder why this is?
 
I certainly hope so. I just checked prices at Mouser---the LM4562 is $1.64; the LME49720 is $2.68. I wonder why this is?

This changed recently - the LM4562 price dropped significantly, but some months back, the two prices were on par with each other.

From an old (1-2 year old, pre-tariff) spreadsheet of mine, the LME49720MAX/NOPB was $1.10 each in 1000 quantity, and the LM4562MAX/NOPB was $1.16 in 1000 quantity from the TI.com store.

Today, the LM4562MAX/NOPB is $0.55 each in 1000 quantity, and the LME49720MAX/NOPB is $1.10 each in 1000 quantity from TI's store.

I can only guess, but either it was a marketing decision to favor one part number over the other, or the LM4562 are now from the new 8" wafer process instead of the older 6" process, so they may be cheaper to manufacture.

NOTE: this is just a wild guess, but what is true is that the LM4562 price has dropped significantly recently, and the parts I ordered recently have a 2019 datecode, so TI is not shipping old stock and doing a closeout or anything like that.

I say enjoy the new low price - it really is a nice amplifier, and at this price, the price justification for using something like an NE5532 grows weaker.
 
Last edited:
Well, that's it, then---I am officially declaring the LM4562, overall, as the best opamp in the world for the buck. I'm ordering a hundred of them at $1.11 each. Nothing else even comes close!!

Well the NE5532 is half that price still and less than half the current noise. Its optimal at about 7k ohms for noise, whereas the LM4562 is optimal for 1.5k. People always forget current noise, yet its often the dominant source of noise in real circuits...

The future is low voltage rail-to-rail CMOS opamps I think, for instance the AD8656 which has good voltage noise(*) and of course no current noise to speak of, pretty good distortion and excellent output current. Its a great choice for 5V systems where conventional opamps have no headroom at all. Its been a while since I searched for the best all-round 5V opamp, so there might be something even better now, but this dual opamp seems damn good for DC precision and audio applications, can drive headphones directly too.


(*) 1/f noise is still not brilliant in CMOS but its getting there.
 
Well the NE5532 is half that price still and less than half the current noise. Its optimal at about 7k ohms for noise, whereas the LM4562 is optimal for 1.5k. People always forget current noise, yet its often the dominant source of noise in real circuits...
Yes, the noise of the NE5532 and the LM4562 is equal at 3KΩ source impedance; below that, the 4562 is better, above that, the 5534 is quieter. BUT the 4562 still has MUCH less (-36db) THD, is faster (~2x), and better CMR (-20db). In PDIP form, 100 qty is 66¢; so about 2/3rds the price.
 
If we take it as a given that the 4562 and 49720 are identical internally (my R+S and AP analysers would tend to suggest such), can anyone quantify if there is a QC difference between the two?

Having tested a handful of each, my initial feeling is that the 4562 (both parts in DIP form) is more prone to the LF burst noise issue. But this is just a hunch at this point, and I wouldn't assert this as fact unless I'd managed to leave a hundred of each part on bake overnight and then audited them via live FFT the next day.

TIA.
 
Again? This has been clarified MANY times by people who have been involved with the products at NS/TI. The original part was dubbed 'LM4562'. Then National Semi decided to redo their part numbering scheme, given their high-performance audio op-amps names starting with 'LME' and with 5-digit numbers. I don't know why, other than marketing reasons, but they had been using 'LM' (which stands for Linear Monolithic) since the 70's (maybe late 60's) and were probably running out of 'empty space' in their numbering scheme.

So they re-dubbed the LM4562 as the LME49720, keeping the old p/n because it's a massive pain for customers to deal with a 'change', even if in substance it's only a paperwork change.

NS/TI folks, including those who were directly involved, have verified that both chips have the same die, go through the same manufacturing and test flow, and the only difference is the p/n marking on the package. One of these people has indicated that he has 'heard differences', but that had to do with parts in TO-99 cans vs the same parts in one of the plastic packages. Again, nothing different in the raw die.

If you are seeing some differences, it is likely attributable to differences between wafer lots. Doug Self does point out in one of his books (Small Signal circuits book?) that he has see some oscillation issues when these parts are used as unity gain buffers and the fix he found was to replace the wired connection between output and the inverting input with a 100 Ohm resistor.

So what you're seeing may be the same thing that Doug comments on. Keep in mind that these are wide-bandwidth parts -- 25 years ago something with 50MHz GBW would have been considered a 'high speed' op-amp and not a 'mere' audio op-amp. Thus high-frequency layout and bypassing practices should be followed in order to avoid stability problems - i.e., other oscillation issues not directly related to whatever phenomenon that Doug reports.
 
Thank you, Brian.

From my perspective, I have found that (in agreement with the thread on the TI forum started by Victor) approx 30% of the 4562 have the LF burst noise issue. The manifestation is usually in the same region, but it varies in terms of frequency of 'burst' and also time spent powered up (hence the concept of leaving them to bake overnight and testing next day - as I have been disappointed to see parts test fine initially, and then go noisy later on). I was asking simply to ascertain if both part No.s are made on the same fab line, and if anyone has seen a difference in QC (i.e. with regards to the burst issue).

If the 4562 and 49720 are from the same fab lines, but with different part numbers, then TI's policy of charging twice as much for the LME part makes zero sense.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.