Yes, a friend of mine in the audio business found the same thing recently (and ordered a bunch of the cheaper p/n). I'll chalk that up to the difficulty of keeping data (even pricing!) correct for 10s of thousands of parts. Just a little over a year ago I was looking at the BUF634 on TI.com and found that the price listed was in the 10's of dollars. I told a friend who works there about it and turns out an extra zero got typed into a spreadsheet or web tool somewhere.
It's definitely an established fact that the two p/n's differ only in marking.
It's definitely an established fact that the two p/n's differ only in marking.
Can you please find out which page that was on? I don't find anything in my D.Self books. It is an interesting fact. By the way, Victor recently switched from LM4562 to OPA1656 because it has even less THD. It also depends on the use case to use it.
What was omitted from the LM4562/LME49720 history is that the part was cancelled, because the 6" wafer line it was sourced from went way, and then resurrected on a new 8" wafer line. The price for the new LM4562 also dropped significantly at the beginning of 2019.
Modern datecode parts (2019 forward?) are in fact from a different wafer fab, and while I have not done any significant tests, I have not seen a new datecode part that has exhibited the annoying popcorn noise problems. So, since popcorn is a fab problem, it is logical to expect that it could be different for older or newer devices.
Modern datecode parts (2019 forward?) are in fact from a different wafer fab, and while I have not done any significant tests, I have not seen a new datecode part that has exhibited the annoying popcorn noise problems. So, since popcorn is a fab problem, it is logical to expect that it could be different for older or newer devices.
Popcorn noise is not a new thing with these parts as I've seen it for years, but would be good to know it's gone. My understanding is that the change-over is not yet completed, but that info is a few months old. Since both p/n's have been through whatever transfer, that doesn't explain the difference in prices.
Ralf, I will have to look at my books this weekend. The same chapter on the merits of various op-amps appears in multiple of Doug's books in slightly different form.
Ralf, I will have to look at my books this weekend. The same chapter on the merits of various op-amps appears in multiple of Doug's books in slightly different form.
There are several hundred LM4562 DIP (UK supplier) that I got up to Oct 2016. I was finding 30-40% to be below-par in terms of burst noise.
I got 1,000 x LME49720 DIP (USA) in Oct 17 (panic buy due to phasing out).
I haven't bought any since.
Over the last couple of days I've been looking at them on a live FFT, leaving batches to soak, powered up for a few minutes before making judgement.
From what I've seen, the 49720s do not have anywhere near the failure rate that the 4562s from 2016 have.
As I said before, unless I made a large test jig and baked a hundred of each type over night, making a time-consuming statistical analysis, I wouldn't stake my life on these claims. But from the batches in the drawers here, the 49720 definitely has lower burst noise QC fails.
Your mileage may vary (it definitely will in this case). I'm not reading anything into the stock code btw. If the ones from the new fab lines cure the issue, it's almost worth trashing the substantial stock I already have, as staring at an FFT for hours at a time really isn't a productive use of one's time. I guess I could program macros on one of the analysers, but I doubt it'd be accurate as you get momentary bursts from people switching irons and stuff on... It's not an easy issue to vet out.
Oh yeah - this IC is definitely not the only modern OA to have popcorn noise. I think you notice it more with this IC as, with the correct source-Z, it's so quiet. I've yet to figure out the exact config to exacerbate the problem. Methinks inverting, medium gain.
I got 1,000 x LME49720 DIP (USA) in Oct 17 (panic buy due to phasing out).
I haven't bought any since.
Over the last couple of days I've been looking at them on a live FFT, leaving batches to soak, powered up for a few minutes before making judgement.
From what I've seen, the 49720s do not have anywhere near the failure rate that the 4562s from 2016 have.
As I said before, unless I made a large test jig and baked a hundred of each type over night, making a time-consuming statistical analysis, I wouldn't stake my life on these claims. But from the batches in the drawers here, the 49720 definitely has lower burst noise QC fails.
Your mileage may vary (it definitely will in this case). I'm not reading anything into the stock code btw. If the ones from the new fab lines cure the issue, it's almost worth trashing the substantial stock I already have, as staring at an FFT for hours at a time really isn't a productive use of one's time. I guess I could program macros on one of the analysers, but I doubt it'd be accurate as you get momentary bursts from people switching irons and stuff on... It's not an easy issue to vet out.
Oh yeah - this IC is definitely not the only modern OA to have popcorn noise. I think you notice it more with this IC as, with the correct source-Z, it's so quiet. I've yet to figure out the exact config to exacerbate the problem. Methinks inverting, medium gain.
Raw date codes themselves may not tell the story. When one decides to pull the plug on one factory and move to another, the wise thing to do is to build ahead so that you have enough stock to supply production in case of the inevitable bumps in the road during the switch-over. If subsequent manufacturing (in this case assembling finished devices from raw wafers) is done at more than one factory (or even if not), it may be that production will use a blend of old and new material. It may not be a binary change where you can count on a date/lot code to be a solid predictor of which is which.
In this case, both 2017 and 2016 material are likely from the old fab. Popcorn noise is definitely varies lot-to-lot. Read the section in Art Kay's book on op-amp noise for background information.
In this case, both 2017 and 2016 material are likely from the old fab. Popcorn noise is definitely varies lot-to-lot. Read the section in Art Kay's book on op-amp noise for background information.
"The Design of Active Crossovers", 2nd edition, page 488, he reports when using the LM4562 as a voltage follower driven from a low source impedance it can be noisy, appearing to demodulate radio stations, due to internal oscillation/heterodyning. Using 100 ohm feedback seems to be a viable fix.
Actual, real life measurements.
This thread has been very interesting, but what are the real differences between these opamps? Are there actually any differences in spite of the datasheet similarities?
So, I decided to find out.
I had a Rotel RA-930AX at hand which I had recently upgraded, reduced the opamp supplies to +/-16.4Vdc and then sequentially put in the dual audio opamps which I happened to have in my inventory into its preamp stage socket: Total amp DC offset and THD
LME49720 (TO-99, Green trace) -1.8/-5.8 mV 0.0026%
LM4562 (DIP, Blue trace) -1,2/+2.3 mV 0.0026%
OPA2134 (DIP, Black trace) +2.4/+0.8 mV 0.0025%
AD8599 (SOIC, Orange trace) +12.9/+8.7 mV 0.0032%
All were left to fully warm up and then I measured the total amp output DC offset, the 1kHz THD by doing a RTA and the opamps' running temperatures.
I also measured the SPL (gain) linearity, and as expected that was straight as a whistle from 2Hz and up for each and all these opamps (Red trace). I didn't measure the upper frequency limits.
Note that all measurements are taken at the output of the entire amp, so this is an actual measure of what to expect if you should decide to change an internal opamp in your amp to something "better". And no, I didn't perform any subjective listening tests to judge the perceived sound quality, so sorry, I can't provide any qualified audiophile journalistic waffle.
Ok, I know the dc offset is a result of each opamp chip's final laser trim and can/will vary. But I guess that it is the overall level that is of interest. Note that the preamp has been direct (DC) coupled to the power stage, ie. no capacitors in the audio path.
The first thing that can be noticed is that there is absolutely no detectable difference between the LME49720 and the LM4562 traces, so the claim that they are identical chips are most probably true.
The OPA2134 is only very marginally better, whereas the AD8599 has a higher noise floor, offset and THD, although not by much. All would probably do absolutely fine in this application - at least from a strict measurement standpoint.
The opamp running temperatures were in the region of 50-55 oC - except the TO-99 metal can which was significantly higher, probably over 70 oC. I burnt my fingertips when removing the can after power down. That could be a closer representation of what the actual inside chip temperature was?
Cheers,
Per
This thread has been very interesting, but what are the real differences between these opamps? Are there actually any differences in spite of the datasheet similarities?
So, I decided to find out.
I had a Rotel RA-930AX at hand which I had recently upgraded, reduced the opamp supplies to +/-16.4Vdc and then sequentially put in the dual audio opamps which I happened to have in my inventory into its preamp stage socket: Total amp DC offset and THD
LME49720 (TO-99, Green trace) -1.8/-5.8 mV 0.0026%
LM4562 (DIP, Blue trace) -1,2/+2.3 mV 0.0026%
OPA2134 (DIP, Black trace) +2.4/+0.8 mV 0.0025%
AD8599 (SOIC, Orange trace) +12.9/+8.7 mV 0.0032%
All were left to fully warm up and then I measured the total amp output DC offset, the 1kHz THD by doing a RTA and the opamps' running temperatures.
I also measured the SPL (gain) linearity, and as expected that was straight as a whistle from 2Hz and up for each and all these opamps (Red trace). I didn't measure the upper frequency limits.
Note that all measurements are taken at the output of the entire amp, so this is an actual measure of what to expect if you should decide to change an internal opamp in your amp to something "better". And no, I didn't perform any subjective listening tests to judge the perceived sound quality, so sorry, I can't provide any qualified audiophile journalistic waffle.
Ok, I know the dc offset is a result of each opamp chip's final laser trim and can/will vary. But I guess that it is the overall level that is of interest. Note that the preamp has been direct (DC) coupled to the power stage, ie. no capacitors in the audio path.
The first thing that can be noticed is that there is absolutely no detectable difference between the LME49720 and the LM4562 traces, so the claim that they are identical chips are most probably true.
The OPA2134 is only very marginally better, whereas the AD8599 has a higher noise floor, offset and THD, although not by much. All would probably do absolutely fine in this application - at least from a strict measurement standpoint.
The opamp running temperatures were in the region of 50-55 oC - except the TO-99 metal can which was significantly higher, probably over 70 oC. I burnt my fingertips when removing the can after power down. That could be a closer representation of what the actual inside chip temperature was?
Cheers,
Per
Attachments
I always wondered why people thought it wouldn't be true. Printing different labels on OpAmps to get higher margins from buyers who still believe in the superiority of anything with "LME" in the name is not that big of a hassle in production, but having different wafers seems completely out of question from a production management standpoint.
That being said, most of the comparisons made here are pretty much worthless from a statistical perspective, due to variances between lots and small sample sizes as has been mentioned plenty before.
Guys, this has been documented multiple times in this and other threads by National Semi / TI employees. the LM4562 and LME49720 are one and the same chip; the only difference is the marking on the package and the orderable part number. National originally created the LM4562 and then decided to create the LMExxxxx p/n's for their high-end audio op-amp family. I suspect both p/n's were kept to avoid confusion with customers, though it surely appears to confuse audio DIYers.
Hi
i am not an expert....
Here we have this discussion again:
https://www.diyaudio.com/forums/class-d/315681-tpa3250-listening-76.html#post5660863
here is a doc were you can find the answer...maybe..
http://www.waynekirkwood.com/images/pdf/US_Patent_7649417_Bias_Current_Cancellation_Audio_Op_Amp.pdf
chris
i am not an expert....
Here we have this discussion again:
https://www.diyaudio.com/forums/class-d/315681-tpa3250-listening-76.html#post5660863
here is a doc were you can find the answer...maybe..
http://www.waynekirkwood.com/images/pdf/US_Patent_7649417_Bias_Current_Cancellation_Audio_Op_Amp.pdf
chris
I have a Tektronix SG-505 Ultra low distortion oscillator, its based on either NE5534 or NE5532 chips can't remember exactly which one, anyway, many years ago when the LME series came out, I thought that if I used an LME49720 or LME49710 instead of the NE553x I would be able to get even lower distortion, I was wrong. I don't know if its due to the higher input noise current of the LME series, or if the LME chips burst into high freq oscillation or what, but the LME series performed worse and the distortion was considerably higher.
The lesson learned is: opamps are not the entire story, the design plays a huge role on the final performance, a good designer with an off the shelf part can do wonders, which is another reason why I don't always believe in opamp rolling to be the best choice.
You just experienced this with the OPA2134, which at least on the datasheet has higher THD than the LM4562, but it exhibits lower THD on your amp, I'm guessing you measured THD+N so the actual improvement might be less noise rather than less distortion, either way a valuable lesson is learned.
Why do you NOT know? Do you not have a scope to put on the LME chips to SEE if there's oscillation? It's advisable to put 100nF bypass caps on each of the op amp's power rails, especially when going to a much higher bandwidth IC such as the LME49720 (55 MHz) replacing an NE5532 (10 MHz). Also recommended if you do suspect oscillation is a 22-100pF (depending on the circuit) across the feedback resistor. The slightly higher current noise of the 49720 should only be an issue in very low-level, high source impedance circuits.
I did not know because basically I didn't bother going any further, I realized it was a "if it ain't broke don't fix it" kind of situation, I didn't want to mod the oscillator, I just wanted to do a drop in replacement and see if it worked, it didn't and I was lazy enough not to pursue any further.
Before I got an AP I used a Boonton 1120 -- same story -- the NE5534 in the oscillator couldn't be beat.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.