NE5534 replacement - National LME 49710?

[Kingston]

More often than not, "remarkable" sonic differences are very apparent in basic bench testing.

Of course they are. In this test rig of mine nearly always they can be seen in a plain 1k sine to FFT display set up, or in IMD tests (RMAA software). Differences are most often apparent in the higher harmonics combined "shape". Not much difference will be seen between 2-3th harmonics (sometimes they are even non-existent), but usually above that. A mess in the "grain" or "better imaging" area after 4th harmonic. Sometimes noisefloor changes.

It's only several times I've bumped into roughly identical harmonics shapes looking at the FFT plots. I'm fully aware that's not the whole story. I could next check how square wave edges look, for example, but most of the time the difference in sound is already so apparent I could not be bothered. I can already pick the better one. If not, it's not worth the effort.

5 stages of opamps in series, all doing different tasks gets easily comparable results quickly.

By the way, I even selected a certain favourite opamp to be used in RME DA converter output stages this way. I already knew there was a measurable (lower THD) and audible difference to the cheapo crap used there because I had run their equivalents in the mixer channel strip. I could only estimate what exactly they had done with those opamps at the output - buffer & differential conversion for each channel probably - but swapped them anyway. I easily repeated the results of the mixer in there: lower THD (if only marginal) and surprisingly lower noise floor and of course more transparent sound. Easily audible comparing two stereo outputs side by side.

I tend to trust this selection method, even though the measurements and comparisons are unorthodox.

[millzners]

Hook up a simple sound card with signal generator and metering to measure basic frequency response.. More often than not, "remarkable" sonic differences are very apparent in basic bench testing.  I have measured many differences that I couldn't hear, but never heard a difference that I couldn't measure,

I'm trying to keep my promise here, as I finally acquired a Hakko for SMD soldering this week.  I swapped all the opamps out of the 1st channel for AD797 (dual servo), LM4974 (quad in insert circuit), and LM6172 (output).

The question is, how is the best way for me to measure the difference between channel 1 and 2?  I have a good scope, and as far as I can remember the only way to use that and the computer signal generator to calculate frequency response is to sweep the frequency and measure the amplitude on the scope along the way.  Are there more sophisticated ways of doing this w/out sending the signal back into the computer?

And, when using the scope for these measurements, are you just sending a line out into the 50 ohm BNC of the scope, or are you using something like a pi network to match impedances to ensure the source is happy?  I can't imagine the output opamp in this case cares much but for transformer based outputs I'm not sure.

[JohnRoberts]

The scope probe is generally a very high impedance so will not load down whatever you look at.

Yes, frequency response is that simple,,, amplitude in vs amplitude out 20-20kHz.

For comparisons between channels of a multichannel device, I like null testing. If you subtract the modified channel from an unmodified channel, any deviation from a full null, is difference between the two channels. If the channels are very similar this may reveal parts tolerance differences, but it could reveal improved phase response, distortion, whatever...

If the channels can be sent to a bus to sum, simple invert the input tp one of the two channels.

JR

[envelope]

sometimes I change OpAmps just to reduce overall current in a current cramped design, or for better noise specs, or slew rate, as long as they are perfectly suitable in the electronics of the application. E.G. OP275 in place of NE5532. (if the new OpAmp is more to your 'taste' all the better) For subjective analysis, test for yourself and ignore every and all 'opinions', put down the scope (momentarily) and prick up your Production Ears,. (again, make sure it is compatible, and change other components as necessary)

[abbey road d enfer]

If the channels are very similar this may reveal parts tolerance differences, but it could reveal improved phase response, distortion, whatever...

I don't get that. Let's say one channel is close to perfect, minimal phase deviation, -100dB THD, VLN, and the other has gross phase shift, high THD and very noisy; doing the difference test will tell you there's a difference, but it won't tell you which one best. You may say that listening to individual channels will tell.
Now if one has no phase-shift but high THD and the other large phase-shift but low THD, what do you conclude from the null test?
For me, null test is just a way to prove/disprove a claimed sonic difference.

[millzners]

The scope probe is generally a very high impedance so will not load down whatever you look at.

Yes, frequency response is that simple,,, amplitude in vs amplitude out 20-20kHz.

For comparisons between channels of a multichannel device, I like null testing. If you subtract the modified channel from an unmodified channel, any deviation from a full null, is difference between the two channels. If the channels are very similar this may reveal parts tolerance differences, but it could reveal improved phase response, distortion, whatever...

If the channels can be sent to a bus to sum, simple invert the input tp one of the two channels.

JR

I do something very similar when I listen to the differences, but it never occured to me to do that with a scope and measure the differences... Great idea.

If the channels are very similar this may reveal parts tolerance differences, but it could reveal improved phase response, distortion, whatever...

I don't get that. Let's say one channel is close to perfect, minimal phase deviation, -100dB THD, VLN, and the other has gross phase shift, high THD and very noisy; doing the difference test will tell you there's a difference, but it won't tell you which one best. You may say that listening to individual channels will tell.
Now if one has no phase-shift but high THD and the other large phase-shift but low THD, what do you conclude from the null test?
For me, null test is just a way to prove/disprove a claimed sonic difference.

Also true.  As I said above I run into the same issue when null listening.  Sure, maybe I'm left with a signal that's from 2kHz to 10kHz and really quiet...  but often that's  such a minor shift I have trouble locating that difference in the original signals.  At least I know there's a difference, but I still struggle to make a decision.

It sounds like I should be looking to AudioTester or some such thing to really try to measure these things... 

[mata_haze]

SOrry to bring up sych an old topic, but I recently found this website:

http://bursonaudio.com/burson_opamp.htm

is it just speculations?
I mean,let's say I have an old console, is it not worth replacing the output opamps with something like the above?

Mattia.

[JohnRoberts]

Smells like hokum...  The output opamp is NOT like the lens on a projector (more correct analogy is like a link in a chain or perhaps a shackle connecting two chains).

I couldn't wade through the hard sell to look at the actual opamp if they do share any actual data in with the marketing spiel.

JR

[Harpo]

I'd better not trust an advertisment without any data except rail voltages, where this Burson will be a direct replacement for a JRC5534 Dual Op-amp.

[mata_haze]

thank you guys for confirming what I thought.
Also if you look at the "5 minutes replacement opamp" the guy is actually putting a lot of stress on the main PCB while inserting the replacement op-amp.
not good practice... 

Best,
Mattia.

[Andy Peters]

SOrry to bring up sych an old topic, but I recently found this website:

http://bursonaudio.com/burson_opamp.htm

Wow, lots of bullsh*t there. Tread carefully.

-a

[Nishmaster]

This part is the best:

"100 hours of burn-in, then enjoy the music!"

Because, clearly, anyone can make 100% objective comparisons on audio between two sittings separated by a minimum of 100 hours.

I thought the prices were cute, too.

-Matt

[Biasrocks]

I must say, I just replaced the 5532's/34's in my MCI with the LME parts and man did it come to life.

Those Burson's look like an accident waiting to happen, plus they're physically huge!

Regards,
Mark

[Kingston]

I must say, I just replaced the 5532's/34's in my MCI with the LME parts and man did it come to life.

Careful there, this rabbit hole goes deep!

[Biasrocks]

Careful there, this rabbit hole goes deep!

Agreed.

Thanks for turning me on to the LME parts Kingston.

My experiment was on my JH-24 which I had previously upgraded with OP176/OP275 about 5 years ago which I felt was a nice improvement over the stock TI 553* parts that were in there.

The thing that I found most satisfying was the 553* and OP* parts were imparting a lot of colouration and the LME's totally took away the nasty compressed sound the older parts had. Currently, the JH-24 is a mixture of OP*'s and now the LME's in select places; LME's are on the playback side. Resulting in a very nice sounding multi-track. 

I have before upgrade and after multi-tracks that I saved for comparisons sake and the difference is anything but subtle.

The LME parts like good power, so additional bypassing may be in order, the GBW is 5x's greater than the 553*'s, so care is required to avoid nasty surprises.

Regards,
Mark

[Brian Roth]

One other possible "gotcha" about the National LME part is the "abs-max" +/- 18 VDC rail limitation.  IIRC, a 553x chip had a +/- 22 VDC "abs-max" rail spec. 

Best,

Bri

[Kingston]

One other possible "gotcha" about the National LME part is the "abs-max" +/- 18 VDC rail limitation.  IIRC, a 553x chip had a +/- 22 VDC "abs-max" rail spec. 

Best,

Bri

Which part? LME49860 does 22VDC rails. Absolute maximum even higher.

LM4562 is the older version with equal specs before they improved the manufacturing process. The only difference was the supply voltage range. Little birdie told me the two are today actually the same part, just sold under two different names.

[edit]

holy crap the price has come down on LME49860! I remember paying more than 5 euros a pop at digikey. It's more than halved now!

[dagoose]

One other possible "gotcha" about the National LME part is the "abs-max" +/- 18 VDC rail limitation.  IIRC, a 553x chip had a +/- 22 VDC "abs-max" rail spec. 

Best,

Bri

Which part? LME49860 does 22VDC rails. Absolute maximum even higher.

LM4562 is the older version with equal specs before they improved the manufacturing process. The only difference was the supply voltage range. Little birdie told me the two are today actually the same part, just sold under two different names.

[edit]

holy crap the price has come down on LME49860! I remember paying more than 5 euros a pop at digikey. It's more than halved now!

was just checking them out at RS.. 9,28 each... ouch...

[Kingston]

was just checking them out at RS.. 9,28 each... ouch...

Never ever buy opamps from RS or farnell. The prices are just wrong.

[mata_haze]

was just checking them out at RS.. 9,28 each... ouch...

Never ever buy opamps from RS or farnell. The prices are just wrong.

do you always get yours from DIGIKEY?

Mattia.