• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Hypex NC1200: Quality of the implementations

JJB70

Major Contributor
Forum Donor
Joined
Aug 17, 2018
Messages
2,905
Likes
6,148
Location
Singapore
One of the oldest tricks in consumer product design (except for stuff where lightness is a virtue) is to add weight as many associate heaviness with hight quality. Now sometimes building in high construction quality and durability will add weight, and weight is important for some things, but a lot of the time it is just a marketing trick. For all that I am a sucker for deluxe build and metal case work etc myself.
 

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
Yes They will provide better technical performance than a 5532, although the 5532 isn't the poor performer many seem to think it is.

My general point about this op amp rolling is that rarely provides any genuine benefit. If you have a properly designed circuit swapping out a good op amp for another good op amp will show little measured difference. This has been demonstrated with Amir's measurements and the additional measurements that Maty has extracted from manufacturers and posted in this thread.

Then of course we have some examples here where these swap out op amps are absolute dogs.

Then we have the problem that people hear differences that aren't there because they have compared sighted and without controls. Mind you those bad performers shown earlier may possibly sound different.

As an ex designer of opamps understanding the trade off benefits is not much of a problem, and I am quite circumspect when I make the exchanges.
One issue that I have with the 5534 family is that, and please correct me if I'm wrong, no distortion measurements are provided in the data sheet, although they were partially assessed by Sam Groner in 2009.
Replacing the 5532s with the 1612s did provide measurable performance improvements in distortion- 2-3 orders of magnitude if I remember correctly- and in the circuits that I looked at some other performance improvements- nothing earth shattering, a bit lower noise in an input stage, less 20kHz roll off in high gain stages due to GBW limitations- it also provided me with a bit of amusement.
I was checking the 5534/2 specs and glanced at the On Semi datasheet for the parts and immediately noticed something strange in the equivalent "schematic" they provided for the 54.
https://www.onsemi.com/pub/Collateral/NE5534-D.PDF
The input stage current mirror biasing cannot work.
The 5532 schematic is correct.
https://www.onsemi.com/pub/Collateral/NE5532-D.PDF.
I believe that the discrete opamp that was alluded to earlier is this one,
https://sparkoslabs.com/wp-content/uploads/2018/03/SS3601_SS3602.pdf
The open loop phase characteristic is very interesting to say the least- supporting Maty's observations earlier- and there are comments made on the ability of IC opamps to include various capabilities which are not correct. It's also interesting that the data sheet doesn't include any square wave response plots that are not in a limiting case (slewing, clipping, current limited).
In any case, at least they make some attempt to be professional. I've tried, without success, to find any detailed information on the Bursons besides this "datasheet"
https://drive.google.com/file/d/0Bxn23njCr8VCR19ELS1YOTMxV00/view
Many people claim performance improvements such as "lower noise", "more extended bass", etc. which one would think would be readily derived from specs. However, as far as I know no one has ever actually measured these improvements and there seem to be no ABX tests to be found to justify the statements of subjective improvement.
Sam Groner was less than complimentary about an earlier Burson amp
http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf
Is anyone aware of any technical analysis/teardown of the recent Burson offerings?
 

kn0ppers

Active Member
Joined
Jul 4, 2019
Messages
192
Likes
258
Location
Germany
I am no EE or expert on OpAmps, but can one just replace 5532s with OPA1612s and 5534s with 1611s without worrying about stability? I always thought these things could start oscillating when used in some kind of feedback loop and the implementation not being optimised for the particular OpAmp?

Apart from that I will always remember what Mr. Douglas Self had to say, recording consoles are stuffed with 5532s and 5534s and have been for a while. "Audiophiles" haven't complained about all the recordings being mixed on those consoles using cheap OpAmps.
 

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
As an ex designer of opamps understanding the trade off benefits is not much of a problem, and I am quite circumspect when I make the exchanges.
One issue that I have with the 5534 family is that, and please correct me if I'm wrong, no distortion measurements are provided in the data sheet, although they were partially assessed by Sam Groner in 2009.
Replacing the 5532s with the 1612s did provide measurable performance improvements in distortion- 2-3 orders of magnitude if I remember correctly- and in the circuits that I looked at some other performance improvements- nothing earth shattering, a bit lower noise in an input stage, less 20kHz roll off in high gain stages due to GBW limitations- it also provided me with a bit of amusement.
I was checking the 5534/2 specs and glanced at the On Semi datasheet for the parts and immediately noticed something strange in the equivalent "schematic" they provided for the 54.
https://www.onsemi.com/pub/Collateral/NE5534-D.PDF
The input stage current mirror biasing cannot work.
The 5532 schematic is correct.
https://www.onsemi.com/pub/Collateral/NE5532-D.PDF.
I believe that the discrete opamp that was alluded to earlier is this one,
https://sparkoslabs.com/wp-content/uploads/2018/03/SS3601_SS3602.pdf
The open loop phase characteristic is very interesting to say the least- supporting Maty's observations earlier- and there are comments made on the ability of IC opamps to include various capabilities which are not correct. It's also interesting that the data sheet doesn't include any square wave response plots that are not in a limiting case (slewing, clipping, current limited).
In any case, at least they make some attempt to be professional. I've tried, without success, to find any detailed information on the Bursons besides this "datasheet"
https://drive.google.com/file/d/0Bxn23njCr8VCR19ELS1YOTMxV00/view
Many people claim performance improvements such as "lower noise", "more extended bass", etc. which one would think would be readily derived from specs. However, as far as I know no one has ever actually measured these improvements and there seem to be no ABX tests to be found to justify the statements of subjective improvement.
Sam Groner was less than complimentary about an earlier Burson amp
http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf
Is anyone aware of any technical analysis/teardown of the recent Burson offerings?

Well this is the distortion of a 5532 from a well known source. Gen distortion underneath. Not too shabby. Yes there is better, and I agreed with you above.

Also as noted directly above, much recording equipment is stuffed full of the things :)

Screenshot_20190707_202154.jpg
 
Last edited:

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
I am no EE or expert on OpAmps, but can one just replace 5532s with OPA1612s and 5534s with 1611s without worrying about stability? I always thought these things could start oscillating when used in some kind of feedback loop and the implementation not being optimised for the particular OpAmp?

Apart from that I will always remember what Mr. Douglas Self had to say, recording consoles are stuffed with 5532s and 5534s and have been for a while. "Audiophiles" haven't complained about all the recordings being mixed on those consoles using cheap OpAmps.

In general, no you just can't drop an opamp in and expect to have no consequences.
However, that's not what I do. My general procedure is to construct the schematic, where possible, and to simulate the circuit with the old and new devices, and where not possible to assess the operation paying particular attention to possible oscillation.
Understanding the data sheet and example circuits helps to assess whether a given circuit is likely to be problematic.
As an example, I evaluated the read back circuit of a friend's Studer A810 which includes a number of NE5532s, and possibly 5534s- I can't remember. He provided the schematic. On simulation it seemed that there was a potential latch up problem with the head preamp- and we discovered that Studer had modified the circuit in a later revision to correct the problem. Simulating the full circuit showed that there were frequency response issues with the preamp at HF that were corrected using the calibration procedure. Replacing the NE5532s in simulation essentially removed the frequency response problems and slightly improved the noise, but did not show any other pathologies. Subsequently the read amp was reworked with new audiophile caps and 1612s without problem.
The owner seemed convinced that it was better sounding. I felt no compulsion to challenge this- it was an interesting experience and a good use of my time.
If you examine the data sheet for the 1612 you will see that the application circuits are pretty minimal and unlike the AD797 the level of concern that is needed to ensure stability is small. Of course, good layout practices and decoupling caps close to the supply pins are de rigeur, as are small value caps across the feedback Rs under certain circumstances, but these are fairly standard requirements and in most cases are already done in the circuits.
The A820 and A810 use the same circuitry. The Otari tape decks from that era also use opamps that are comparable to the NE5534 family as did most professional audio gear. So, Douglas Self had a good point. Heck I even designed CLX , PEQ and other equalizers for pro audio using the NE5534 family and they sounded just fine to me then.
Knowing this doesn't stop me from some degree of ridiculous self flagellation for my purchase of an AHB2 amp as Bruno Pudseys' new amp seems to have a smidgen less IMD.
 
Last edited:

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
Well this is the distortion of a 5532 from a well known source. Gen distortion underneath. Not too shabby. Yes there is better, and I agreed with you above.

Also as noted directly above, much recording equipment is stuffed full of the things :)

View attachment 29057
What were the test conditions?
 

kn0ppers

Active Member
Joined
Jul 4, 2019
Messages
192
Likes
258
Location
Germany
In general, no you just can't drop an opamp in and expect to have no consequences.
However, that's not what I do. My general procedure is to construct the schematic, where possible, and to simulate the circuit with the old and new devices, and where not possible to assess the operation paying particular attention to possible oscillation.

Yes that's kind of what I expected when you said you work with OpAmps a lot professionally. As mechanical engineers, we would take the same route: simulate it extensively, prototype it, measure it in the real world, make adaptations if necessary and repeat until the requirements are met or the option is deemed unfit for the application. Do you use LTSpice or something else? I am not that proficient in the electrical engineering world.

Edit: I wanted to add that I think 5532s etc. are perfectly fine in all things HiFi and Pro-Audio. Only when it comes to the "High-End" stuff, you would also expect better OpAmps if there is a notable, measurable improvement by using them at certain points in the circuit. On the other hand, I think it was in the chapter about Baxadall active volume controls in Self's Small Signal Design Book where there was an example of using multiple NE5532s in a parallel configuration to reduce noise. Considering how cheap these things are, if space isn't constrained, you can just use lots of them. So depending on the use case, multiple 5532s could still be just as good and cheaper. But there are few cases where space-constraints aren't an issue nowadays. And OPA1612s are really good.
 
Last edited:

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
Yes that's kind of what I expected when you said you work with OpAmps a lot professionally. As mechanical engineers, we would take the same route: simulate it extensively, prototype it, measure it in the real world, make adaptations if necessary and repeat until the requirements are met or the option is deemed unfit for the application. Do you use LTSpice or something else? I am not that proficient in the electrical engineering world.

Edit: I wanted to add that I think 5532s etc. are perfectly fine in all things HiFi and Pro-Audio. Only when it comes to the "High-End" stuff, you would also expect better OpAmps if there is a notable, measurable improvement by using them at certain points in the circuit. On the other hand, I think it was in the chapter about Baxadall active volume controls in Self's Small Signal Design Book where there was an example of using multiple NE5532s in a parallel configuration to reduce noise. Considering how cheap these things are, if space isn't constrained, you can just use lots of them. So depending on the use case, multiple 5532s could still be just as good and cheaper. But there are few cases where space-constraints aren't an issue nowadays. And OPA1612s are really good.
Correct me if I'm wrong, but putting opamps directly in parallel seems to me to not be a good idea for several reasons. It's not a good idea if you want to DC couple stages as the amplifier output stages will act against one another to try to produce whatever offset the individual amps require. Also, the equivalent input voltage noise may be reduced but the input noise current will be increased, as will the net input bias current, so operating the stage with the same resistances/impedances in the inputs and feedback paths may be problematic, etc. etc.
These issues can, of course, be overcome, but with a significant increase in complexity.
It seems better to just use an amp with improved specs.
Yes, I use LTspice now that I have to pay for my simulation environment and I'm cheap.
It works quite well, once you get to understand it's foibles, and I can use it for tube and SS designs.
 

kn0ppers

Active Member
Joined
Jul 4, 2019
Messages
192
Likes
258
Location
Germany
Correct me if I'm wrong, but putting opamps directly in parallel seems to me to not be a good idea for several reasons.

No, I just said parallel for lack of a better way to describe it. You can see the example in his 2012 Preamplifier Circuit for Elektor.

Edit: I will try to be more precise with terminology in the future, I know how important it is and I am always annoyed by people who don't do so. But I am still very much a newbie to this part of engineering. I hope the quite extensive class on Electronics that I am planning to take next semester will change that, after all it's the class the EEs at my University also take in their fourth semester.
 
Last edited:

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
No, I just said parallel for lack of a better way to describe it. You can see the example in his 2012 Preamplifier Circuit for Elektor.

Edit: I will try to be more precise with terminology in the future, I know how important it is and I am always annoyed by people who don't do so. But I am still very much a newbie to this part of engineering. I hope the quite extensive class on Electronics that I am planning to take next semester will change that, after all it's the class the EEs at my University also take in their fourth semester.
Yes, the circuit that he designed addresses my objections, and with a significant increase in complexity.
 

kn0ppers

Active Member
Joined
Jul 4, 2019
Messages
192
Likes
258
Location
Germany
Well I never stated it as being a drop in replacement for existing circuits or not being more complex. I just said in some applications it could be a feasible option.

But I think we are digressing ;) How did we get here anyway?
 

pirad

Active Member
Joined
Apr 22, 2018
Messages
178
Likes
61
I have conducted auditory tests with several buffers/opamps including the top/expensive SIL.
None of them performed better with hypex NC500 than the standard 30$ hypex converter.
Coloring of sound is one thing, a matter of taste. Resolution is not- either you can hear something
clearly or not. Most of the "designer" buffers robbed NC500 of resolution.
 

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
I have conducted auditory tests with several buffers/opamps including the top/expensive SIL.
None of them performed better with hypex NC500 than the standard 30$ hypex converter.
Coloring of sound is one thing, a matter of taste. Resolution is not- either you can hear something
clearly or not. Most of the "designer" buffers robbed NC500 of resolution.

Interesting, if anecdotal ;) you know what I'm going to ask, did you do the test blind?

One of the strengths of the Hyoex amps is that they are so neutral, so to my mind it sort of defeats the object to colour the sound deliberately. If you want tone controls use variable tone controls to suit the sound of any particular recording. Don't colour everything with a non transparent buffer.

I think the lesson is to ask the vendor to show measurements of their buffer before purchase.
 

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
So basically maximum 0,0005% in the audible band, about -106dB.
Unfortunately, without knowing the source R, gain, load R, supply voltage and output level, the distortion plot is not particularly useful in making a direct comparison against, say, the OPA1612 from a purely technical perspective.
 

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
Unfortunately, without knowing the source R, gain, load R, supply voltage and output level, the distortion plot is not particularly useful in making a direct comparison against, say, the OPA1612 from a purely technical perspective.
I will try and dig up the the details. For info the data is from Mr Self. I don't think there is any doubt the 1612 will be better.

Edit: rin is 1k, vs +-18v vout is 5v rms. Not sure about rload.
 
Last edited:

pirad

Active Member
Joined
Apr 22, 2018
Messages
178
Likes
61
Interesting, if anecdotal ;) you know what I'm going to ask, did you do the test blind?

One of the strengths of the Hyoex amps is that they are so neutral, so to my mind it sort of defeats the object to colour the sound deliberately. If you want tone controls use variable tone controls to suit the sound of any particular recording. Don't colour everything with a non transparent buffer.

I think the lesson is to ask the vendor to show measurements of their buffer before purchase.
Anecdotal, but can be validated by others. I strongly encourage this simple test. Normally I would engage a couple of my usual suspect musicians and audio engineers and test blind. With buffers I never got to that stage. The underwhelming performances of designer buffers are so obvious that even my mediocre ears can tell. I am afraid that the vendor measurements are not enough. You must hear it to believe it.
 

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
I will try and dig up the the details. For info the data is from Mr Self. I don't think there is any doubt the 1612 will be better.

Edit: rin is 1k, vs +-18v vout is 5v rms. Not sure about rload.
Thanks for the lead. I searched and found this.
https://www.eetimes.com/document.asp?doc_id=1278866#
I still have to digest the test results, but they do appear to be pretty impressive.
One can only wonder why Signetics neglected to include distortion in their datasheet.
 

wynpalmer

Active Member
Technical Expert
Joined
Dec 14, 2018
Messages
175
Likes
214
Thanks for the lead. I searched and found this.
https://www.eetimes.com/document.asp?doc_id=1278866#
I still have to digest the test results, but they do appear to be pretty impressive.
One can only wonder why Signetics neglected to include distortion in their datasheet.

Distortion measurements for the 5532 are taken from the eetimes article.

The best comparison is at 20kHz- otherwise the "distortion" measurements are mostly overwhelmed by the noise.
The worst case is in the voltage follower mode, which causes the opamps to modulate their input device base current and produces distortion.
I believe that to a first order this is linearly dependent on voltage, so the result should be more or less independent of amplitude.
Also, the best comparison is with RS=0 and RL=2k for the 1612 as the NE5532 tests are all with comparable or higher load R.
The 1612 is also measured at 3v RMS rather than 5, but the effect of this is unclear as for the 1612 performance the SINAD is still largely due to noise, not distortion, and for example, a small improvement occurs at 1kHz as the output is increased. So, I'll assume that this is a wash.
The OPA1612 also has a moderately complex pole-zero compensation scheme. Around 10MHz the response changes from a single pole to two poles, then a compensating zero is added to bring up the phase and ensure stable operation at unity gain.
This means that for a gain of 1 or 2 the amount of feedback that is applied is about 12dB greater than that of the NE5532 which has a unity gain bandwidth of 10MHz, although no detailed information is available in the 1980 era datasheet. However, I believe that the compensation is single pole.
At gains >= about 10 the effective GBW product of the 1612 increases to about 80MHz, providing even more feedback.
Thus, if we have identical open loop equivalent input distortion in the two parts we would expect 12dB-18dB lower output distortion for the 1612 under the same gain, frequency etc. conditions.

So, what's the result?

Eyeballing it, at 20kHz, gain of +1, load 2k, 0 Rs, the 1612 has a distortion of about -131dB.
The NE5532 has a distortion, under similar conditions, of 0.0014% or about -97dB- a difference of c. 34dB.
Taking into consideration the differences in open loop gain, it suggests that the actual open loop distortion of the 1612 is about an order of magnitude lower than that of the NE5532, and about 2 when the additional GBW is taken into account.
This is a very cursory look. If I've made any errors please let me know.
Wyn
 
Top Bottom