Hypex pre-amp stage upgrade

[kukynas]

this thread is about pre-amp stage of the Hypex N-core amps so anyone interested can chime in and contribute or ask, this is purely technically oriented thread so don't ask about amp sound or comparison between class D vs AB or DBT or what ever else OT already mentioned in other threads (including those closed down)

this also covers MP range which in case of rebuild requires rework station but nothing what hopeful diyer shouldn't have at his disposal  

everything discussed here is available to anyone interested, it has been found out by third party op-amp producer and till now (as far as I can see) hidden behind the DIYaudio in over 2 years old thread

I wouldn't ask Allan if he is aware or even care about it but coz he's member here why not ask his POV

 

so what's going on, for those not familiar with Hypex pre-amps (input buffer) also called OEM eval board or OEM adapter in their original design is flawed, some might ask who cares all designers with certain care will use their own design anyway right? you might be lucky and buy an amp with this sorted out...

what's going on 2, said pre-amp stage is as noisy as my grandma and drastically reduce amp performance, now we can go back to the previous sentence and you might not care because your amp might contain unique p-stage without this problem but here comes the aha moment because this is also directly related to MP range which contains said pre-amp stage so all potential DIYers prepared to buy their MP range of cheap plate amplifiers are affected as well,  Dave you might be aware but just in case 

Art should be aware by now (if last post in said thread about buffer is yours) as well if he went through the entire thread on diyaudio but just in case 

 

so what's going on 3:

below problematic section of the stage, 2x47k series resistor with another 2.2M resistor grounded

 

MP range of the amps: besides that they added noisy NE5532 opamp instead of LM4562 and MP range has slightly lowered gain via 1.8K resistor instead of 1.5K resistor used in NC500 above (R11)

 

 

how to fix it mentioned below, I'm posting this specifically for potential diyers who are willing to take the risk and rework it themselves (I did it with my previous 2xNC500 and plan to do the same with NC252MP once I get my hands on one

 

https://www.diyaudio.com/forums/vendor-s-bazaar/281361-hypex-ncore-nc500-build-45.html#post4655884

 

so basically 47K resistors are replaced by 1.2K resistors and 2.2M is replaced by 1K resistor, few caps added to add filtering, they aren't required but essential

 

the end result even without the scope is clearly audible, distortion I and many others were complaining about is gone 

 

all required is available on Mouser, Farnell, RS or any other smd shop for few bucks and of course rework station with regulated heat gun and sthin nozzle, I went with Vishay thin film resistor in 1206 package such as this  https://eu.mouser.com/ProductDetail/Vishay-Dale/TNPW12061K20BEEN?qs=sGAEpiMZZMu61qfTUdNhGzXW%2Fw%2BNPaau%2F3%2F9oTgINyw%3D

 

@March Audio@Art Vandelay@davewantsmoore

 

 

 

 

[Art Vandelay]

1 hour ago, kukynas said:

this thread is about pre-amp stage of the Hypex N-core amps so anyone interested can chime in and contribute or ask, this is purely technically oriented thread so don't ask about amp sound or comparison between class D vs AB or DBT or what ever else OT already mentioned in other threads (including those closed down)

this also covers MP range which in case of rebuild requires rework station but nothing what hopeful diyer shouldn't have at his disposal  

everything discussed here is available to anyone interested, it has been found out by third party op-amp producer and till now (as far as I can see) hidden behind the DIYaudio in over 2 years old thread

I wouldn't ask Allan if he is aware or even care about it but coz he's member here why not ask his POV

 

so what's going on, for those not familiar with Hypex pre-amps (input buffer) also called OEM eval board or OEM adapter in their original design is flawed, some might ask who cares all designers with certain care will use their own design anyway right? you might be lucky and buy an amp with this sorted out...

what's going on 2, said pre-amp stage is as noisy as my grandma and drastically reduce amp performance, now we can go back to the previous sentence and you might not care because your amp might contain unique p-stage without this problem but here comes the aha moment because this is also directly related to MP range which contains said pre-amp stage so all potential DIYers prepared to buy their MP range of cheap plate amplifiers are affected as well,  Dave you might be aware but just in case 

Art should be aware by now (if last post in said thread about buffer is yours) as well if he went through the entire thread on diyaudio but just in case 

 

so what's going on 3:

below problematic section of the stage, 2x47k series resistor with another 2.2M resistor grounded

 

MP range of the amps: besides that they added noisy NE5532 opamp instead of LM4562 and MP range has slightly lowered gain via 1.8K resistor instead of 1.5K resistor used in NC500 above (R11)

 

 

how to fix it mentioned below, I'm posting this specifically for potential diyers who are willing to take the risk and rework it themselves (I did it with my previous 2xNC500 and plan to do the same with NC252MP once I get my hands on one

 

https://www.diyaudio.com/forums/vendor-s-bazaar/281361-hypex-ncore-nc500-build-45.html#post4655884

 

so basically 47K resistors are replaced by 1.2K resistors and 2.2M is replaced by 1K resistor, few caps added to add filtering, they aren't required but essential

 

the end result even without the scope is clearly audible, distortion I and many others were complaining about is gone 

 

all required is available on Mouser, Farnell, RS or any other smd shop for few bucks and of course rework station with regulated heat gun and sthin nozzle, I went with Vishay thin film resistor in 1206 package such as this  https://eu.mouser.com/ProductDetail/Vishay-Dale/TNPW12061K20BEEN?qs=sGAEpiMZZMu61qfTUdNhGzXW%2Fw%2BNPaau%2F3%2F9oTgINyw%3D

 

@March Audio@Art Vandelay@davewantsmoore

 

 

 

 

 

I haven't actually read the entire (long) thread in DIY so thanks for bringing this to our attention. 

 

As stated previously and on DIY A, I intend to drive the modules directly from a low impedance preamp (without input buffer) so the changed gain distribution should also help to lower the noise floor.  

 

As I stated earlier, the noise figure of a NC Hypex amp will not be determined by the module itself, and that's something you have very clearly demonstrated.  

 

 

 

[davewantsmoore]

3 hours ago, kukynas said:

https://www.diyaudio.com/forums/vendor-s-bazaar/281361-hypex-ncore-nc500-build-45.html#post4655884

Oh yeah... I remember seeing that thread, wow.

 

Quote

so basically 47K resistors are replaced by 1.2K resistors and 2.2M is replaced by 1K resistor

 

 

The red text in this image:   http://www.sonicimagerylabs.com/misc_stuff/UltimateFix_4NIQ.jpg

 

Concerning the resistors is in general, wrong.  Lowering the resistance will increase the noise, not decrease it.   (The circuit picture posted is, of course, incomplete.... and looking at the complete circuit will make more sense).

 

If these circuit changes help their opamp, then great (I'm surprised, but) .... but the justification for them in the stock circuit is flawed (the resistors), and a stretch (the capacitors).

 

Quote

Dave you might be aware but just in case 

I'll be interested to know if you get an "audible hiss" in yours.   I don't.

 

The way I see it, the thread you are looking at is about:

 

• Fixing an issue (an audible hiss) that I don't have
• The measurements they took of their device don't correspond to what other people report
• The logic they use for the design of their "fix" (the lowering of the resistors to remove the "resistor noise", is completely flawed.

 

That being said, they may need such a circuit change(s) to accommodate their own opamp.

 

 

I shouldn't say much more.  I'm not an electronics design expert.

Edited December 3, 2019 by davewantsmoore

[oohms]

I'm not familiar with hypex amps, but couldn't you bypass that stage by changing J15 and J16?

[aussievintage]

1 hour ago, Art Vandelay said:

so basically 47K resistors are replaced by 1.2K resistors and 2.2M is replaced by 1K resistor,

didn't we have a discussion in another thread complaining about input impedances being too low?  This results in 2.2k input impedance does it not?

[davewantsmoore]

24 minutes ago, aussievintage said:

didn't we have a discussion in another thread complaining about input impedances being too low?

The input impedance of the naked modules are low.   Given they don't know how customer will use them, then it's the right thing for them to do.

 

The circuit you're looking at here, is a "input stage" (preamp, buffer, whatever).   That you can put infront of the module to get more closer to "consumer level" and a high impedance.

[March Audio]

Firstly I haven't read the diyaudio thread yet, so if I miss something I apologise. 

 

Thing is the stock circuit isn't actually noisy or flawed and you should be wary of the comments of "a third party op amp designer", who may well have vested interests in selling "upgraded" input boards for Hypex products.

 

We have the measurements and for the Nc252mp the overall amps output noise (max unweighted) is 30uV.  Maybe not the quietest power amp ever, but you have to remember it's market positioning. SNR is 121dB

 

The nc1200 is typically 20uV unweighted. It's snr is 128dB Nothing wrong with this at all. 

 

I would have to do the calculations but I suspect you are only going to make a small difference with this mod. 

 

I will dig out some Texas Instruments info on the noise and affect of input resistors.  Yes they do make a difference but you have to make a trade off between benign input impedance and lowest noise. It's a judgement call for the designer between most compatible and lowest noise. 

 

 

Edited December 4, 2019 by March Audio

[March Audio]

http://www.ti.com/lit/pdf/slva043

 

General info from an OPA1612 datasheets section 8.2

 

http://www.ti.com/lit/gpn/opa1612

 

 

Edited December 4, 2019 by March Audio

[aussievintage]

24 minutes ago, davewantsmoore said:

The input impedance of the naked modules are low.   Given they don't know how customer will use them, then it's the right thing for them to do.

 

The circuit you're looking at here, is a "input stage" (preamp, buffer, whatever).   That you can put infront of the module to get more closer to "consumer level" and a high impedance.

I know that Dave, but this mod gives this preamp/buffer a low input impedance - sort of defeating some of it's purpose.

[Art Vandelay]

57 minutes ago, oohms said:

I'm not familiar with hypex amps, but couldn't you bypass that stage by changing J15 and J16?

 

Hypex NC modules require 11.5V rms and coupled with a low input impedance it's beyond a single op-amp's capability. That's why the balanced drive configuration is favoured.  

 

 

 

  

 

Above comes from the NC500 application notes. 

 

1 hour ago, aussievintage said:

didn't we have a discussion in another thread complaining about input impedances being too low?  This results in 2.2k input impedance does it not?

 

It's too low for most op-amps at the required voltage swing, yes. 

 

In order to drive this Hypex "buffer" for best noise figure you will actually need to use a real buffer. 

 

[davewantsmoore]

I see what you mean now, sorry.   Depends on what you want to drive it with.

[Art Vandelay]

The above discussion is for academic purposes only. 

 

In the real world the 47k input impedance will be fine, and the noise floor is likely to be defined upstream anyway. 

[Art Vandelay]

3 hours ago, davewantsmoore said:

Concerning the resistors is in general, wrong.  Lowering the resistance will increase the noise, not decrease it.   (The circuit picture posted is, of course, incomplete.... and looking at the complete circuit will make more sense).

 

Lowering the resistance increases current noise. What's being discussed here is thermal noise which will be dominant. However, Daniel mentions (audible) distortion which doesn't really make sense in the context of the cct.

 

I found this comment which may shed some light.... 

 

2.5.1 Audio input

Do not drive the input with fully floating sources, be it electrically floating ones like line driver IC’s intended for driving XLR outputs or transformers. Using a floating source will always result in a common mode component that will exceed the common mode input range and will manifest itself as gross distortion. Make sure to set the outputs of your distortion analyser to grounded, not floating.

 

I'll converse with the guys on DIYA when I have some spare time. 

 

 

[March Audio]

@kukynas

 

Have you actually measured properly the difference after the mods to your NC500?

 

To hear distortion frankly it has to be gross in level.  The performance of the amps is measured and known and this does not show itself.

Edited December 4, 2019 by March Audio

[Art Vandelay]

 

Daniel might not have been testing with a dc coupled gen, and with those resistor values selected the op-amps would not have sufficient bias current. 

 

Should be no 2M2 resistor as per the app notes...

 

 

Also, the source impedance will hopefully define input impedance, so we need to know if the noise test was done with the inputs terminated or not.  

[March Audio]

I havent used the NC500 and its evaluation buffer as its not part of my product range, but the MP modules are.  I have not been able to create a circumstance where by gross distortion results.  This suggests the 2m2 isnt a problem and CM isnt going out of bounds.  

 

I have a lot of these out in the wild now and I would have thought that a customer would have got back to me by now about overtly audible distortion.

 

Source impedance.  Generally I would expect any source to be low impedance.  SS pre might be typically 100 ohms, a DAC may be lower.  Tube, welll that might be a different case.

Edited December 4, 2019 by March Audio

[davewantsmoore]

4 minutes ago, Art Vandelay said:

What's being discussed here is thermal noise which will be dominant.

... but if you include the blocking capacitor (not shown in circuit posted here) then reducing the R lessens the filtering out of this noise.   There was a spice of this posted in the DIYA thread.

 

.... and even if it DID work the way they're saying (spice says it doesn't, but) .... then like you noted, it's not going to be "audible distortion", or cause measurements of the sort they charted (which were pretty nasty looking).

 

If it were all as simple as Hypex changing a few resistor values to fix a "fault" with the circuit, then ....  (?!)

 

 

The capacitors suggested are going to be waay specific to the active device.... and the circuit shown is for a different opamp to the ones used in the datasheet circuit for the OEM modules, or the shipped versions of the MP modules.    So doing any of this to the MP plate amplifiers is unwarranted.

[Art Vandelay]

1 minute ago, davewantsmoore said:

 

.... and even if it DID work the way they're saying (spice says it doesn't, but) .... then like you noted, it's not going to be "audible distortion", or cause measurements of the sort they charted (which were pretty nasty looking).

 

If it were all as simple as Hypex changing a few resistor values to fix a "fault" with the circuit, then ....  (?!)

 

Looking at the noise plots, the difference can't be explained by thermal noise. Even if the input is un-terminated you're not going to see anywhere near a 30dB improvement.

 

Something else at play, maybe related to switching noise from the Hypex SMPS. 

 

 

 

[March Audio]

Just took a look at the thread.

 

Firstly I see BAVMIKE  (Mike Mivera) was involved.  I rolled my eyes when I saw this.  He has been banned from practically every audio forum.  He has no technical knowledge.  My first reaction at audible hiss is its something Mike has done wrong.

 

You are simply not going to get 30dB improvement from changing the resistors.  The point re the feedback caps is valid, however you would need to see how that op amp behaves when plugged into the NC500 module to see if there is any peaking or instability or oscillation. You cant test that in isolation.  As mentioned the caps will be specific to the op amp in question, the final bandwidth you want and the load.

 

Lastly it should be pointed out that the NC500 board that Hypex supply is just a simplistic "get you going" board for testing and evaluation purposes, its not meant to go into final products.

 

 

Edited December 4, 2019 by March Audio

[davewantsmoore]

Just now, March Audio said:

Firstly I see BAVMIKE  (Mike Mivera) was involved.  I rolled my eyes when I saw this. 

I have quite a bit of time for people with "no technical" ... but not people like that 

 

Just now, March Audio said:

Lastly it should be pointed out that the NC500 board that Hypex supply is just a simplistic "get you going" board for testing and evaluation purposes, its not really meant to go into final products.

Yes, it's interesting.  They seem to infer that there is no ESD/RFI parts on the eval board.   That's strange as hypex clearly say they're needed.

[March Audio]

4 minutes ago, davewantsmoore said:

I have quite a bit of time for people with "no technical" ... but not people like that 

 

Yes, it's interesting.  They seem to infer that there is no ESD/RFI parts on the eval board.   That's strange as hypex clearly say they're needed.

Yeah Mikes a technical liability.

 

There isnt, the NC500 eval board is not a finished product.  Its just literally to get you going.  Its for use by designers that will use it "intelligently" within its limitations.  It seems that some may have been using it naively and trying to make final designs with it. 

Edited December 4, 2019 by March Audio

[davewantsmoore]

4 minutes ago, March Audio said:

There isnt, the NC500 eval board is not a finished product.  Its just literally to get you going.  Its for use by designers that will use it "intelligently" within its limitations.  It seems that some may have been using it naively and trying to make final designs with it. 

This is starting to make more sense now.

 

Perhaps that's why they didn't respond to the "your circuit diagram is incomplete" comments....  or the comments around the RC filter formed by the DC blocking capacitor.

 

..... as they thought  "no, that's exactly the circuit we were using".

[kukynas]

yes, it's eval board and it should never be used in end product but some got to the market at early stage  

 

measurements from diyaudio thread supposed to be related only to OEM board itself before and after applied changes, not entire amp

 

I don't buy bavmike's exaggerated crap either, it's his way or noway...

 

I didn't measure the before and after coz I didn't have scope at that time, I should have add as mentioned somewhere in the thread that the noise was combination of HF leak somewhere clearly audible at tweeter level no matter if with connected or disconnected amp from source and distortion itself audible at high frequencies during reproduction, RC filter together with other filter has been suggested top of resistor change and as said clearly audible after applied changes not only by me but all others

 

my main interested in this thread is MP range, after reading MP range thread on diyaudio some guys weren't happy with noise level out of their amp so hence the thread, I don't expect same distortion level as with OEM boards otherwise measurement's from Hypex and Amir's plots would clearly show it but coz it has the same simplified pre-amp stage I'd like to question if there's anything still to be improved or what has been done to eliminate the problem earlier mentioned, I'm clearly aware of the fact that these are cheaper offerings and we shouldn't expect same level of performance as from the NC5 range but what if we could do something to it?  

[Addicted to music]

If I was in your shoes with your issue, I’d be putting it out and asking the man himself Bruno.   Why don’t you forward and email to him, worth a shot, all that can happen is that he’ll ghost you......  

 

when a PSU for a nc400 blew up and took out a 30A circuit breaker to the main building, he was kind enough to send a replacement free of charge....  so no harm in trying.

[Addicted to music]

https://www.audiosciencereview.com/forum/index.php?threads/review-and-measurements-of-nord-one-nc500-amp.7704/

 

Noisier than the nc400....  so im glad I never taken the task for the nc500.

You can see that looking at these measurements, you wonder why what you have is noisy.