Hypex NCxxxMP buffer bypass. Someone?

[Davide]

Hi everyone.
Someone by chance tried to modify a Hypex NCxxxMP module bypassing the buffer according to the manual? (for those who do not know it details which resistors to remove and add).
I would like to try it by having an Ultralite Mk5 that delivers 8.7v.
However I have no experience in modifying SMD components so I wanted to know if anyone had tried their hand and could give me suggestions.
Thanks

 

[ppataki]

Very good point @Dlomb11
I have multiple NC250MPs so I would be really interested in this topic too!
I have contacted Hypex to ask them about it and the replied that that operation is not supported on those amps and I should not do it....
So if anybody here has done that it would be great to know how it shall be done and how it changed the sound quality (if at all)
Thank you

 

[boXem]

That modification is not supported for old modules, only the latest ones.
You need to keep in mind that the input impedance becomes 2 x 1.8k with all the associated consequences.

 

[Davide]

That modification is not supported for old modules, only the latest ones.
You need to keep in mind that the input impedance becomes 2 x 1.8k with all the associated consequences.

Yes I know.
I'm not sure my Ultralite Mk5 can drive that load because it's not written in the specs and tech support doesn't know anything about it (absurd).
However, the Motu M4 from the teardown done here on ASR seems to mount the OPA1678 on the output buffer, which I think reaches 600 ohms (there is a THD graph on the specification with this load).
Now I drive the NC252MP module with a passive attenuator having a corresponding load of about 1.1k per leg, and it works fine.
So I would like to eliminate the buffer since I have to put this attenuator ... so maybe I also gain some SNR.

In theory, however, it is easy as a modification, 3 resistors to remove and two bridges to make (per channel).
More than anything else I have a doubt: the Hypex specification asks for 0 ohm resistors in 0603 format.
But if I make a bridge with solder, isn't it the same?

 

[boXem]

My comment was more of a general comment for people who would be tempted. If your setup works well with 1.1k, there is no reason why this would change.
Solder bridge is equivalent to 0Ohm, that is indeed the issue with manual soldering 0603 components, solder bridges are a bit too easy to do.

 

[antcollinet]

Hi everyone.
Someone by chance tried to modify a Hypex NCxxxMP module bypassing the buffer according to the manual? (for those who do not know it details which resistors to remove and add).
I would like to try it by having an Ultralite Mk5 that delivers 8.7v.
However I have no experience in modifying SMD components so I wanted to know if anyone had tried their hand and could give me suggestions.
Thanks

Why not just use a 3 resistor U attenuator between the Mk5 and the amp? (see this thread for a discussion)

Balanced attenuator pad with impedance mismatch

Hello everybody, I'm trying to understand how and if it is possible to create a balanced attenuator (pad) that allows to maintain an adequate ratio between the output impedance of the source (in my case a dac) and the input impedance of the driven device (an amplifier) , as well as an adequate...

www.audiosciencereview.com

Or just buy a 10dB inline attenuator, eg

Loops -10db XLR In-Line Balanced Attenuator Adapter - Volume/Noise Reduction Audio: Amazon.co.uk: Electronics & Photo

Shop Loops -10db XLR In-Line Balanced Attenuator Adapter - Volume/Noise Reduction Audio. Free delivery and returns on eligible orders.

www.amazon.co.uk

(RCA versions also available)

 

[Davide]

Why not just use a 3 resistor U attenuator between the Mk5 and the amp? (see this thread for a discussion)

Balanced attenuator pad with impedance mismatch

Hello everybody, I'm trying to understand how and if it is possible to create a balanced attenuator (pad) that allows to maintain an adequate ratio between the output impedance of the source (in my case a dac) and the input impedance of the driven device (an amplifier) , as well as an adequate...

www.audiosciencereview.com

Or just buy a 10dB inline attenuator, eg

Loops -10db XLR In-Line Balanced Attenuator Adapter - Volume/Noise Reduction Audio: Amazon.co.uk: Electronics & Photo

Shop Loops -10db XLR In-Line Balanced Attenuator Adapter - Volume/Noise Reduction Audio. Free delivery and returns on eligible orders.

www.amazon.co.uk

(RCA versions also available)

Yes, I am the author of that thread
The truth is, I hate attenuating the signal because the buffer has too much gain... It is easier and cleaner to have the right gain, or as in my case where I have enough voltage output from the source, bypass the buffer completely and increase SNR.
It's a hobby basically, so I like the idea of tinkering with aiming for the best.
For the record, I need that pad on the NC502MP, where the output of my Mk5 is not enough to guarantee the full power of the amplifier (subwoofers).
With NC252MP and NHT C3, where 10W are enough for normal listening levels, I would bypass the buffer.

 

[Davide]

---

Solder bridge is equivalent to 0Ohm, that is indeed the issue with manual soldering 0603 components, solder bridges are a bit too easy to do

Nice! Thanks!

 

[antcollinet]

Yes, I am the author of that thread

Doh!

 

[mdsimon2]

Although I do think that eliminating the buffer is probably worth it for the SNR increase I should note that the Ultralite Mk5 seems surprisingly sensitive to low input impedance from a THD perspective. On my Cosmos ADC which has variable input impedance depending on voltage range I've seen a 14 dB increase in THD from the highest setting (3480 ohm) to the lowest setting (640 ohm). See below for a quick plot.

Michael

 

[Davide]

Although I do think that eliminating the buffer is probably worth it for the SNR increase I should note that the Ultralite Mk5 seems surprisingly sensitive to low input impedance from a THD perspective. On my Cosmos ADC which has variable input impedance depending on voltage range I've seen a 14 dB increase in THD from the highest setting (3480 ohm) to the lowest setting (640 ohm). See below for a quick plot.

View attachment 198514

Michael

Interesting.
But are you sure this is not due to the measurement error of the ranges?
Does the lower impedance correspond to a larger measurement range?
By the way, does it mean differential impedance or per leg?

 

[mdsimon2]

Interesting.
But are you sure this is not due to the measurement error of the ranges?
Does the lower impedance correspond to a larger measurement range?
By the way, does it mean differential impedance or per leg?

No, it is the opposite, lower measurement range is lower impedance. Lower measurement range gives the best noise measurement but the worst THD measurement. Not all DACs show this behavior but the Ultralite Mk5 and M4 definitely do.

I assume the quoted impedance is differential but I am not sure.

Michael

 

[Davide]

No, it is the opposite, lower measurement range is lower impedance. Lower measurement range gives the best noise measurement but the worst THD measurement. Not all DACs show this behavior but the Ultralite Mk5 and M4 definitely do.

I assume the quoted impedance is differential but I am not sure.

Michael

Ok, thanks for your contribution!
This leads me to think that it is not worth bypassing the buffer if the THD then goes up for the lower impedance.
Also doing the math, I get the same output level if I use a 20db passive attenuator with the input buffer, and -6.5db of digital attenuation without the buffer.
But in the second case I sacrifice 6.5db of SNR, probably as much as I gain by bypassing the buffer.
So if we consider the worsening of THD due to the lower impedance, perhaps the overall performance is even worse without the buffer...
It would be nice alternatively to know how to modify the buffer to reduce the gain, but I have no idea ...

 

[sarieri]

I’m in a similar situation. My soundcard(Motu Ultralite mk5) has an output impedance of 220ohm. I connect it to two LA90D (low gain @9.4dB) and two NCx500 (buffered @20dB gain). I want to lower the gain of NCx500 and match them to the LA90D.

I could bypass the buffer and lower the gain of Ncx500 to 11.5dB. That case I get an extra 3 dB SINAD. But since the motu doesn’t like 1.5k ohm input impedance very much, according to measurement from @mdsimon2, I lost a few dB comparing to and a line level attenuator and keep the impedance at 47k ohm. (i still have to apply -2dB at DSP/attenuator to get an exact match).

If I add a -10dB line level attenuator, SINAD of NCx500 is 3dB worse, but motu should be happier with 47k ohm input impedance.

I guess in the end none of these subtle differences can be heard so it doesn’t matter at all…

 

[antcollinet]

I guess in the end none of these subtle differences can be heard so it doesn’t matter at all…

This

 

[sarieri]

This

I would probably just solder few resistors to make a lone level attenuator between dac and amp…. Do you know if those attenuators will affect the frequency response in anyway? I guess NO?

 

[antcollinet]

I would probably just solder few resistors to make a lone level attenuator between dac and amp…. Do you know if those attenuators will affect the frequency response in anyway? I guess NO?

Not if they are purely resistive, and located close to the amp. It's even possible to build into an XLR input connector if you make your own cable.

 

[sarieri]

Not if they are purely resistive, and located close to the amp. It's even possible to build into an XLR input connector if you make your own cable.

Thank you! I did all my cables. They are TRS to XLR cables, so I guess I will just put those resistors at the XLR end (amp only takes XLR input). Should that just be a u-pad attenuator?

Edit: I wonder if I can also do impedance matching by add a u-pad here. For example, R1/2=8.2, R1=270, and the output impedance is now 56 ohm?

 

[antcollinet]

You are putting qute a heavy load on your DAC with 220ohm. That means your attenuation will be higher than you expect due to the voltage dropped across the output impedance of the DAC. EG another 6dB if the DAC output impdedance is as high as 220 ohm. I'm also not sure how that input impedance has been calculated - looks wrong to me.

In any case you don't need to impedance match with audio interconnect. Just make sure you are not overloading the output of the DAC, and take into account the output impedance of the DAC, and input impedance of the amp to calculate the attenuation.

 

[sarieri]

You are putting qute a heavy load on your DAC with 220ohm. That means your attenuation will be higher than you expect due to the voltage dropped across the output impedance of the DAC. EG another 6dB if the DAC output impdedance is as high as 220 ohm. I'm also not sure how that input impedance has been calculated - looks wrong to me.

In any case you don't need to impedance match with audio interconnect. Just make sure you are not overloading the output of the DAC, and take into account the output impedance of the DAC, and input impedance of the amp to calculate the attenuation.

Thank you for the insight. In order to put a light load on the DAC, Z1 should be at least 10 times the output impedance of the DAC right? The motu Ultralite’s output impedance is 220 ohm, so I guess 20k ohm will make more sense. I don’t know what value should Z2 be however… the amp’s input impedance is 47k ohm, should I just pick a small value? I’m also quite confused about the L/U Type matching option of the calculator. I will read more to learn.