Hypex NC400 Low Gain Modification

[mppix]

Summary​

This post summarizes information to changes the Hypex NC400 gain to 12dB "low gain" (from 26dB "high gain") by removing the Rg resistor from the instrumentation section. With lower gain, the amp clips above 9.6V rms instead of 2V*.
An added bonus is that padding, i.e. gain reduction, cuts noise in half and results in an additional 6dB of SNR resulting in a performance comparable with the AHB2, Purifi, and Nilai amps.

Warning: this modification requires soldering and will void any warranty.

* The lower line level is useful when the amp is used with pro audio gear like RME or MOTU. Line level consumer audio typically clips between 2-4V, line level pro audio gear above 12V.

Principle​

The NC400 has an instrumentation amp section with gain is Av=(1+2*Rf/Rg)=4.7=13.4dB, where Rg=1.2k and Rf=2.2k.

It is noted that the instrumentation amp is omitted in the Purifi amp. This reduces the gain by 4.7x and, results in differential amp at 4.4x or 12.8dB gain, but with a low input impedance.

It is possible to achieve the same gain optimization by setting removing Rg=inf, i.e. removing Rg (R141 on the PCB - red rectangle). The modification retains the NC400 high input impedance.

My modified 4xNC400 "low gain"

Measurements (credit Langston @ Klipsch community)​

Benchmark AHB2 (for reference):​

NC400 before mod "high gain":​

NC400 after mod "low gain":​

Thoughts​

The gain is mostly relevant in setups that run professional (or high-gain) interfaces or DACs. I only did this modification to achieve a better gain match in my system (without inline attenuators).
The 6dB noise improvement is a nice benefit. It is quite plausible that is also observed in the Hypex Nilai500DIY measurements (link). It is good to see that the NC400 benefits from this too.

Reference and measurements​

miniDSP SHD Power & Hypex NC400 Class D Amps

community.klipsch.com

 

[AnalogSteph]

A 6 dB reduction in noise from a 13.4 dB reduction in gain suggests that input stage noise post-mod has dropped substantially below power amp noise. A 9.6 Vrms clipping level (almost +22 dBu) also seems a bit high. I'd think that aiming for an input stage gain of around +7 dB would make for a better compromise - then you'd be hitting the ca. 4.2 Vrms that a lot of balanced DACs will output.

This would require replacing Rg with a thin film resistor of 3.32 or 3.40 kOhms (the footprint looks big enough for a 1206 if I am not mistaken). My estimate for dynamic range thereafter would be 127.1 dB. Resulting total gain would be about 10.6 or +20.5 dB.

 

[mppix]

A 6 dB reduction in noise from a 13.4 dB reduction in gain suggests that input stage noise post-mod has dropped substantially below power amp noise. A 9.6 Vrms clipping level (almost +22 dBu) also seems a bit high. I'd think that aiming for an input stage gain of around +7 dB would make for a better compromise - then you'd be hitting the ca. 4.2 Vrms that a lot of balanced DACs will output.

This would require replacing Rg with a thin film resistor of 3.32 or 3.40 kOhms (the footprint looks big enough for a 1206 if I am not mistaken). My estimate for dynamic range thereafter would be 127.1 dB. Resulting total gain would be about 10.6 or +20.5 dB.

A ~22dB "medium gain" is likely the best solution when combined with "consumer" balanced DACs (~4V).

"Pro" interfaces (RME, MOTU, ...) happily put out 20/21dBu and more. For example, this can be used to run for professional reference levels -18dBFS=0dBu (European Broadcasting Union, EBU) or -20dBFS=+4dBu (Society of Motion Picture and Television Engineers, SMPTE).

This gives some spectacular headroom (and dynamic range), at least in theory.

Graphic reference: SOS (link)