Hypex UcD180HG HxR amplifier module analysis and review

[pma]

THD distortion plot vs. power for 4ohm, 6.8ohm and 8ohm resistive load. Not that it is very important, however it is interesting to see the nonlinearity change with load. It seems that it is the output voltage that is important (8ohm load needs 1.414x more voltage to get same power as 4ohm load). The 8ohm plot ends at 90W only for the reason that I did not use enough input voltage in the amplitude step. Again to mention, the maximum achievable power is influenced by my 100VA transformer used.

 

[pma]

It seems that it is the output voltage that is important (8ohm load needs 1.414x more voltage to get same power as 4ohm load).

This hypothesis has proven to be correct. It is the output voltage that corresponds with rise of distortion of the UcD topology. Below is the graph with output voltage as X-axis variable.

It should be mentioned that speaker distortion is not much affected by this. It is the speaker current that tells about speaker distortion and even with ideal amplifier with zero output voltage distortion we get distortion in speaker current, because the speaker is a nonlinear load. And, it acts according Ohm's law, so ideal non-distorted voltage divided by nonlinear load creates nonlinear current, Iout = Vout/Z(v,f).

 

[sarumbear]

Is it me or this amplifier's SINAD of less than 70dB will land it in the "fair" category? The forum declared Rega IO to be poor and it measured 69 dB.

Am I missing something?

 

[Matias]

Is it me or this amplifier's SINAD of less than 70dB will land it in the "fair" category? The forum declared Rega IO to be poor and it measured 69 dB.

Am I missing something?

SINAD at 5W 4 ohms 1kHz is 76, the average in the list and in the orange "fair" category.

Amplifier SINAD list ASR + 3rd parties

docs.google.com

 

[voodooless]

Is it me or this amplifier's SINAD of less than 70dB will land it in the "fair" category? The forum declared Rega IO to be poor and it measured 69 dB.

Am I missing something?

It’s not only the SINAD number that gained it the poor rating. It’s the bad thermal handling, instability, poor PSU noise rejection and a few other gripes.

 

[sarumbear]

It’s not only the SINAD number that gained it the poor rating. It’s the bad thermal handling, instability, poor PSU noise rejection and a few other gripes.

How about the fact that it is in the fair category which is purely based on measurements? This amplifier is borderline in the poor category.

 

[voodooless]

How about the fact that it is in the fair category which is purely based on measurements?

Sure. No problem with that. This this is roughly the median of performance I think. For a more than 15 year old class D design, still quite good.

 

[sarumbear]

Sure. No problem with that. This this is roughly the median of performance I think. For a more than 15 year old class D design, still quite good.

When around 60% of the SINAD ranking list is above 78 dB I don’t think the median will be less than 60dB at below the rated power.

 

[pma]

Is it me or this amplifier's SINAD of less than 70dB will land it in the "fair" category? The forum declared Rega IO to be poor and it measured 69 dB.

Am I missing something?

If you want to be specific, it is 76dB/1kHz/5W/4ohm, please see post #1. However, SINAD, the mix of noise and distortion at one level and one frequency, as used here in the ASR, is absolutely unimportant measure to me and I am very unhappy that it is used as a universal measure of amplifier quality here and even the comparison chart is built this way. It is simple but wrong, absolutely wrong. It is the speaker nonlinearity that defines the resulting distortion and this distortion is never better than the distortion measured in output current which is defined by nonlinear speaker impedance function. For those interested, please read explanation of current distortion on my web

Current drive of speakers and speaker distortion

You can also try the ABX test posted here on ASR (which does not cover speaker distortion, only amplifier)

Blind test - can you tell the class D amp from original loop in ABX

This is the test of the looback file vs. the file recorded through a class D amplifier. The files for download are here: https://www.dropbox.com/s/z13zpe0k08opfma/Mutter.zip?dl=0 If you are interested, please download the zip file, unzip it and listen. Dropbox account is not needed, just click...

www.audiosciencereview.com

and check if you can tell the difference, of course supported by a valid ABX protocol.
Waving with SINAD without a wide set of other parameters makes no sense.

P.S.: the main difference (except for price) between UcD and NCore is the gain of feedback loop. Simply, NCore has much more feedback thus reduced distortion. Other than that both circuits are very similar, including frequency response invariant with load and low output impedance even at high frequencies. I do not think there would be a chance to find a difference in a double blind test.

 

[sarumbear]

If you want to be specific, it is 76dB/1kHz/5W/4ohm, please see post #1. However, SINAD, the mix of noise and distortion at one level and one frequency, as used here in the ASR, is absolutely unimportant measure to me and I am very unhappy that it is used as a universal measure of amplifier quality here and even the comparison chart is built this way. It is simple but wrong, absolutely wrong. It is the speaker nonlinearity that defines the resulting distortion and this distortion is never better than the distortion measured in output current which is defined by nonlinear speaker impedance function. For those interested, please read explanation of current distortion on my web

The fact of the matter is, SINAD is the criteria used in ASR. According to that criteria, this amplifier is at best mediocre.

 

[Matias]

I have seen UcD modules showing up on Aliexpress, until then there were none. Could it be legit or some kind of fake copy?

139.5US $ 10% OFF|Hypex Ucd Power Amplifier Board Ultra-low Distortion Hifi Audio Pure Discrete Digital Power Amplifier 1 Pair - Operational Amplifier Chips - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

https://www.aliexpress.com/item/4000987433868.html

 

[boXem]

I have seen UcD modules showing up on Aliexpress, until then there were none. Could it be legit or some kind of fake copy?

139.5US $ 10% OFF|Hypex Ucd Power Amplifier Board Ultra-low Distortion Hifi Audio Pure Discrete Digital Power Amplifier 1 Pair - Operational Amplifier Chips - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

https://www.aliexpress.com/item/4000987433868.html

Both expensive and fake .

 

[pma]

I would like to mention one probably less known (to general public) feature of self-oscillating class D amplifiers. It applies to both hysteresis switching (TPA32xx like) and phase shift controlled oscillations (UcD, NCore). Switching frequency is not stable and is a function of output voltage amplitude. With higher output we get lower switching frequency. (For better understanding, please read Hypex papers linked in the thread below)

Class D must read papers and publications

I would like to open the thread dealing with important publications on class D amplifiers principles, topologies, characteristics and limitations. One of the great sources is the Hypex publications library at https://www.hypex.nl/p/application-notes-white-papers/ The most explaining and...

www.audiosciencereview.com

And really dramatically lower when we are near to clipping. I will demonstrate it on 2 measurements of my UcD180 sample amplifier.

1. Output swing is 57Vpp, load 6.8 ohm

We have output power of 60W/6.8ohm. Switching frequency falls from 420kHz at idle to some 225kHz at peaks. Trace broadening is visible as output LC filter is less effective.

2. Output swing is close to clipping

We have output power of 92W/6.8ohm (limited by my 100VA power supply). Switching frequency falls now from 420kHz at idle to some 80kHz at peaks. This is a great decrease. Trace broadening is visible and is even more pronounced. Output LC filter is less and less effective. The result is elevated noise floor in THD+N and spectrum measurements. Further increase of amplitude leads to radical rise of THD+N and visible drop-outs in switching frequency.

 

[pma]

CMRR measurement

I have just measured a CMRR (common mode rejection ratio) which tells how common mode voltage is attenuated as a function of frequency. We can see that 50Hz is attenuated of about 70dB and 500Hz about 80dB, 10kHz of 63dB. These are quite excellent results of an audio power amplifier. It tells the user how well the potential ground-loop hum and buzz would be attenuated.

This kind of measurement should be a standard for every audio product with balanced input.

 

[pma]

Measurements into 2.5 ohm load

Please note that maximum power is limited by the 100VA transformer in my power supply.

Please note that Hypex UcD180HG has almost no issues with increasing distortion with frequency and that even with demanding 2.5 ohm load. This measurement was made especially for @restorer-john , he will know why (90 kHz measuring bandwidth, 192kHz Fs) to show that properly designed class D has no violent rise of distortion with frequency. Yes that is Bruno and Hypex and proper class D topology.

 

[Harmonie]

Measurements into 2.5 ohm load


Please note that maximum power is limited by the 100VA transformer in my power supply.




Please note that Hypex UcD180HG has almost no issues with increasing distortion with frequency and that even with demanding 2.5 ohm load. This measurement was made especially for @restorer-john , he will know why (90 kHz measuring bandwidth, 192kHz Fs) to show that properly designed class D has no violent rise of distortion with frequency. Yes that is Bruno and Hypex and proper class D topology.

Good to know, as some Electrostatic speakers make some suffer pretty close to 2 ohms in some frequencies.

 

[pma]

Now the measurement of THD vs. output power with 2.5 ohm load at 1kHz, 5kHz, 10kHz and 15kHz, again with @90kHz bandwidth. From these plots we can see increasing distortion with frequency above 50W.

 

[EJ3]

Measurements into 2.5 ohm load

View attachment 170320
Please note that maximum power is limited by the 100VA transformer in my power supply.


View attachment 170324

Please note that Hypex UcD180HG has almost no issues with increasing distortion with frequency and that even with demanding 2.5 ohm load. This measurement was made especially for @restorer-john , he will know why (90 kHz measuring bandwidth, 192kHz Fs) to show that properly designed class D has no violent rise of distortion with frequency. Yes that is Bruno and Hypex and proper class D topology.

I run my subs as a 2 ohm (stereo) load (they have dual 4 ohm voice coils). I guess that I could run them as a 4 ohm (stereo) load but I would probably lose about 3 DB efficiency. It would certainly require adjustments. My NAD 2200 on my subs can put out over a KW of power into 4 ohms ([over 500 a stereo channel] as measured in AMIRM's review on this site). He did not test the 2 ohm load, which is the way I run it but it should be able to put out around 1.6 KW into 2 ohms (800+ per stereo channel). I am looking for a good reason to change but haven't seen it so far (not to mention HYPEX's unwilling to sell an amp module when something does go wrong).

 

[pma]

Here we we are limited by my power supply used. With UcD400HG and SMPS400A400 there should not be a problem with power. These modules are safe for use with load above 1ohm. It seems, based on my experience with UcD180HG, that lower impedance of the load the better. Distortion here is about 0.02%/100W/2.5ohm. I am sure that with SMP400A180 the power would be considerably higher, with my UcD180HG.

 

[EJ3]

Here we we are limited by my power supply used. With UcD400HG and SMPS400A400 there should not be a problem with power. These modules are safe for use with load above 1ohm. It seems, based on my experience with UcD180HG, that lower impedance of the load the better. Distortion here is about 0.02%/100W/2.5ohm. I am sure that with SMP400A180 the power would be considerably higher, with my UcD180HG.

I was unaware that they could pull 1 ohm. More & more I am starting to see a reason to switch. Perhaps when I (if) retire in 2 years...I may switch. Or get one for the subs & then, later one for each channel.