Hypex UcD180HG HxR amplifier module analysis and review

[pma]

Neat. Also note that one solder joint that hasn't flowed correctly. Some components are frighteningly close to the mounting screws.

Tom

View attachment 163973

I think that the resistor close to the heatsink screw might be an optical illusion because of the camera lens axis vs. PCB angle during taking the photo, so the bolt head apparently overlaps the SMD resistor. And it is a wide-angle lens. Yes the bolt head is quite close.

Regarding the solder joint mentioned, again it is a question if it is a failure or light reflection

 

[tomchr]

That's entirely possible. I'm sure Hypex wouldn't design a PCB where the components were sheared off by a mounting screw.

Thanks for the closeups. In the second shot you can clearly see the void in the solder joint on that one 3-pin PTH component. And also solder splatter by R35. It looks like a typical IPC-A-610 Class I assembly. I have my boards assembled to IPC-A-610 Class III (and inspected to Class II). You can read more here: https://neurochrome.com/pages/the-north-american-advantage

Tom

 

[tmtomh]

I was unaware of this module (became familiar with these Class D modules more recently) - thanks for the detailed testing and write-up!

I'm thinking if these modules ever became available at something like 80-100 Euro then they could really be a low-cost alternative. They are already lower-cost than Purifi and NCore of course. I'm just wondering about the market for a module that costs about 130EU with wiring and such and provides about 70dB SINAD, when a discerning buyer can put together a lower-end Ncore-based DIY for something like maybe 300EU give or take? I'm just wondering how many folks who can spend 130 for 70dB SINAD would not instead prefer to spend 300 for 95+dB instead.

I'm not making an argument about that necessarily, just raising the question.

 

[pma]

I was unaware of this module (became familiar with these Class D modules more recently) - thanks for the detailed testing and write-up!

I'm thinking if these modules ever became available at something like 80-100 Euro then they could really be a low-cost alternative. They are already lower-cost than Purifi and NCore of course. I'm just wondering about the market for a module that costs about 130EU with wiring and such and provides about 70dB SINAD, when a discerning buyer can put together a lower-end Ncore-based DIY for something like maybe 300EU give or take? I'm just wondering how many folks who can spend 130 for 70dB SINAD would not instead prefer to spend 300 for 95+dB instead.

I'm not making an argument about that necessarily, just raising the question.

That's a legitimate question for sure. It might be about priorities as well. From my personal point of view, I emphasize reliability and parameters that are or may become audible.

1) Reliability
This module, if properly implemented (means sufficient heatsink as the most important point), has no thermal issues even under rated power and its continuous power is not time limited. It runs mildly warm and this should result in reasonably long live span. Another advantage is the simplest possible implementation even with the conventional power supply that is providing no control signals to the module.

2) Audible parameters
The load invariant frequency response is the most important parameter to me. And the nonlinearity predictable with load and frequency. On the other hand, 1kHz 5W SINAD (provided that spectrum is constituted of signal harmonics with absence of non-harmonic interferences) is pointless to me unless it becomes audible, which is far above the distortion numbers of this module.

There is no question that 350 EUR NC400 is much better in distortion. It may be a decisive parameter to many users.

My view is a bit different from yours. To me, TPA325X based cheap amplifiers like A07 are wasting of money, though they are so cheap. No defined continuous power, overheating and frequency response violently depending on complex load impedance are the reasons for me to avoid such products. Again, it is a matter of personal view, each to his own.

P.S.: UcD modules are much older than NCore, they are their predecessors and they originate in Philips. NCore is the successor with improved feedback loop to reduce distortion.

The Consumer Electronics Hall of Fame: Philips UcD Audio Amplifier

A compact and efficient amplifier changed how we listen to music

spectrum.ieee.org

 

[voodooless]

My view is a bit different from yours. To me, TPA325X based cheap amplifiers like A07 are wasting of money, though they are so cheap. No defined continuous power, overheating and frequency response violently depending on complex load impedance are the reasons for me to avoid such products. Again, it is a matter of personal view, each to his own.

The real problem here is a chip vs a full amplifier. The TPA amps vary vastly depending on the amplifier design around the chip. Post filter feedback models are very rare, and for some reason raises the cost of those amps quite a bit. By then you're getting into Ncore price territory, especially the fully integrated ones.

 

[abdo123]

I was unaware of this module (became familiar with these Class D modules more recently) - thanks for the detailed testing and write-up!

I'm thinking if these modules ever became available at something like 80-100 Euro then they could really be a low-cost alternative. They are already lower-cost than Purifi and NCore of course. I'm just wondering about the market for a module that costs about 130EU with wiring and such and provides about 70dB SINAD, when a discerning buyer can put together a lower-end Ncore-based DIY for something like maybe 300EU give or take? I'm just wondering how many folks who can spend 130 for 70dB SINAD would not instead prefer to spend 300 for 95+dB instead.

I'm not making an argument about that necessarily, just raising the question.

You’re forgetting that Class D has been traditionally used for subwoofers for decades now. Even in Hi-Fi applications. These things have their own market.

 

[pma]

The real problem here is a chip vs a full amplifier. The TPA amps vary vastly depending on the amplifier design around the chip. Post filter feedback models are very rare, and for some reason raises the cost of those amps quite a bit. By then you're getting into Ncore price territory, especially the fully integrated ones.

And even if it is a post-filter feedback (in TPA325X), the load dependence is suppressed only partially. Because there is a main feedback inside the chip and the PFFB is just a 2nd feedback loop that must be designed as a compromise to keep stability. And even with some efforts the circuit has troubles when the load is disconnected (no load). This is explained in the TI datasheet on PFFB and Bruno's paper [2] - see post #1.

 

[boXem]

And even if it is a post-filter feedback (in TPA325X), the load dependence is suppressed only partially. Because there is a main feedback inside the chip and the PFFB is just a 2nd feedback loop that must be designed as a compromise to keep stability. And even with some efforts the circuit has troubles when the load is disconnected (no load). This is explained in the TI datasheet on PFFB and Bruno's paper [2] - see post #1.

Exactly! The pffb TI sauce has about 6 dB loop gain for stability reasons. To compare with 40 dB for an UcD, 60 dB for Ncore and 80 dB for Eigentakt.

 

[tmtomh]

You’re forgetting that Class D has been traditionally used for subwoofers for decades now. Even in Hi-Fi applications. These things have their own market.

Oh absolutely - there were many aspects I was forgetting or just unaware of - that's what I was admitting!

 

[tmtomh]

That's a legitimate question for sure. It might be about priorities as well. From my personal point of view, I emphasize reliability and parameters that are or may become audible.

1) Reliability
This module, if properly implemented (means sufficient heatsink as the most important point), has no thermal issues even under rated power and its continuous power is not time limited. It runs mildly warm and this should result in reasonably long live span. Another advantage is the simplest possible implementation even with the conventional power supply that is providing no control signals to the module.

2) Audible parameters
The load invariant frequency response is the most important parameter to me. And the nonlinearity predictable with load and frequency. On the other hand, 1kHz 5W SINAD (provided that spectrum is constituted of signal harmonics with absence of non-harmonic interferences) is pointless to me unless it becomes audible, which is far above the distortion numbers of this module.

There is no question that 350 EUR NC400 is much better in distortion. It may be a decisive parameter to many users.

My view is a bit different from yours. To me, TPA325X based cheap amplifiers like A07 are wasting of money, though they are so cheap. No defined continuous power, overheating and frequency response violently depending on complex load impedance are the reasons for me to avoid such products. Again, it is a matter of personal view, each to his own.

P.S.: UcD modules are much older than NCore, they are their predecessors and they originate in Philips. NCore is the successor with improved feedback loop to reduce distortion.

The Consumer Electronics Hall of Fame: Philips UcD Audio Amplifier

A compact and efficient amplifier changed how we listen to music

spectrum.ieee.org

Thanks for your detailed reply. I would actually agree with you about the A07 and other so-called "chip amps" - they have their applications, but for me personally they're not attractive. For my main stereo and any other application where I am looking for true hi-fi amplification, I'm going to pay the money for an Ncore or Purifi type unit. For applications where the level of fidelity isn't as important to me, I'm either going for the cheapest Ncore amp I can find, or else I'll just use an old AB amp or receiver that's in my basement stash of old equipment. Given the current state of the art, I personally don't feel there's any sonic benefit to be had in trying to find a hidden gem in the sub-$300 amp market (unless you shop used and are able to test and repair stuff yourself).

 

[pma]

My setup for the listening test:

Marantz SA7001 CD ==> PMA preamplifier RCA/XLR ==> UcD180HG amplifier ==> Quadral Ascent 90

The preamp serves also as a SE/balanced signal converter and the schematics is below:

and the whole setup

Please note the necessary face mask on the top of the Quadral .

BTW the RCA link cable between Marantz and the preamp is 1m of RG174 . Marantz and the preamp are class II instruments, so we have no ground loops here.

 

[somebodyelse]

IMG Stageline use the UcD250LP and one of the SMPS400s in their STA-400D, and it's probably cheaper than buying the bits to DIY. I last saw it at €250 ex VAT but I think it was usually closer to €300 so borderline for the sub $300 bracket. That product line has another UcD, a couple of Pascals and a couple where they don't say whose module they're using. If you don't need that much power it's not much of a step up to the Audiophonics MPA-125NC or FusionAmp FA122 (which you'd need to case, and skip the DSP if you want to use it for stereo I think), but for equivalent power the nCores are another step up the price ladder.

 

[Tangband]

IMG Stageline use the UcD250LP and one of the SMPS400s in their STA-400D, and it's probably cheaper than buying the bits to DIY. I last saw it at €250 ex VAT but I think it was usually closer to €300 so borderline for the sub $300 bracket. That product line has another UcD, a couple of Pascals and a couple where they don't say whose module they're using. If you don't need that much power it's not much of a step up to the Audiophonics MPA-125NC or FusionAmp FA122 (which you'd need to case, and skip the DSP if you want to use it for stereo I think), but for equivalent power the nCores are another step up the price ladder.

Just take some caution with the IMG stageline , though it looks like very good value ( maybe it is ) , we dont know if it has a HP filtering around 10-15 Hz that makes it useless for hifi-subwoofer use. Its very common with PA amplifiers.

 

[somebodyelse]

Just take some caution with the IMG stageline , though it looks like very good value ( maybe it is ) , we dont know if it has a HP filtering around 10-15 Hz that makes it useless for hifi-subwoofer use. Its very common with PA amplifiers.

I should have linked to the other thread where there's a photo of the internals of both models and some discussion, but no measurement, so we don't know if they've hurt performance with their input buffer implementation either. I must admint I was looking at them more as a cheap source of modules than as a complete amp.

 

[pma]

I was curious what noise/hum level I would get from the whole audio chain as described, so I have measured the complete chain Marantz SA7001 + PMA preamp + UcD180HG at the speaker terminals and here is the result, total noise voltage 20Hz-22kHz flat is 152uV and the graph is calibrated in dBV. This is the example of measurement of an audio chain as it exists under real life conditions, something I am repeatedly calling for. UcD gain is 26dB.

 

[pma]

I made a controlled level matched A/B test between this UcD180HG HxR amplifier and AIYIMA A07, with Quadral Ascent 90 speaker. There is a clear, undisputable difference in the top octave, AIYIMA sounds too sharp.

I made a measurement that clearly explains it. Frequency response at speaker terminals of both amplifiers was measured and plotted as A over B. A is AIYIMA and B is UcD.

We can se that at 10kHz AIYIMA boosts the treble at 0.8dB and the FR rise starts at 4kHz.

The distortion into Quadral Ascent 90 was measured as well, at 1Vrms, which is not a low volume of the sine sweep. One might suppose that AIYIMA would have lower distortion in this test, based on wrong deductions from SINAD measurements into purely resistive load. But alas, speaker is a complex nonlinear load so the result is different.

We can see that AIYIMA has higher distortion which is much more modulated by speaker complex impedance. So it goes, 1kHz 5W SINAD is not everything and worse, it is a bad measure to compare amplifiers.

 

[restorer-john]

I made a controlled level matched A/B test between this UcD180HG HxR amplifier and AIYIMA A07, with Quadral Ascent 90 speaker. There is a clear, undisputable difference in the top octave, AIYIMA sounds too sharp.

I made a measurement that clearly explains it. Frequency response at speaker terminals of both amplifiers was measured and plotted as A over B. A is AIYIMA and B is UcD.

View attachment 164499
We can se that at 10kHz AIYIMA boosts the treble at 0.8dB and the FR rise starts at 4kHz.

The distortion into Quadral Ascent 90 was measured as well, at 1Vrms, which is not a low volume of the sine sweep. One might suppose that AIYIMA would have lower distortion in this test, based on wrong deductions from SINAD measurements into purely resistive load. But alas, speaker is a complex nonlinear load so the result is different.
View attachment 164506
We can see that AIYIMA has higher distortion which is much more modulated by speaker complex impedance. So it goes, 1kHz 5W SINAD is not everything and worse, it is a bad measure to compare amplifiers.

We know that.

5W is an arbitrary, absolutely stupid number and bears no relationship to anything other than a random line in the sand, some person on the internet decided was remotely useful. Hanging all your SINAD results off 5W is unfortunately a mistake.

1 Watt (2.83V@8R) has been used forever and is a standard.

 

[tmtomh]

We know that.

5W is an arbitrary, absolutely stupid number and bears no relationship to anything other than a random line in the sand, some person on the internet decided was remotely useful. Hanging all your SINAD results off 5W is unfortunately a mistake.

1 Watt (2.83V@8R) has been used forever and is a standard.

But is 1W any less random or any more useful as a measurement? The fact that it’s been a standard is an advantage only if the 1W measurement actually gives you apples to apples results that correlate with how amps will perform relative to each other under real world loads.

 

[pma]

We know that.

5W is an arbitrary, absolutely stupid number and bears no relationship to anything other than a random line in the sand, some person on the internet decided was remotely useful. Hanging all your SINAD results off 5W is unfortunately a mistake.

1 Watt (2.83V@8R) has been used forever and is a standard.

Any single number is stupid. Regarding amplifiers, if I had to make a choice of simple metrics, I would have chosen two:

1) Output continuous sine power at THD+N = 1% for 4ohm and 8ohm.

2) maximum deviation of frequency response between 20Hz and 20kHz from straight line, with 2ohm, 4ohm and 8ohm load, the worst number is sufficient. (not speaking about complex load, I agree there are infinite variations. But see what the capacitive character of Quadral impedance above 5kHz makes with A07 response).

The reason why I insist on 2ohm are the impedance dips (and EPDR even worse) of most existing speakers, se below e.g. my Quadral in the test results above, which is believed to be a "light load" for an amplifier.

If the amp cannot handle 2ohm at all, even at lower power, it should have been stated.

BTW, the A07 and UcD frequency responses with Quadral Ascent 90.

 

[restorer-john]

Regarding amplifiers, if I had to make a choice of simple metrics, I would have chosen two:

What's wrong with a rated maximum THD, 20Hz-20KHz from 250mW to full rated power? (the FTC spec). Why not include noise and make it SINAD/THD+N. Sure the numbers will get worse, but the playing field would be levelled.

You can look at that single metric and know that nowhere across the spectrum will the number be exceeded. A worst case number, not a best case cherry-picked number.

maximum deviation of frequency response between 20Hz and 20kHz from straight line, with 2ohm, 4ohm and 8ohm load, the worst number is sufficient.

So, pretty much the full power equivalent of the old power bandwidth (full rated -3dB) test. I run FR plots at full rated power which is a tough test and compare to the 1W FR- that is where a lot of amplifiers fall down, especially at the frequency extremes and into half rated load etc.