DIYclassD.com UcD400 stereo (power amp) kit: Electricity consumption & heat measurements

[cybercobra]

HiFi-n00b here. After reading various threads (here and elsewhere) about amplifier wattage requirements, and making some detailed spreadsheets about available amplifiers and about my particular wattage needs, and thereby getting depressed (at the lack of consensus/confidence), I decided to go for "definite overkill and future-proofing, on an increased (by my standards) budget".
Also, this amp will be part of a dual-purpose stereo speaker system, hooked up to both music and my TV. So it needed either HDMI input, or an auto-on+off feature, or low idle power consumption (climate change is real and bad, folks).
Ultimately I decided on a DIY Class D UcD400 stereo kit amplifier, using Hypex UcD technology. I was eager to finally make some decision after so many hours spent researching...

If you're here for sonic performance measurements, stop reading now; I don't have any of the equipment for that. But let's see what the actual power consumption and waste heat are..

Vendor's Data

Hypex's FAQ states (emphasis mine):

9 Q: The blue heatsink on the module gets very warm and the output coil even gets warmer! I thought UcD Class D stayed cool at all times!
A: [...] An idling UcD amplifier draws up to 2% of rated power. That’s 8W for a UcD400 amplifier. Whether that translates into a small or large temperature rise depends on the size of the heatsink. Please refer to the Thermal design application note for more information.

Their datasheet for the raw mono amp board "UcD400HG with HxR" quotes 400W into 4Ω @ 1% THD. So their math checks out (2% * 400W = 8W).
Since the kit is stereo and includes 2 of those boards, I expected (at most) twice the quiescent wattage, i.e. 16W.
And likely less than that, since the single included SMPS400A400 shared power supply is rated "only" 600W max & 400W max @ 20Hz; versus the amp boards' combined 800W rating. The power supply's datasheet states:

Idle Losses @ 230V AC, 50 Hz: 7.5 W

The kit quotes 200W @ 4Ω and 200W @ 8Ω. The webpage isn't crystal clear, but I presume these are watts-per-channel figures, given the power supply.
I am unclear on how 8Ω speakers impact the consumed (vs. provided) wattage.
After assembly, the kit connects directly to a wall outlet via IEC cable; no external power brick. Being a power amp, it lacks any built-in volume control.

Test Setup

• Poniie PN1500 (kill-a-watt clone from Amazon)
  • Claimed resolution: 0.01W
  • No maximum delay or responsiveness figure quoted.
  • No "maximum [Watts/etc.] over time period" function. Maximums are entirely manual.
• Audio source: JDS Labs EL DAC II+
  • Connected via Bluetooth using aptX HD, to an Android phone
• Noise meter: Triplett SoniChek Pro 3550
  • Ordered from Fry's Electronics days before the bankruptcy... Barely dodged that bullet; RIP.
• Listening distance: Approx. 7 ft
• Speakers: Triangle BR03
  • Claimed sensitivity: 90
  • Rated impedance: 8Ω
  • Power handling: 100W
• U.S. power grid. ~120V, 60 Hz.
• Thermometer: Digital instant-read meat thermometer off Amazon with flip-out probe.

Test Results

• <31 W; Meter reading 30.XX W with the fractional portion fluctuating, under all the following conditions:
  • Idle
    • No signal (Amp inputs disconnected)
    • Quiet signal (Inputs connected to DAC. DAC powered on. No digital audio playing.)
  • Playing Alternative Rock music off my phone with Bluetooth volume @ 50%
  • Playing a YouTube video with people conversing normally, via my phone with Bluetooth volume @ 50%
• When playing a non-quiet passage of "Audiophile Music - Bass Test Reference - The Best of Audiophile Music Collection - NbR Music" for a several seconds at maximum volume (praying the neighbors don't complain)
  • 36W
    • Given how brief instantaneous peaks are in music, and our cheap consumer-grade meter, this is likely an undercount.
  • Noise meter in "Max" mode: 82 dB(A) achieved
    • Subjectively, this felt at least discomfortingly loud.
• 90.5°F at idle, probing several random parts of the top and side of the amplifier's metal case, holding steady until multiple seconds elapsed without further increase, and taking the maximum overall.
  • Room ambient temperature: 72.3°F
  • Temperature delta: 18.2°F
  • Amplifier is sitting on a wooden table. At least 6 inches of clearance in each planar direction. Multiple feet of clearance upwards. Ventilation isn't an issue.
  • Top part of the case feels towel-dryer-gripped-through-a-towel hot. Not gonna burn you, but I wouldn't stack on top of it or put it in an enclosed space.

Ordering and Assembly
I'm happy to report that 2-day EU-US UPS shipping operated without delay, despite COVID, etc. I am not a lawyer, but amplifiers apparently being tariff-free in the US probably helps. On the downside, no significantly cheaper shipping options were offered by the vendor.
Annoyingly, the vendor's payment solution declined transactions with both of my US credit cards. I verified with customer service for one of the cards that no transaction had even been attempted on the network, so the issue was definitely on the vendor's side. There's also a surcharge for any payment method besides wire transfers or Netherlands bank accounts. Ultimately I was forced to use PayPal, which worked.

I've assembled computers before; this wasn't any more difficult. I'd say the most time-consuming parts were determining, with a ruler, process of elimination, and skimming an article on the metric screw naming scheme, which screws and washers were which in several steps. And of course, there's a moment of tension where you pray you got the grounding and A/C wiring right, and aren't going to either blow the amp's fuse or electrocute yourself. Still wish I had a multimeter. But power-up and an audio test went fine; I took off the rubber gloves, and touched the unit without any tingle even. They do give you 3 spare fuses, in case you're less lucky than me.

The daughterboard with the amplifier's power LED also has a mysterious three-way switch which the case labels simply "0-1-2" and which the instruction manual doesn't seem to explain. Experimentation shows this is a brightness setting for the power LED (0 = Disable/Defeat, 1=Low, 2=High).
The cable management seems meh. You route them through some partial corridors and compress the wires together with zipties in 3 places, but their stiffness fights you and you never bind them to the case, so the routes are vague at best.
It's weird to me how acute some of the bends are in the CK-4 and CK-7 electrical wires, but then again I'm not an electrician. And they sell this within the EU, where they're more regulation-happy; so it's probably fine.

Conclusion
So we came into this hoping for 16W idle, but actually got ~30W idle, a bit less than twice as much. Perhaps someone who's an electrician or EE can make an educated guess about the discrepancy?
30W is dim incandescent lightbulb territory. Idling 24/7 at my electric rates, that's $42/year for high-fidelity sound anytime. Imagine it's the '80s and you have a dim porch light constantly on. So, acceptable to me, though higher than I'd like. On the bright side, for non-metalhead listening at reasonable volumes, we're barely pulling any extra watts.

 

[restorer-john]

So we came into this hoping for 16W idle, but actually got ~30W idle, a bit less than twice as much. Perhaps someone who's an electrician or EE can make an educated guess about the discrepancy?

At idle, the modules may be dissipating their rated 8W each giving 16W and the SMPS is dissipating the rest (14W). Don't forget, idle losses for the PSU don't include the modules pulling their idle current.

It's pretty much spot on if you ask me.

 

[voodooless]

Nice test and writeup! Welcome!

Just a question though: why go for an almost two decades old Class-D design? You could probably get an NCore kit for just about the same price?

 

[somebodyelse]

30W is dim incandescent lightbulb territory. Idling 24/7 at my electric rates, that's $42/year for high-fidelity sound anytime.

You can probably hook a trigger to the PSU's standby input for less than the cost of a year's idle power consumption.

 

[MCH]

These things are very personal and what is not much for someone can be a lot to someone else. Your reasoning is yours and i respect it, and even agree, but let me share some figures:
48 eur/ year is what i pay for amazon prime, and it has been ages since i don't have a 30 w bulb at home, let alone always on, i think the most powerful one i have now is 12 w.

I have a Hypex amp that consumes <0.5 w when idle, and i paid for it ca. 250 eur more than this kit costs. I plan it to last many years. Being realistic with energy price increases in ca. 5 years (probably less) the difference in iddle energy consumption will pay for the amp price difference, and in 15 years (probably less) for more than the cost of the amp.

 

[cybercobra]

Just a question though: why go for an almost two decades old Class-D design? You could probably get an NCore kit for just about the same price?

(Someone should tell that to NAD . I thoroughly researched their range. Only their high-end stuff explicitly uses NCore or better. They have several current products which are explicitly UcD, and some just vaguely "HybridDigital" ones. The mid-range stuff seemed low-wattage for the price; but I'm a programmer with an HTPC who believes in downloadable music, so I don't need a streamer device. And/or perhaps my original ~$500 target was unreasonable for "audiophile-grade"? Or the Internet successfully bullied me into thinking I need too many watts? *shrug*)

I fully admit to not exploring DIY/boutique options as thoroughly in my research. I was in a bit of a stupor, having pored over (integrated & power) amp specs and reviews for over a week straight. My lack of soldering capability also limits my options. What US-compatible product would you have recommended, for posterity? My stereo kit's base price is €550. The only NCore kit on Hypex's site is "NC400 mono kit", which is €600, and you'd need 2 of them for stereo. So that's a significant price increase, and that mono kit is (as of the time of writing) out of stock anyway. One of my other constraints was instant gratification .

Ghent Audio is in China; what with the lockdowns, their work is on pause until ~5th May, and IDK how reliable <1month US-China shipping is in today's environment. The Buckeye Amps guy here in the US has substantial lead-time due to queue length (Good for him!). And CE/UL certification makes me feel better, fire-wise; boutiques seldom offer that. I suppose I could've tried cobbling together components from Audiophonics.fr, but I'm more comfortable with all-in-one kit, given my electronics inexperience. Audiophonics' prebuilds do look competitive, now that I look again. I didn't come across anyone within the US offering unassembled kits.

Anything Purifi-based has a large price bump. I figure UcD vs. NCore is likely difficult to notice to non-golden ears, based on already-great UcD THD/SINAD charts from ASR. And if I'm going to the trouble of international ordering, then directly supporting one of the innovator companies gives me some warm-fuzzy feelings as a fellow engineer, and some additional confidence about product quality. Any of these are a significant upgrade from my previous ~20 year-old fancy-looking Sony CD-player w/ speakers, or my new TV's built-in speakers.

 

[cybercobra]

You can probably hook a trigger to the PSU's standby input for less than the cost of a year's idle power consumption.

Interesting. Could you provide a pointer or some more details? It'd need auto-on & auto-off for convenient usage with my TV.

 

[MCH]

There is a member here that says he buys amps in Audiophonics and does not need to pay taxes when he receives them in US.
If that is true for you too, at the current usd/eur rate you can get their assembled amps at a very very attractive price (Audiophonics shows the price without vat just below the ttc price)

 

[somebodyelse]

In the SMPS400A400 PSU datasheet see the 'SMPS Standby' pin. Unfortunately it's the inverse of the usual trigger signal - you need to apply voltage (3.3V to 12V) to to put it into standby - so a bit of DIY will be needed. Exactly what depends on what you want to trigger from - one or more 12V trigger outputs, audio signals, a remote control, home control system or whatever. It could be as simple as a phone charger to supply the voltage and a relay switched by a conventional 12V trigger, something based on the audio detection part of ESP Project 38, or whatever.

 

[deercreekaudio]

Nice test and writeup! Welcome!

Just a question though: why go for an almost two decades old Class-D design? You could probably get an NCore kit for just about the same price?

Keep in mind that the UcD400 amplifier module has been significantly upgraded over the last few decades, specifically with the addition of HG and HxR regulation technology.

Here is a summary of the current UcD400 state-of-the-art features:

1. A mathematically exact understanding of self-oscillation. This allows optimization of large signal performance.
2. Improved comparator circuit ensures that actual behavior matches the theoretical model as closely as possible.
3. New gate drive circuitry improves open-loop distortion at moderate signal levels while significantly reducing idle losses.
4. HxR regulation technology providing power stability and a regulation rejection ratio of a 110 dB.

Deer Creek Audio is an authorized Hypex dealer.

 

[Andersonix]

I leave my amps turned off when unattended. Why would I want a device made to amplify every input by 1000 to be active 24/7 and ready to blow my speakers?
And it still gets hot (in a tiny spot somewhere under there?), so you're sucking away its lifespan, even if drawing only a small parasitic amount of energy the whole time. You could run this test: Leave one powered on, and switch one on/off, and then report back here which one lasts longer.

 

[Willem]

I think it is time to connect to the modern world of climate change and skyrocketing energy prices. Leaving this on will use about 250 kWh per year, or some 10% of the electricity consumption of an average Dutch home. At current Dutch market prices for electricity that will cost about 200 euros. Get a master/slave power strip if switching on/off is too much hassle. I use one because the power switch on my Quad 606-2 poweramp is hard to reach at the rear. It works fine.

 

[Head_Unit]

Leaving this on will use about 250 kWh per year, or some 10% of the electricity consumption of an average Dutch home

Wait, "~30W idle" is 10%?!? Really? Doesn't even a television use a lot more?
I get 262,800W over a year so yeah 250 kWh...jeez I gotta look at my electric bill and see how much I use (though it includes charging a Chevrolet Volt some of the time).