Sokel
Master Contributor
- Joined
- Sep 8, 2021
- Messages
- 6,118
- Likes
- 6,195
Name the best or forever stay silent
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Class D amp long term reliability
- Thread starter sophie smith
- Start date
JSmith
Master Contributor
restorer-john
Grand Contributor
Name the best or forever stay silent
Oh, that's tough.
If I was to say an electrolytic brand I've never seen a failure, when in a normal operating environment, it would be Elna, Panasonic/Matsushita (old), Cornell Dubilier, BHC Aerovox or Illinois Capacitor (acquired by CDE).
I'd take the best of the US and the best of Japan for my caps.
And no, Nichicon (much as I love their range) is, IME, one of the least reliable of the 'Japanese' brands. Nippon Chemicon the same.
Fair enough. But the main problem for me is not the one failure but the bent pcb. Which is not a one off and looks like a design failure. If so, it is a recipe for poor longevity.
The good thing about this readership is that it is going to take above average care of the gear. Imagine if the Class D amplifier was more prone to water damage. A large population study may show lots of Class D failures but it’s really due to poor care. In contrast, people may be driving harder loads in this group.
I think premium Class D is more *reliable* since heat is an issue for all equipment but premium class AB is more *repairable* based upon what I have read here. I have seen premium class A where there are isolated areas of heatsink discoloration that over time which haven’t failed yet but suggests that it is at greater risk for failure.
Burning Sounds
Addicted to Fun and Learning
I had a 3 year-old NC502MP module in an Apollon NCMP8350 that began distorting heavily in one channel. It drove the lower-mid (120Hz-1kHz) in my Linkwitz LX521s and the failure did no damage. Apollon sent me a replacement at a reasonable price (excellent customer service BTW) and I fitted it myself. The 4 modules in the NCMP8350 are arranged vertically (like in @pma amp) each with an additional heatsink. I've run temp checks on several internal components including the heatsinks and nothing runs very hot, although the small caps that sit just above the output coils in this orientation always looked vulnerable to me, I have measured about 60deg C on these.
Out of curiosity I sent the failed module to Apollon to see if it was worth repairing. They diagnosed several failed transistors and an ic chip in one channel. Transformers and caps were all OK.
Would I buy another Hypex-based amp again? - yes, but I was hoping that this amp would see me through the next 10 years or so.
I have several class AB multichannel amps and the Hypex-based amp is superior to all of them in terms of clarity, power and from an environmental perspective. Hopefully it was just an isolated failure.
Out of curiosity I sent the failed module to Apollon to see if it was worth repairing. They diagnosed several failed transistors and an ic chip in one channel. Transformers and caps were all OK.
Would I buy another Hypex-based amp again? - yes, but I was hoping that this amp would see me through the next 10 years or so.
I have several class AB multichannel amps and the Hypex-based amp is superior to all of them in terms of clarity, power and from an environmental perspective. Hopefully it was just an isolated failure.
Oh man. I've replaced more IC caps than any other brand. The TTA series used in many cheap guitar amps are horrid.
Class D: Too little experience for me to comment. At work we use SMPS "everywhere" and they have lasted operating 24/7 in the field at high temp (80-100+ degC) for 5-10 years with too few failures to count. OTOH, we have had several bad batches of video cards that were caused by capacitor failures, and a similar problem occurred on one of our evaluation boards (which get stressed pretty good during validation).
LED lights: I replaced almost all the lights in our home with LEDs 5-10 years ago. I have had a number of failures, all power supply best I can tell. Many of them are inexpensive China-made off-brands and I suspect low quality components and perhaps poor design is to blame. I have others, from name brands but also made in China, that have had zero problems. I tend to think it is a matter of quality components and design (and perhaps quality control) rather than country of manufacture.
High-rel components: I know a fair bit about them, having spent about half my career in military, space, and some medical product R&D and production. Some of it is simply better screening but there are fundamental differences in some component materials. Space (S-level) tends to focus on vibration and radiation resistance, at least for most of the electronics, since they are inside a relatively benign environment otherwise (not talking about things outside the body). Quality components, sure, but a lot is in the circuit design for longevity, so plenty of margin in the designs and back-up systems everywhere. Flight hardware actually tended to be harder to design, mainly due to the wide temperature range (-55 to 125 degC, some to 140+ degC; space components I designed were often -40 to 85 degC, some to 125 degC) and generally rough environment. Like vibration; space stuff wasn't usually constantly shaking once in orbit, just had to survive take-off, and the satellite electronics were off until deployment so a mechanical rather than electrical issue. Flight hardware (like fighter jets, missiles) is constantly subject to high-G stress, all sorts of vibration, and rapid temperature changes (jets and missiles get from dessert deck to high altitude very quickly).
LED lights: I replaced almost all the lights in our home with LEDs 5-10 years ago. I have had a number of failures, all power supply best I can tell. Many of them are inexpensive China-made off-brands and I suspect low quality components and perhaps poor design is to blame. I have others, from name brands but also made in China, that have had zero problems. I tend to think it is a matter of quality components and design (and perhaps quality control) rather than country of manufacture.
High-rel components: I know a fair bit about them, having spent about half my career in military, space, and some medical product R&D and production. Some of it is simply better screening but there are fundamental differences in some component materials. Space (S-level) tends to focus on vibration and radiation resistance, at least for most of the electronics, since they are inside a relatively benign environment otherwise (not talking about things outside the body). Quality components, sure, but a lot is in the circuit design for longevity, so plenty of margin in the designs and back-up systems everywhere. Flight hardware actually tended to be harder to design, mainly due to the wide temperature range (-55 to 125 degC, some to 140+ degC; space components I designed were often -40 to 85 degC, some to 125 degC) and generally rough environment. Like vibration; space stuff wasn't usually constantly shaking once in orbit, just had to survive take-off, and the satellite electronics were off until deployment so a mechanical rather than electrical issue. Flight hardware (like fighter jets, missiles) is constantly subject to high-G stress, all sorts of vibration, and rapid temperature changes (jets and missiles get from dessert deck to high altitude very quickly).
If you can see them partially working, it's probably because you can see the individual LEDs. LED signal displays have used diffusion covers for some years now, so if you can see individual LEDs, the indication is already some years old.
Most agencies think LED signal indications have a five-year lifespan. That compares to incandescent bulbs that have an 8000-hour mean lifespan (with a lot of variability). In the days of incandescent bulbs, many agencies routinely replaced the bulbs every year. Believe me, that's expensive, but it's not as expensive as replacing them as they burned out. Even in the early days of LED traffic signals, the indication used several circuits of LEDs so that if one failed the color would still be displayed. They were more subject to damage from bullet holes than anything else.
I am not guessing when I say that the lifecycle costs of LEDs in traffic signal service is much, much lower than incandescent bulbs. I'm also not guessing when I assert that the cost of maintaining traffic signals, as a portion of the budget of the agency at large, is a percentage so low that if it was THD people would line up for the amp. Go look for waste somewhere else--most agencies have cut their traffic signal budgets so low their maintenance struggles to stay afloat, let alone have enough left over to kill budgets.
Residential LED bulbs have improved enormously in recent years, but they still have mini-power supplies in them made from price-point components and that's what fails. The power supplies in traffic-signal indications are more robustly designed, but heat is still the enemy, and South Carolina is more hot than cold.
It is true that LEDs don't generate enough heat to melt blown snow, and that is an issue with several different workarounds.
By the way, my Buckeye Hypex NC502MP Class D amp is still going strong with no issues.
Rick "who spends a LOT less time changing even residential bulbs now than when using incandescent bulbs" Denney
The Hypex SMPS is the same basic design and comparable form factor as the MP series amp modules.
If there is a problem with reliability, they could get hit on both fronts. If SMPS, Purifi goes along for the ride too. I know @restorer-john has mentioned some other questionable practices and could add a few more, but dealt with reps for both companies and they are responsive and seem to care. Doubt an engineer like Bruno would have his new company rely on product from Hypex if he thought they were untrustworthy.
There are many other companies than Hypex and Purifi doing Class D amps. To the OP’s question, is Class AB better, and if so, which vendors? We have Benchmark, but is really a higher end offering. Before I bought Hypex, was seriously considering Outlaw’s monoblock. Any others?
If there is a problem with reliability, they could get hit on both fronts. If SMPS, Purifi goes along for the ride too. I know @restorer-john has mentioned some other questionable practices and could add a few more, but dealt with reps for both companies and they are responsive and seem to care. Doubt an engineer like Bruno would have his new company rely on product from Hypex if he thought they were untrustworthy.
There are many other companies than Hypex and Purifi doing Class D amps. To the OP’s question, is Class AB better, and if so, which vendors? We have Benchmark, but is really a higher end offering. Before I bought Hypex, was seriously considering Outlaw’s monoblock. Any others?
The OP and several other posts contrasts the reliability of class D amps to class AB amps. The thing is, the type of amp is completely irrelevant when it comes to reliability - it's all down to the design goals of the manufacturer. You can make a very reliable class D amp and a very unreliable class AB amp, or the other way around.
As a point of reference, here's the insides of an industrial, 400 W (total), three channel "class D amplifier" (it's an Omron servo drive):
Servo drives are commodity items - they're mass produced by multiple manufacturers - and since the buyers are mainly large corporations with factories the focus is function, reliability (if one of these fails it's likely to halt a production line which is expensive, and as a result these are much more reliable than most consumer electronics!), cost, and also good, factually correct documentation.
You can actually buy these as an individual; the cost is about 1650€+VAT in single quantities, so in the same ballpark as our audio amplifiers.
What do you get for this money? Look at the heatsinking solution: a custom, beautifully cast aluminium heatsink makes up the main body of the unit, and although it's not visible in the photo I can assure you that there's no PCB under tension involved in securing the beefy switching elements to its milled bottom side! The capacitors are Nippon Chemi-Con, Nichicon and United Chemi-Con. (Ironically, this amp had a capacitor failure, but due to user error - 400 VAC applied instead of 230 VAC - so not due to any design issue.) There's also some pretty advanced DSP inside to control the current through the coils (stator) and the movement of the diaphragm (well, rotor - to millionths of a turn).
This is what's possible to manufacture when the customer is "aware" (focused on functionality, reliability, and cost - and more or less impervious to BS).
As a point of reference, here's the insides of an industrial, 400 W (total), three channel "class D amplifier" (it's an Omron servo drive):
Servo drives are commodity items - they're mass produced by multiple manufacturers - and since the buyers are mainly large corporations with factories the focus is function, reliability (if one of these fails it's likely to halt a production line which is expensive, and as a result these are much more reliable than most consumer electronics!), cost, and also good, factually correct documentation.
You can actually buy these as an individual; the cost is about 1650€+VAT in single quantities, so in the same ballpark as our audio amplifiers.
What do you get for this money? Look at the heatsinking solution: a custom, beautifully cast aluminium heatsink makes up the main body of the unit, and although it's not visible in the photo I can assure you that there's no PCB under tension involved in securing the beefy switching elements to its milled bottom side! The capacitors are Nippon Chemi-Con, Nichicon and United Chemi-Con. (Ironically, this amp had a capacitor failure, but due to user error - 400 VAC applied instead of 230 VAC - so not due to any design issue.) There's also some pretty advanced DSP inside to control the current through the coils (stator) and the movement of the diaphragm (well, rotor - to millionths of a turn).
This is what's possible to manufacture when the customer is "aware" (focused on functionality, reliability, and cost - and more or less impervious to BS).
restorer-john
Grand Contributor
Interesting it has no overvoltage protection for the control supply- the bridge rectifier has clearly got very hot, the cap has vented and likely the TNY smps controller is totalled too. Did the venting damage the digital board? I can see elctrolyte 'blast' on the connector shield can.
I repaired this amp to a working state (although obviously not for professional use). The only totalled components were the capacitor and a soldered-in fuse (so I guess they do have some overvoltage protection, at least to prevent fire). The transformer of the control SMPS has been pretty well cooked, but survived. The discoloration of the bridge rectifier was mostly electrolyte I think; it cleaned up well with some IPA. Interestingly, the capacitor has a plastic cap on it:
Not sure exactly about the purpose of that plastic cap - isolation between control voltage and IO board perhaps? Anyway, it acted as a kind of blast shield here!
I should add that the history of this amp and the cause of the failure are not known; overvoltage was a (hopefully) educated guess on my part. (It was sold as "new unused" on Ebay, lol. I did get my money back.)
Maybe you should get one of these on Ebay - perhaps one sold as for parts only - just to take it apart and enjoy the fabulous construction. They ooze quality in a way I never see in consumer items.
Not sure exactly about the purpose of that plastic cap - isolation between control voltage and IO board perhaps? Anyway, it acted as a kind of blast shield here!
I should add that the history of this amp and the cause of the failure are not known; overvoltage was a (hopefully) educated guess on my part. (It was sold as "new unused" on Ebay, lol. I did get my money back.)
Maybe you should get one of these on Ebay - perhaps one sold as for parts only - just to take it apart and enjoy the fabulous construction.
They ooze quality in a way I never see in consumer items.
Very similar to the devices I designed and project managed over a 35 year career. I worked on drives for AC induction machines - but the principles are very similar, including cost of failure. We designed devices going up to 250kW (Higher in other parts of the organisation)
These devices are typically three phase full wave uncontrolled rectification onto a DC link (Bulk capacitor), direct from mains supply. Protection against double+ the supply voltage is not really feasible. Also designed to be installed by qualified engineers.
On our 230V drives we used to put a "230V" label over the supply terminals so they couldn't be wired without removing (and hopefully looking at) the label.
Last edited:
This is one of my "class D" measuring systems via optical fibre.
It works reliably in this environment since 2006:
and for audio, it would perform like this () :
However, "distortion" requested is <1% only. There are quite different requirements, like BW (300kHz) and immunity to RF EMI fields in the lab.
It works reliably in this environment since 2006:
and for audio, it would perform like this (
) :However, "distortion" requested is <1% only. There are quite different requirements, like BW (300kHz) and immunity to RF EMI fields in the lab.
Like Nicola Tesla, right? But this is an independent and internationally accredited High Voltage Laboratory.
Similar threads
