Class D amp long term reliability

[dlaloum]

There were several AVR models where DSP chips failed, but heat wasn't the culprit; rather, it was defective DSP chips and poor soldering—issues that subsequently led to thermal problems.
One of the most well-known examples of this involved the Pioneer AVR series LX-73/83/75/85/76/86. This caused lasting damage to Pioneer's reputation, as the company failed to address the issue, leaving €2,500 AVRs with nothing but scrap value after just 3 to 5 years.

One also needs to look more closely at capacitor failures in AVRs.
Especially in the past—roughly from the 90s to around 2010—capacitors used in switching regulator applications (DC/DC converters, power supplies, Class-D amplifiers, SMPS units, etc.) were frequently underspecified or incorrectly rated; consequently, they would overheat and fail. This was easily identified by bulging tops or electrolyte leaking from the base, often occurring within just a few months to a few years. Again, this had nothing to do with high ambient temperatures.

In both instances, uninformed people spread the idea that thermal issues were causing the failures—a claim that is absolute nonsense and technically incorrect regarding the root cause.
The same thing happened both here on the forum and elsewhere—for example, with the Topping PA5 amplifiers (first version), which were prone to failure. In those cases, too, there was never a heat problem.

Do you see the pattern?

I understand what you are saying - what I in turn am pointing out, is that even with bad batches of capacitors, when the temperature is kept low, their life is extended.

The bad batch of capacitors that went to several manufacturers around 2008 led to AVR failures.

So you can say that the capacitor manufacturer was to blame (yes they were) - but if the AVR's had been designed to run at lower operating temperatures, then the bad capacitors would probably not have failed - the flaw of the bad capacitors meant that effectively rather than being a 75C design they were a 65C design in reality - and that due to the AVR's running "HOT" close to the spec margin, the capacitors ended up running above the temperature where they would degrade.

So there are multiple contributing factors - but heat is definitely one of them - and if you drop the temperature below 50C - most of those AVR's would still be working fine today.

 

[whazzup]

In both instances, uninformed people spread the idea that thermal issues were causing the failures—a claim that is absolute nonsense and technically incorrect regarding the root cause.
The same thing happened both here on the forum and elsewhere—for example, with the Topping PA5 amplifiers (first version), which were prone to failure. In those cases, too, there was never a heat problem.

Do you see the pattern?

Yes, the pattern is pretty clear that you are consistently dismissive of users who buy these amplifiers and fear that heat related issues might shorten the lifespan of the amp (not the chip, the entire amp as a product) and try to improve heat dissipation.

Question to Google AI: "Topping pa5 problems were not heat related, is that true?"
Answer: https://share.google/aimode/knJ1HGff9t0FAXL6c

That statement is partially true, but mostly a matter of semantics. The root cause of the original Topping PA5 failures was triggered by heat, but the heat didn't fry or burn out the electronic chips themselves. [1, 2]
Instead, users and community reverse-engineers on platforms like Audio Science Review found that the failure was mechanical and structural: [1, 2]

• • The Potting Compound Issue: Topping enclosed the input/op-amp circuitry (the D01 "black box" module) in a hard resin potting compound to protect their design from being copied. [1, 2, 3]
  • Differential Thermal Expansion: As the internal components warmed up during normal use and cooled down when turned off, the hard resin and the underlying circuit board expanded and contracted at different rates. [1]
  • The Result: This constant physical stress from temperature changes sheared off tiny surface-mount resistors or cracked fragile solder joints inside the solid block. [, 2]
  • The Evidence: Simply melting/reflowing the solder joints or removing the resin entirely often brought "dead" modules back to life without replacing any components, proving the underlying silicon chips were thermally undamaged. [1, 2, 3]

So, while it is true that heat didn't melt the electronics, heat cycling (thermal expansion) was the physical force that caused the joints to break. [1, 2]

 

[somebodyelse]

I used off the shelf pipes. I have not researched doing my own, but take into account that the inside is a sort of foam to increase the surface area, similar to electrolytic capacitors electrodes. I see two challenges to diy one: to get that in a controlled way, and to close them without blowing everything. But maybe it is doable.... In any case straight ones, they are super cheap. What I did was to use a radiator in which the flat pipes fit without having to solder them (what would be another challenge) and a fan to cool the radiator

The porous inner surface is there to return liquid to the hot end via capillary action when gravity can't do the job, making them less sensitive to orientation. I made one some years back just to see if I could, not for any practical reason.

 

[EJ3]

heat cycling (thermal expansion) was the physical force that caused the joints to break.

So, if you never turned it on, it would not have HEAT cycled and would have never failed:
And IF you did turn it on but did something to mitigate the HEAT part of the cycling
(say, causing the HEAT part to be lower by 10 C), you are saying that it would not have lasted longer,
despite having lower HEAT in the cycling.
Therefore HEAT was not the reason for the failure?
I'm not sure about that.
Would you please enlighten me about how and why HEAT was NOT any part of the issue?

 

[whazzup]

So, if you never turned it on, it would not have HEAT cycled and would have never failed:
And IF you did turn it on but did something to mitigate the HEAT part of the cycling
(say, causing the HEAT part to be lower by 10 C), you are saying that it would not have lasted longer,
despite having lower HEAT in the cycling.
Therefore HEAT was not the reason for the failure?
I'm not sure about that.
Would you please enlighten me about how and why HEAT was NOT any part of the issue?

The question was phrased from roland68's perspective as I understand it. I asked AI and it said it's only partially true.

I'm in the 'less heat and more active cooling is better' camp, while roland68's perspective seems to be 'but the chip has no problem with 70 degrees' without acknowledging that some users are more concerned about not just the chip but whether the entire amp is affected by constant heat, because even if one capacitor were to fail, for most of us that's it.

 

[TheWalkman]

There were several AVR models where DSP chips failed, but heat wasn't the culprit; rather, it was defective DSP chips and poor soldering—issues that subsequently led to thermal problems.
One of the most well-known examples of this involved the Pioneer AVR series LX-73/83/75/85/76/86. This caused lasting damage to Pioneer's reputation, as the company failed to address the issue, leaving €2,500 AVRs with nothing but scrap value after just 3 to 5 years.

One also needs to look more closely at capacitor failures in AVRs.
Especially in the past—roughly from the 90s to around 2010—capacitors used in switching regulator applications (DC/DC converters, power supplies, Class-D amplifiers, SMPS units, etc.) were frequently underspecified or incorrectly rated; consequently, they would overheat and fail. This was easily identified by bulging tops or electrolyte leaking from the base, often occurring within just a few months to a few years. Again, this had nothing to do with high ambient temperatures.

In both instances, uninformed people spread the idea that thermal issues were causing the failures—a claim that is absolute nonsense and technically incorrect regarding the root cause.
The same thing happened both here on the forum and elsewhere—for example, with the Topping PA5 amplifiers (first version), which were prone to failure. In those cases, too, there was never a heat problem.

Do you see the pattern?

Someone correct me if I’m wrong, but as the owner of a Gen 1 PA5, I thought the problem with the PA5 was, ultimately, heat: Topping potted their op-amps to keep competitors from determining what their secret sauce was, causing the op-amps to overheat and in many cases, fail.

Am I mistaken in my understanding the failure mechanism was ultimately heat in the old PA5?

 

[whazzup]

Simple guidelines from Arendal:

Why Amplifiers and AV Receivers Get Warm?

Amplifiers and Audio/Video Receivers (AVRs) are essential components in enhancing the audio experience in various entertainment systems. However, they often get warm during operation. Here’s why this…

arendalsound.com

Precautions Regarding Heat​

To ensure that your amplifier or AVR operates safely and maintains its lifespan, consider the following precautions regarding heat management:

1. Ventilation: Ensure that your device has adequate ventilation. Most devices have vents either on the top or sides; these should not be obstructed. Placing the device in a cabinet or shelf with good airflow is ideal. Avoid stacking other electronic equipment directly on top of or underneath amplifiers and AVRs.
2. Environment: Avoid placing the amplifier or AVR in an already warm environment. Increased ambient temperatures can exacerbate the heating issues and push the device beyond its safe operating temperature.
3. Regular Cleaning: Dust accumulation can block vents and insulate components, leading to increased internal temperatures. Regularly cleaning the vents and the area around your amplifier or AVR can help mitigate this issue.
4. Use of Cooling Fans: For high-power devices or those used in confined spaces, additional cooling fans can be employed to help dissipate heat more effectively.
5. Spacing: Ensure there is enough space around the amplifier or AVR for heat to dissipate naturally. This includes not only horizontal spacing around the sides but also vertical spacing if the device is placed in an entertainment center or similar enclosure.

 

[stuartkohl]

The question was phrased from roland68's perspective as I understand it. I asked AI and it said it's only partially true.

Please don't use "AI" for this type of thing. It's just not reliable for answering questions that require factual answers.

 

[whazzup]

Please don't use "AI" for this type of thing. It's just not reliable for answering questions that require factual answers.

It's a good aggregator of online information. In certain cases it's probably more accurate than a random individual purporting 'many years of experience' posting on a forum. Anyway the sources the AI is drawing from are included, so anyone can drill deeper and pinpoint any inaccuracies if so desired.

 

[stuartkohl]

AI always "wants" to give you an answer. It's not designed to say, "I don't know," or "I'm not sure."

Yes, they often train the models with online information. GIGO.

Random individual on a forum? But that's often exactly what the LLM has been trained on.

AI certainly has its uses, but don't ask it questions you don't already know the answer to.

 

[DavidR]

It's a good aggregator of online information. In certain cases it's probably more accurate than a random individual purporting 'many years of experience' posting on a forum. Anyway the sources the AI is drawing from are included, so anyone can drill deeper and pinpoint any inaccuracies if so desired.

I have often used google before AI for technical/audio searches as well as others, then of course evaluated any results as to whether or not the source is/appears to be credible, often was a university or technical web site. Recently, however, some results have been surprsinging good and detailed in ways that were a summary. Often that was enough, but as you said the sources are included. Quite frequently those have been AudioScienceReview, although even then it would be a specific post by a specific member that, occasionally, was not necessarily the best answer. But that often provided a link to a post in a thread that had coverage of my question by other very quaified members here.

Hit or miss, though, as it was pre-AI. Ai seems to be an improvement depending on the topic.

 

[Roland68]

It's a good aggregator of online information. In certain cases it's probably more accurate than a random individual purporting 'many years of experience' posting on a forum. Anyway the sources the AI is drawing from are included, so anyone can drill deeper and pinpoint any inaccuracies if so desired.

AI always "wants" to give you an answer. It's not designed to say, "I don't know," or "I'm not sure."

Yes, they often train the models with online information. GIGO.

Random individual on a forum? But that's often exactly what the LLM has been trained on.

AI certainly has its uses, but don't ask it questions you don't already know the answer to.

Most people don't understand how standard AIs (not the specialized ones used in professional fields) work.
The operating principle is quite simple: millions of flies can't be wrong—shit tastes good.

In reality, ASR has—at least to some extent—an astonishingly large influence on certain AI responses in the Hi-Fi world. Nevertheless, much of that is only half-true or completely wrong, as was the case with the PA5.

 

[SigbergAudio]

On the topic of AI; personally I go to forums to discuss with human beings. If I wanted an AI answer or comment on a topic, I could just ask an AI directly.

In summary: I appreciate if people reply with their own words and ideas, not copy and paste AI.

 

[Roland68]

Yes, the pattern is pretty clear that you are consistently dismissive of users who buy these amplifiers and fear that heat related issues might shorten the lifespan of the amp (not the chip, the entire amp as a product) and try to improve heat dissipation.

Question to Google AI: "Topping pa5 problems were not heat related, is that true?"
Answer: https://share.google/aimode/knJ1HGff9t0FAXL6c

Someone correct me if I’m wrong, but as the owner of a Gen 1 PA5, I thought the problem with the PA5 was, ultimately, heat: Topping potted their op-amps to keep competitors from determining what their secret sauce was, causing the op-amps to overheat and in many cases, fail.

Am I mistaken in my understanding the failure mechanism of the old PA5?

No, that is completely incorrect.
Even under the worst possible conditions, I was unable to measure a temperature exceeding 60–70°C on the module itself (and that figure is an intentional exaggeration). Given the components used in the module, heat-related failure would not have occurred even after 20 to 30 years of operation.
After all, there are many first-generation PA5 units still running without issues today.

I have personally restored over 20 failed PA5 units, all of which have been running flawlessly ever since, without any further breakdowns.
In every single one of these cases, the module's potting compound was the root cause of the problem. Due to a lack of knowledge regarding the process, insufficient preparation, poor cleaning prior to potting, a lack of protection or insulation, and so on, leakage currents or creeping short circuits developed beneath the potting compound as a result of corrosion or chemical processes.
This explained the failure symptoms—such as noise, interference, and channels gradually dropping in volume until failing completely—which did not necessarily occur symmetrically across channels.

Once the cover and potting compound were removed, a thorough cleaning was often all that was needed to get the PA5 working again. Any torn-off components or damage were the result of the removal process itself. I have since optimized my removal method to the point where this has not happened with the last 10 modules.
Since then, I only resolder as a preventative measure.

 

[Panelhead]

Roland your analysis and repair match what I thought the issue is what the PA5. Same as the rack mount version that had a remote. The TP3 I think.
The thermal expansion differences between the circuit board and the potting agent is present in all potted products. Would be a much more widespread failure mode. Plus with the size of these boards the size changes are extremely small. The difference even smaller.
Hats off for identifying the issue, finding a solution, and implementing it. It is good to hear that these fine amplifiers are back in operation and not in a landfill.
How much do you charge to repair one?

 

[whazzup]

AI always "wants" to give you an answer. It's not designed to say, "I don't know," or "I'm not sure."

Yes, they often train the models with online information. GIGO.

Random individual on a forum? But that's often exactly what the LLM has been trained on.

AI certainly has its uses, but don't ask it questions you don't already know the answer to.

Precisely. So you look through the given content and links and decide whether those random individuals collectively are making sense or nonsense. AI gathers them. Hence 'aggregator of online information'. It is a tool.

No, that is completely incorrect.
Even under the worst possible conditions, I was unable to measure a temperature exceeding 60–70°C on the module itself (and that figure is an intentional exaggeration). Given the components used in the module, heat-related failure would not have occurred even after 20 to 30 years of operation.
After all, there are many first-generation PA5 units still running without issues today.

I have personally restored over 20 failed PA5 units, all of which have been running flawlessly ever since, without any further breakdowns.
In every single one of these cases, the module's potting compound was the root cause of the problem. Due to a lack of knowledge regarding the process, insufficient preparation, poor cleaning prior to potting, a lack of protection or insulation, and so on, leakage currents or creeping short circuits developed beneath the potting compound as a result of corrosion or chemical processes.
This explained the failure symptoms—such as noise, interference, and channels gradually dropping in volume until failing completely—which did not necessarily occur symmetrically across channels.

Once the cover and potting compound were removed, a thorough cleaning was often all that was needed to get the PA5 working again. Any torn-off components or damage were the result of the removal process itself. I have since optimized my removal method to the point where this has not happened with the last 10 modules.
Since then, I only resolder as a preventative measure.

Sure, I stand corrected on the Topping PA5 issue given your extensive experience on fixing them.

 

[TheWalkman]

No, that is completely incorrect.
Even under the worst possible conditions, I was unable to measure a temperature exceeding 60–70°C on the module itself (and that figure is an intentional exaggeration). Given the components used in the module, heat-related failure would not have occurred even after 20 to 30 years of operation.
After all, there are many first-generation PA5 units still running without issues today.

I have personally restored over 20 failed PA5 units, all of which have been running flawlessly ever since, without any further breakdowns.
In every single one of these cases, the module's potting compound was the root cause of the problem. Due to a lack of knowledge regarding the process, insufficient preparation, poor cleaning prior to potting, a lack of protection or insulation, and so on, leakage currents or creeping short circuits developed beneath the potting compound as a result of corrosion or chemical processes.
This explained the failure symptoms—such as noise, interference, and channels gradually dropping in volume until failing completely—which did not necessarily occur symmetrically across channels.

Once the cover and potting compound were removed, a thorough cleaning was often all that was needed to get the PA5 working again. Any torn-off components or damage were the result of the removal process itself. I have since optimized my removal method to the point where this has not happened with the last 10 modules.
Since then, I only resolder as a preventative measure.

Thanks for the clarification.

 

[EJ3]

The question was phrased from roland68's perspective as I understand it. I asked AI and it said it's only partially true.

I'm in the 'less heat and more active cooling is better' camp, while roland68's perspective seems to be 'but the chip has no problem with 70 degrees' without acknowledging that some users are more concerned about not just the chip but whether the entire amp is affected by constant heat, because even if one capacitor were to fail, for most of us that's it.

The AI doesn't know the answer. It just spouts out what is the most hear/say on the internet about something that you ask it.
It is just an aggregator of the prevalent opinion. And opinions are like buttholes: almost everybody seems to have one (right or wrong or partially both).
But real tech's (many of whom say that they are just hobbyist because [in my experience they tend to be humble about their EMPIRICAL] knowledge):
'Know the truth'.
Which many (possibly more often than not???) times does not align with AI.

 

[whazzup]

The AI doesn't know the answer. It just spouts out what is the most hear/say on the internet about something that you ask it.
It is just an aggregator of the prevalent opinion. And opinions are like buttholes: almost everybody seems to have one (right or wrong or partially both).
But real tech's (many of whom say that they are just hobbyist because [in my experience they tend to be humble about their EMPIRICAL knowledge):
'Know the truth'.
Which many (possibly more often than not???) times does not align with AI.

Yup, and it's a fantastic tool for that.

 

[umr]

The AI doesn't know the answer. It just spouts out what is the most hear/say on the internet about something that you ask it.
It is just an aggregator of the prevalent opinion. And opinions are like buttholes: almost everybody seems to have one (right or wrong or partially both).
But real tech's (many of whom say that they are just hobbyist because [in my experience they tend to be humble about their EMPIRICAL] knowledge):
'Know the truth'.
Which many (possibly more often than not???) times does not align with AI.

Many technical people can be very wrong as well.

I worked as an engineering consultant for decades and it was common for technical people to disagree about solutions to problems or root cause analysis.

AI is a useful tool to search for information and evaluate options. It can also find information you may have missed by its ability to explore more or different resources. The quality of the results is heavily influenced by the accuracy of the data and AI's interpretation of that information. I have had it present options or viewpoints that were useful.

I also find AI useful in reviewing large amounts of data for small samples of anomalies.

I personally pit multiple AI tools against each other to look for errors in their responses.

It is up to the user to decide what to do with this type of information.