Stabilizing DIY Mining Rigs: Tips on Component Repair and Heat Management

[Rupok]

Hi community,
After working on various DIY solar-powered mining setups, I’ve realized that long-term stability often depends on the quality of the components we use. In regions with high voltage fluctuations and humidity, standard electronics fail more often than we expect. I want to share some practical repair tips I've learned.


Common Failure Points in DIY Drivers & Converters:

1. Capacitor Longevity: Most budget 50W LED drivers or buck converters come with low-grade capacitors. In a 24/7 mining environment, these are the first to bulge. I’ve had much better results replacing them with 105°C rated Japanese capacitors. It adds a bit to the cost but saves the hardware from sudden shutdowns.

2. Upgrading Rectifiers (The 10A10 Advantage): If you're noticing inconsistent DC output, check your bridge rectifier or protection diodes. Replacing standard diodes with 10A10 diodes has significantly improved the durability of my solar-integrated circuits, especially when handling back-current issues.

3. Soldering and Humidity: In South Asia, humidity causes rapid oxidation of poor soldering joints. I highly recommend using a protective lacquer coating (conformal coating) over your repaired joints to prevent corrosion and "phantom" restarts of your mining rig

A question for the experts:
When scaling up DIY hardware, do you prefer active cooling (fans) for your converters, or have you found passive heatsinks to be enough if the ambient airflow is managed well?
I am sharing this based on my personal repair bench experience. I would love to get your feedback or hear about the common hardware failures you face in your setups.

[Cricktor]

In my opinion your topic fits better in board Bitcoin > Mining > Hardware and you can move it yourself if you agree or a moderator will do it (I haven't reported it to be moved).

What is your reason to post it here when it's mainly in the context of mining?

[Rupok]

In my opinion your topic fits better in board Bitcoin > Mining > Hardware and you can move it yourself if you agree or a moderator will do it (I haven't reported it to be moved).

What is your reason to post it here when it's mainly in the context of mining?

Hello @Cricktor, thank you for the feedback. I have moved the topic to the Mining support section as it seems most relevant to my hardware repair discussion. I'm still trying to understand the better placement for these technical topics, so I appreciate your help.

[Rupok]

In my opinion your topic fits better in board Bitcoin > Mining > Hardware and you can move it yourself if you agree or a moderator will do it (I haven't reported it to be moved).

What is your reason to post it here when it's mainly in the context of mining?

Thanks again for the merit, @Cricktor! I really appreciate the support. I'm taking your advice on board and will be more careful about selecting the right sub-board for future technical discussions.

[FP91G]

I'd like to add a few points.
Capacitors: Even on average-quality writing units, capacitors typically last for three years or more without any problems. I wouldn't buy such equipment and then have to replace the capacitors. There are a lot of counterfeit Japanese capacitors, diodes, and transistors out there these days.
Transistors require active cooling most of all; they can explode from overheating.

[Rupok]

I'd like to add a few points.
Capacitors: Even on average-quality writing units, capacitors typically last for three years or more without any problems. I wouldn't buy such equipment and then have to replace the capacitors. There are a lot of counterfeit Japanese capacitors, diodes, and transistors out there these days.
Transistors require active cooling most of all; they can explode from overheating.

Thank you for sharing your valuable insights! You made an excellent point about counterfeit Japanese capacitors and diodes; that’s a major issue in our local markets.

The difference between a standard 85°C and a 105°C rated Japanese capacitor is indeed huge for 24/7 mining stability. It's becoming increasingly difficult to source genuine parts locally, as the market is flooded with clones that look identical but fail under load.

Regarding transistors, your point about active cooling is vital. In my experience, even with large passive heatsinks, the thermal runaway can happen quickly in confined spaces. I’ve started using high-static pressure fans specifically for the power stages to mitigate this.

Your warning about transistors 'exploding' due to heat is something every DIYer should take seriously. I’ll make sure to prioritize better airflow and genuine components in my future setups. Appreciate the feedback.

[Nexus9090]

Capacitors are a lot more complicated than you first think.

Even 105°C rated capacitors may not perform the way you expect them to; many are only rated for 1000-2000hours at the rated temperature.

Just using a 105°C rated capacitor does not guarantee it will last longer or perform better.

Key factors you've overlooked include ESR (equivalent series resistance) of the capacitors, maximum ripple current and voltage and temperature de-rating.

High ESR and low ripple current ratings result in self heating of capacitors and thus increase the affects of aging.

Running capacitors at their rated voltage will also increase the affects of aging.

Depending on where the capacitor is used, it may not be beneficial to add a low ESR device as this can lead to instability in some SMPS and DCDC converters as it affects the loop compensation of the converter. Which increases conduction and switching losses in other switching elements like transistors, FET's and transformers; leading to their premature failures.

However, as a general rule of thumb for bulk capacitors in rectification

• Select a low ESR capacitor for electrolytic this will depend on the application
• Select a capacitor in the category of >5000h@105°C but be sure to run the device well within this operating temperature allow a margin of at least 20% So max 85°C, this will help extend the operating life of the part.
• Select a capacitor with the highest available ripple current rating that will physically fit into your application
• Select a capacitor with a higher operating voltage than you intend for it to be constantly used at. Typically this results in improvements in other characteristics also. So if you're running at 12V use a >=35V rated part, 24V use a >=63V and so on

Check the manufacturers datasheet for voltage and current de-rating. You will find that some capacitors only provide a small fraction of their capacitance at higher voltages and may reduce by as much as 80% So where you thought you had a 100uF capacitor you find at the rated voltage it may only be capable of providing 20uF of capacitance; which will increase ripple current and self heating.

Always check the datasheet for de-rating data.

Arbitrarily changing capacitors can make things a whole lot worse. Be sure you understand the circuit they are going into before you make a decision on what or if to change it.

[Rupok]

Capacitors are a lot more complicated than you first think.

Even 105°C rated capacitors may not perform the way you expect them to; many are only rated for 1000-2000hours at the rated temperature.

Just using a 105°C rated capacitor does not guarantee it will last longer or perform better.

Key factors you've overlooked include ESR (equivalent series resistance) of the capacitors, maximum ripple current and voltage and temperature de-rating.

High ESR and low ripple current ratings result in self heating of capacitors and thus increase the affects of aging.

Running capacitors at their rated voltage will also increase the affects of aging.

Depending on where the capacitor is used, it may not be beneficial to add a low ESR device as this can lead to instability in some SMPS and DCDC converters as it affects the loop compensation of the converter. Which increases conduction and switching losses in other switching elements like transistors, FET's and transformers; leading to their premature failures.

However, as a general rule of thumb for bulk capacitors in rectification

• Select a low ESR capacitor for electrolytic this will depend on the application
• Select a capacitor in the category of >5000h@105°C but be sure to run the device well within this operating temperature allow a margin of at least 20% So max 85°C, this will help extend the operating life of the part.
• Select a capacitor with the highest available ripple current rating that will physically fit into your application
• Select a capacitor with a higher operating voltage than you intend for it to be constantly used at. Typically this results in improvements in other characteristics also. So if you're running at 12V use a >=35V rated part, 24V use a >=63V and so on

Check the manufacturers datasheet for voltage and current de-rating. You will find that some capacitors only provide a small fraction of their capacitance at higher voltages and may reduce by as much as 80% So where you thought you had a 100uF capacitor you find at the rated voltage it may only be capable of providing 20uF of capacitance; which will increase ripple current and self heating.

Always check the datasheet for de-rating data.

Arbitrarily changing capacitors can make things a whole lot worse. Be sure you understand the circuit they are going into before you make a decision on what or if to change it.

Thanks for this detailed technical breakdown, You've raised some critical points that I hadn't fully considered, especially regarding ESR and Ripple Current.
It’s a great reminder that simply looking at the temperature rating isn't enough for 24/7 mining stability. The point about voltage de-rating and checking the manufacturer’s datasheet for actual capacitance at higher voltages is particularly eye-opening.
I’ll definitely be more careful about matching these specs and not just arbitrarily swapping components in the future. I really appreciate you taking the time to share this knowledge—it helps me (and others) improve our DIY repair standards.

[NotFuzzyWarm]

Points like that ^^ are the reason that so many BitAxes are crap. Being an open source design there are dozens of manufacturers in China and elsewhere making them for a myriad of resellers who told them to use the absolute lowest cost components with zero regard to their lifetimes. Even worse, many also seem to have 0 QC other than 'yes it worked before it was sent to a reseller we built them for'. There are a few good resellers that spec'd higher quality construction but they are not easy to find using simple searches.

Because the Official BOM in the OSMU git also specs parts based mainly on cost, when sidehack made & sold them he redesigned parts of the circuits and spec'd top quality parts picked with their application as the main driver. Even went as far as publishing the modified BOM & board layouts as req by the open source license. https://www.gekkoscience.com/product/bitaxe-gamma-gekkoscience-edition-plus/ and https://www.gekkoscience.com/product/gekkoaxe-gt-2xbm1370-solo-miner/
 The twits behind OSMU (owners of the design) still bitched about that.  

[FP91G]

Capacitors are a lot more complicated than you first think.

Even 105°C rated capacitors may not perform the way you expect them to; many are only rated for 1000-2000hours at the rated temperature.

Just using a 105°C rated capacitor does not guarantee it will last longer or perform better.

Key factors you've overlooked include ESR (equivalent series resistance) of the capacitors, maximum ripple current and voltage and temperature de-rating.

High ESR and low ripple current ratings result in self heating of capacitors and thus increase the affects of aging.

Running capacitors at their rated voltage will also increase the affects of aging.

Depending on where the capacitor is used, it may not be beneficial to add a low ESR device as this can lead to instability in some SMPS and DCDC converters as it affects the loop compensation of the converter. Which increases conduction and switching losses in other switching elements like transistors, FET's and transformers; leading to their premature failures.

However, as a general rule of thumb for bulk capacitors in rectification

• Select a low ESR capacitor for electrolytic this will depend on the application
• Select a capacitor in the category of >5000h@105°C but be sure to run the device well within this operating temperature allow a margin of at least 20% So max 85°C, this will help extend the operating life of the part.
• Select a capacitor with the highest available ripple current rating that will physically fit into your application
• Select a capacitor with a higher operating voltage than you intend for it to be constantly used at. Typically this results in improvements in other characteristics also. So if you're running at 12V use a >=35V rated part, 24V use a >=63V and so on

Check the manufacturers datasheet for voltage and current de-rating. You will find that some capacitors only provide a small fraction of their capacitance at higher voltages and may reduce by as much as 80% So where you thought you had a 100uF capacitor you find at the rated voltage it may only be capable of providing 20uF of capacitance; which will increase ripple current and self heating.

Always check the datasheet for de-rating data.

Arbitrarily changing capacitors can make things a whole lot worse. Be sure you understand the circuit they are going into before you make a decision on what or if to change it.

Capacitors rarely get hot. They usually get hot from heatsinks, chips, and transistors. I mined with video cards for a very long time, and the capacitors on my cards were exposed to hot heatsinks of 80 degrees Celsius or more for over five years. These cards were later sold to gamers, and I never had a single complaint. I'm not sure if these were Japanese capacitors.

[Nexus9090]

Capacitors are a lot more complicated than you first think.

Even 105°C rated capacitors may not perform the way you expect them to; many are only rated for 1000-2000hours at the rated temperature.

Just using a 105°C rated capacitor does not guarantee it will last longer or perform better.

Key factors you've overlooked include ESR (equivalent series resistance) of the capacitors, maximum ripple current and voltage and temperature de-rating.

High ESR and low ripple current ratings result in self heating of capacitors and thus increase the affects of aging.

Running capacitors at their rated voltage will also increase the affects of aging.

Depending on where the capacitor is used, it may not be beneficial to add a low ESR device as this can lead to instability in some SMPS and DCDC converters as it affects the loop compensation of the converter. Which increases conduction and switching losses in other switching elements like transistors, FET's and transformers; leading to their premature failures.

However, as a general rule of thumb for bulk capacitors in rectification

• Select a low ESR capacitor for electrolytic this will depend on the application
• Select a capacitor in the category of >5000h@105°C but be sure to run the device well within this operating temperature allow a margin of at least 20% So max 85°C, this will help extend the operating life of the part.
• Select a capacitor with the highest available ripple current rating that will physically fit into your application
• Select a capacitor with a higher operating voltage than you intend for it to be constantly used at. Typically this results in improvements in other characteristics also. So if you're running at 12V use a >=35V rated part, 24V use a >=63V and so on

Check the manufacturers datasheet for voltage and current de-rating. You will find that some capacitors only provide a small fraction of their capacitance at higher voltages and may reduce by as much as 80% So where you thought you had a 100uF capacitor you find at the rated voltage it may only be capable of providing 20uF of capacitance; which will increase ripple current and self heating.

Always check the datasheet for de-rating data.

Arbitrarily changing capacitors can make things a whole lot worse. Be sure you understand the circuit they are going into before you make a decision on what or if to change it.

Capacitors rarely get hot. They usually get hot from heatsinks, chips, and transistors. I mined with video cards for a very long time, and the capacitors on my cards were exposed to hot heatsinks of 80 degrees Celsius or more for over five years. These cards were later sold to gamers, and I never had a single complaint. I'm not sure if these were Japanese capacitors.

The capacitors used on graphics cards are typically low voltage solid polymer (POSCAP), tantalum electrolytic, niobium electrolytic or ceramic, none of these suffer the same aging and thermal issues that standard and extended range aluminium electrolytic capacitors do.

They are very different devices with vastly different characteristics.

[philipma1957]

Capacitors are a lot more complicated than you first think.

Even 105°C rated capacitors may not perform the way you expect them to; many are only rated for 1000-2000hours at the rated temperature.

Just using a 105°C rated capacitor does not guarantee it will last longer or perform better.

Key factors you've overlooked include ESR (equivalent series resistance) of the capacitors, maximum ripple current and voltage and temperature de-rating.

High ESR and low ripple current ratings result in self heating of capacitors and thus increase the affects of aging.

Running capacitors at their rated voltage will also increase the affects of aging.

Depending on where the capacitor is used, it may not be beneficial to add a low ESR device as this can lead to instability in some SMPS and DCDC converters as it affects the loop compensation of the converter. Which increases conduction and switching losses in other switching elements like transistors, FET's and transformers; leading to their premature failures.

However, as a general rule of thumb for bulk capacitors in rectification

• Select a low ESR capacitor for electrolytic this will depend on the application
• Select a capacitor in the category of >5000h@105°C but be sure to run the device well within this operating temperature allow a margin of at least 20% So max 85°C, this will help extend the operating life of the part.
• Select a capacitor with the highest available ripple current rating that will physically fit into your application
• Select a capacitor with a higher operating voltage than you intend for it to be constantly used at. Typically this results in improvements in other characteristics also. So if you're running at 12V use a >=35V rated part, 24V use a >=63V and so on

Check the manufacturers datasheet for voltage and current de-rating. You will find that some capacitors only provide a small fraction of their capacitance at higher voltages and may reduce by as much as 80% So where you thought you had a 100uF capacitor you find at the rated voltage it may only be capable of providing 20uF of capacitance; which will increase ripple current and self heating.

Always check the datasheet for de-rating data.

Arbitrarily changing capacitors can make things a whole lot worse. Be sure you understand the circuit they are going into before you make a decision on what or if to change it.

Capacitors rarely get hot. They usually get hot from heatsinks, chips, and transistors. I mined with video cards for a very long time, and the capacitors on my cards were exposed to hot heatsinks of 80 degrees Celsius or more for over five years. These cards were later sold to gamers, and I never had a single complaint. I'm not sure if these were Japanese capacitors.

The capacitors used on graphics cards are typically low voltage solid polymer (POSCAP), tantalum electrolytic, niobium electrolytic or ceramic, none of these suffer the same aging and thermal issues that standard and extended range aluminium electrolytic capacitors do.

They are very different devices with vastly different characteristics.

also most gpus are made by a "real" company vs a fly by nite asic builder.

Asus has been around for over ten years
PNY
MSI
Sapphire

all of them have a reputation to try to protect and are likely to have better qc on parts.

Note I listed a few companies not all.

But I have own hundreds of gpus maybe 1000 my failure rate is under 2% other than fans.

and these gpus ran at 80 tdp watts for over 2 or 3 years.

[Nexus9090]

-SNIP-

also most gpus are made by a "real" company vs a fly by nite asic builder.

Asus has been around for over ten years
PNY
MSI
Sapphire

all of them have a reputation to try to protect and are likely to have better qc on parts.

Note I listed a few companies not all.

But I have own hundreds of gpus maybe 1000 my failure rate is under 2% other than fans.

and these gpus ran at 80 tdp watts for over 2 or 3 years.

Very true PC hardware manufacturers get crucified if they don't maintain highest of standards, it costs them millions if they don't successfully uphold their reputation.

They will as you say maintain very strict quality controls over what parts are used and where they're manufactured; especially where graphics cards are concerned, they'd loose their license from the likes of AMD and NVidia if they produce junk since its not just their own reputation at stake.

Even though companies like BitMain have a good reputation for ASIC gear, I can tell you from my own experience with them that their build quality is poor at best. on the few boards S17 and S19 I've had to repair the solder quality has been abysmal. Its clear they use the lowest cost components and cheap out on manufacturing quality so they can to maximize their return.

Their ASIC packages aren't ideal either, I've mentioned this before elsewhere that they are prone to moisture ingress, so if they're not handled properly prior to assembly you cant guarantee they'll last.

I've not had to work on any other ASIC machines, so cant comment on the build quality elsewhere with the notable exception of a few BitAxe boards which have been pretty hit and miss on build quality.