Voltage Scaling
Voltage Scaling is the process of increasing or decreasing the operating voltage supplied to ASIC chips to achieve a desired balance between performance, power consumption, energy efficiency, temperature, and stability. In Bitcoin mining, voltage is closely connected to chip frequency. Higher frequencies may require higher voltage for stable operation, while lower frequencies may allow ASIC chips to operate reliably at reduced voltage.
Voltage Scaling Explained in Simple Terms
ASIC chips need electrical voltage to perform SHA-256 calculations reliably.
If the voltage is too low for the selected operating frequency, the chips may become unstable.
If the voltage is higher than necessary, the miner may consume additional electricity and generate more heat without providing a proportional increase in hash rate.
The goal of voltage scaling is therefore to provide enough voltage for stable operation without using unnecessarily high settings.
A simplified relationship is:
Too Little Voltage → Instability
Suitable Voltage → Stable Hashing
Excessive Voltage → More Power and Heat
The ideal voltage depends on the ASIC chip, frequency, temperature, miner design, and tuning objective.
How Voltage Scaling Works
ASIC firmware and power-control hardware regulate the voltage supplied to mining chips.
A simplified tuning process works as follows:
An Operating Frequency Is Selected
The miner determines how fast the ASIC chips should operate.A Voltage Level Is Applied
The chips receive sufficient voltage to support the target frequency.ASIC Performance Is Monitored
Firmware observes hash rate, hardware errors, temperature, and stability.Voltage Is Adjusted
Voltage may be increased if chips are unstable or reduced if stable operation is possible at lower power.A Stable Operating Point Is Selected
The miner operates at a frequency-voltage combination appropriate for the tuning target.
The process can be summarized as:
Frequency Target → Voltage Setting → Stability Test → Adjustment → Operating Point
Example of Voltage Scaling in Practice
Suppose an ASIC miner operates at its standard frequency and voltage.
The operator wants to improve energy efficiency.
The voltage is gradually reduced while the miner's frequency remains within a range that the chips can support.
If the ASIC remains stable, power consumption may decline without an equivalent reduction in hash rate.
This can improve J/TH efficiency.
However, if voltage is reduced too far, the chips may begin producing errors or become unstable.
The operator must therefore find a balance between lower voltage and reliable hashing.
Voltage and Power Consumption
Voltage has a particularly important effect on semiconductor power consumption.
A simplified expression for dynamic power in digital circuits is:
Power ∝ Capacitance × Voltage² × Frequency
Because voltage appears as a squared term in this simplified relationship, relatively small voltage changes can have a meaningful effect on dynamic power.
However, real ASIC power consumption also includes leakage and other electrical losses, so actual miner behavior is more complex than this formula alone.
In practice, reducing voltage can significantly reduce power consumption when the ASIC remains stable.
Voltage and Frequency
Voltage and frequency are closely connected.
Higher clock frequencies generally require ASIC circuits to switch faster.
To maintain reliable operation, higher frequencies may require increased voltage.
This creates a common relationship:
Higher Frequency → Higher Required Voltage → Higher Power → More Heat
At lower frequencies, the chip may be able to operate at reduced voltage.
Therefore, ASIC tuning often adjusts frequency and voltage together rather than treating them as completely independent settings.
What Is ASIC Undervolting?
Undervolting means reducing the voltage supplied to ASIC chips compared with a higher-power operating configuration.
The objective is usually to reduce:
electrical consumption
heat generation
cooling requirements
If the chip remains stable, undervolting can improve mining efficiency.
For example:
Lower Voltage + Suitable Frequency → Lower Power → Better Potential J/TH
However, excessive undervolting can cause unstable hashing.
What Happens If ASIC Voltage Is Too Low?
Insufficient voltage can prevent ASIC circuits from operating reliably at the configured frequency.
Possible symptoms include:
increased hardware errors
unstable chips
reduced effective hash rate
missing or disabled chips
rejected work
miner resets or crashes
A miner may therefore report a high configured frequency while delivering poor effective performance if voltage is insufficient.
What Happens If ASIC Voltage Is Too High?
Increasing voltage can sometimes improve stability at high frequencies, but excessive voltage has disadvantages.
It can result in:
higher power consumption
increased heat production
higher die temperatures
increased cooling requirements
greater electrical stress
reduced efficiency
More voltage does not automatically produce more useful hash rate.
Once sufficient voltage is available for stable operation, additional voltage may provide little benefit.
Voltage Scaling and Energy Efficiency
Bitcoin mining efficiency is commonly measured in joules per terahash (J/TH).
Voltage scaling can be used to search for an efficient operating point.
Consider three simplified profiles:
Profile | Voltage | Hash Rate | Power | Main Objective |
Efficiency | Lower | Lower/Moderate | Lower | Better J/TH |
Standard | Standard | Standard | Standard | Balanced operation |
Performance | Higher | Higher | Higher | More hash rate |
The actual relationship varies significantly between ASIC models and individual chips.
Voltage Scaling and Heat
Nearly all electricity consumed by an ASIC miner ultimately becomes heat.
If voltage scaling increases miner power consumption, the cooling system must handle the additional thermal load.
This can increase:
die temperature
board temperature
fan speed
rack heat output
facility cooling requirements
The relationship can be simplified as:
Higher Voltage → Higher Power → More Heat → Greater Cooling Demand
This is one reason aggressive voltage settings require adequate cooling infrastructure.
Voltage Scaling and ASIC Autotuning
Advanced firmware may automatically search for appropriate voltage-frequency combinations.
A simplified autotuning process might be:
Select Frequency → Apply Voltage → Test Chips → Measure Errors and Efficiency → Adjust → Repeat
The firmware can attempt to find the lowest practical voltage that maintains stable operation at a particular frequency.
Depending on the system, tuning may occur:
per miner
per hashboard
per voltage domain
per chip or chip group
The exact level of control depends on ASIC hardware and firmware design.
Per-Chip Voltage Scaling
ASIC chips are not perfectly identical.
Because of manufacturing variation, one chip may require slightly more voltage than another to operate at the same frequency.
For example:
Chip A: stable at lower voltage
Chip B: requires more voltage
Chip C: unstable at the target frequency
Where supported, more granular tuning can account for these differences.
However, not every ASIC architecture provides independent voltage control for each individual chip.
Voltage Scaling and the Silicon Lottery
The silicon lottery describes natural variations between semiconductor chips produced from the same design.
These variations can affect:
voltage requirements
maximum stable frequency
power consumption
thermal behavior
energy efficiency
Some chips can operate at a given frequency using less voltage than others.
These chips may therefore achieve better efficiency under optimized tuning.
Voltage Scaling vs Frequency Scaling
Voltage scaling and frequency scaling are closely related but different.
Voltage Scaling
Changes:
electrical voltage supplied to ASIC chips.
Primarily influences:
stability
power consumption
heat generation
efficiency
Frequency Scaling
Changes:
ASIC operating clock frequency.
Primarily influences:
potential hashing performance
power demand
required voltage
In practical ASIC tuning, the two are usually considered together:
Voltage + Frequency = Operating Point
Voltage Scaling and Overclocking
ASIC overclocking often involves increasing frequency.
At some point, the standard voltage may no longer be sufficient for stable operation.
Additional voltage may then be required.
The progression can become:
Higher Frequency → Higher Voltage → Higher Hash Rate → Higher Power → More Heat
This creates practical limits based on:
chip capability
power supply capacity
cable and connector ratings
board power circuitry
cooling capacity
facility electrical infrastructure
Aggressive voltage increases can therefore create risks beyond the ASIC chips themselves.
Dynamic Voltage Scaling
Voltage settings can potentially change together with frequency as operating conditions change.
For example:
Efficiency Mode → Lower Frequency + Lower Voltage
Performance Mode → Higher Frequency + Appropriate Higher Voltage
Dynamic tuning may be useful when miners respond to:
electricity prices
power availability
thermal conditions
demand-response events
profitability changes
This allows the same ASIC hardware to operate under different power and performance profiles.
Risks of Incorrect Voltage Scaling
Improper voltage settings can cause:
unstable hashing
excessive hardware errors
overheating
excessive power consumption
reduced efficiency
power supply overload
electrical stress
unexpected shutdowns
Voltage should therefore be adjusted together with frequency, temperature, power limits, and hardware stability.
The objective is not the lowest or highest possible voltage, but a stable and efficient operating point.