SEGA Dreamcast Pix3lmods PSU power and temperature analysis
Index
To finalize the design of the power supply unit (PSU) for the Sega Dreamcast, we decided to carry out a power and thermal consumption analysis - two closely related aspects. For this purpose, we used a European Sega Dreamcast and a Pix3lmods PCB (first revision, which still includes integrated LDOs, although unpopulated and bypassed with a wire. See previous post for more details (SEGA Dreamcast PSU).
1. Power consumption analysis
Power consumption measurements were taken on the two main voltage rails: 3.3V and 5V. The expected current draws are approximately:
- 5V rail: < 1A (up to ~5W). Used by digital logic, the expansion port, legacy peripherals, and the GD-ROM drive.
- 3.3V rail: ~2.5–3 A (up to ~10W). Dedicated to the CPU, GPU, RAM, and most of the core digital components.
To perform the measurements, we used a basic multimeter. We desoldered the bypass between the LDO input and output pins, connected the multimeter leads, and set it to ammeter mode with the range set to amperes (A).
The game used for testing was Quake III Arena, as it's one of the most technically demanding titles in the Dreamcast library.
Figure 1. European Sega Dreamcast (left, center). Setup used to measure 3.3V rail current consumption.
Current consumption remained fairly stable across all typical usage scenarios: boot, main menu (game, memory, music selection), game loading, and in-game performance.
The 3.3V rail showed an average current consumption of 2.6 A, reaching up to 2.8 A (using the DMM’s HOLD function) while playing a Quake III match.
Figure 2. Measuring 3.3V rail current consumption in different usage scenarios.
We repeated the same procedure for the 5V rail. It quickly became clear that this rail draws significantly less current than the 3.3V rail. Average consumption was around 0.45 A, with spikes up to 0.6 A during disc read operations.
Figure 3. Measuring 5V rail current consumption in different usage scenarios.
2. Conclusions
In conclusion: The measurements align well with expectations on the 3.3V rail and remain safely within acceptable limits on the 5V rail.
Table 1 summarizes the measurement results and includes calculated power consumption on both rails, overall power drawn from the 12V input (12V 2A 24W wall-mount power supply), and power dissipated by each TPS62135 regulator, assuming an efficiency of approximately 90%.
Table 1. Power consumption measurements.
3. Temperature analysis
For the thermal analysis, we used the same multimeter with a thermocouple probe. The thermocouple was placed directly on top of the TPS62135 regulator for each power rail and secured carefully using Kapton tape.
Figure 4. Thermocouple placed on top of the TPS62135 3.3V rail regulator.
The initial temperature of the regulator was around 25 °C (ambient temperature, TA). During the test, the temperature of the 3.3 V regulator gradually increased to a maximum of approximately 65 °C, reached during a gaming session.
Figure 5. Measuring 3.3V TPS62135 regulator temperature in different usage scenarios.
For the 5 V rail, the same procedure was followed. In this case, the temperature remained stable at around 35 °C, which is consistent with the low current consumption previously observed on this rail (~0.5 A).
Figure 6. Thermocouple placed on top of the TPS62135 5V rail regulator.
To determine whether these temperature increases are within the expected range, we compared our readings with the thermal specifications from the TPS62135 Datasheet (Table 7.4, also shown in Figure 7).
Figure 7. TPS62135 thermal information.
In particular, we used the junction-to-case (top) and junction-to-ambient thermal resistance values to estimate the theoretical surface temperature of the device.
A good explanation of the thermal model can be found at:
https://www.analog.com/en/resources/design-notes/package-thermal-analysis-calculator-tutorial.html
Figure 8. An industry-standard thermal model and formulas for integrated circuit devices.
Table 2 summarizes the theoretical temperature rise for each voltage rail, calculated based on the power dissipated in each scenario.
Table 2. TPS62135 theoretical temperature increase calculations.
In our case, for the 3.3 V rail, we measured 64 °C — a realistic value that closely matches the theoretical estimate. It's worth noting that the PSU is mounted directly above the console's motherboard, and the internal ventilation is somewhat limited.
For the 5 V rail, the measured value was around 35 °C, while the expected case temperature was approximately 33 °C.
4. Conclusions
The TPS62135 regulator is rated for junction temperatures (TJ) up to 125 °C. Our measurements are well below that limit (roughly half), meaning the regulator operates within a thermally safe zone, which also results in reduced long-term component aging.