From LogiWiki
Jump to navigation Jump to search

piernov's FAQ

Why INTRUDER# measures 3.0V or lower on the multimeter?

INTRUDER# has a very weak pull-up (1MΩ). If you touch the probes, skin resistance is enough to pull down the voltage a lot (like less than 1.5V). Even without touching the probes, the internal impedance of the DMM (typically around 10MΩ) will cause it to drop to around 10MΩ*3.3V/(10MΩ+1MΩ) = 3.0V.

Can I use a different SPI ROM part number?

Better use the original part number.
If you can't get it and you have an Intel platform, use UEFITool to check what part numbers are registered in the VSCC table and use one of these.
If you have an AMD platform, I'm not aware of something similar to the VSCC table, but compatibility could still be an issue. Be sure to check operating voltage, size, maximum frequency and dual/quad SPI mode.

I am missing VCCAXG/VCCGFX/VCCGT (iGPU core voltage), how do I solve it?

The iGPU core voltage is almost never relevant. It typically only comes up when there's a significant load on the iGPU. It can often be seen spiking a couple of times on the oscilloscope when POSTing.

In other words, it's generally normal that it's missing. If you have some kind of issue with the machine, the cause is most likely elsewhere.

How can I read sensors on >=2015 Macs?/Where can I download HWSensors?

Latest version of HWSensors is available here as far as I'm aware:

SourceForge and other websites have older versions with worse support for 2015 and newer machines.

How can I select a replacement MOSFET if I can't find the original part?

It depends on the circuit it is in.

If it is just used as a power switch, eg. DC-in MOSFETs or MOSFETs turning on/off S0 rails sourced from S5 rails, then you mostly want to pay attention to these parameters:

  • Package and pinout (must be the same of course)
  • Vgs and Vds breakdown voltage (must be same or higher)
  • Id current rating (must be same or higher)
  • Rdson resistance (must be same or lower)

Vds is a hard requirement but it's rarely a problem, you just have to make sure it's at least 30V.

If they are part of a buck converter, they will be switching at a high frequency, so some other parameters are important too (in addition to the ones above):

  • Ci input capacitance (similar or lower)
  • Qg total gate charge (similar or lower)
  • Tdon turn-on delay time (similar or lower)
  • Tdoff turn-off delay time (similar or lower)
  • Body diode forward voltage (similar or lower)

These are especially critical in multiphase buck converters like CPU VCore and GPU VCore since the MOSFETs are stressed a lot, these buck converters must suddenly deliver dozens of amps when there's a load spike. Even up to a few hundred amps on high-performance desktop parts.

Thermal characteristics also play a small role in this so it's better to have a MOSFET with low thermal resistance overall. (this is often dictated by the package, but there are still variations)

Additionally, they must have the same part number across all phases in a multiphase buck converter, and preferably come from the same batch.

MOSFETs with higher current rating often have a lower Rdson, that's a good thing (less losses when turned on), but they conversely are often slower and harder to drive (higher input capacitance/total gate charge and slower turn-on/off delay times), so you can't got crazy on the current rating because it'll be counterproductive: the MOSFET will be much less efficient when switching and the MOSFET driver will have a hard time driving it.

Some MOSFETs (very often the low-side on synchronous buck converters) have a schottky-type body diode, that has a lower voltage drop than a "regular" body diode. (otherwise it'd be less efficient)