Analysis of power supply circuit of BP6
and stability check of various voltages

I analyzed the power supply circuit of BP6. It was very interesting for me.
Swinging of the VTT voltage was discussed. But there were several other voltage descents and swingings.

Vcore generation part

These four MOSFET(a-b, c-d) generate Vcore.
The first step(a, c) generates 5V, and the second step generates the core voltage.
There are differences between rev 0.4 and 0.41.
Rev 0.4 has better Vcore stability than Rev 0.41.
Rev 0.4 is only 0.2V descent. Rev 0.41 is 0.5-0.6V descent while running heavy application such as 3DMARK2000.
The MOSFET(b) supplies Vcore to W83782D tip(for XBUS)
In the case of Rev 0.41, I think the modification is necessary.

VTT (Vcc 1.5) generation part

This regulator(s) is already changed. to LM338T.

Vcc 2.5V generation part

This regulator(e) is the same as Vtt regulator.
This 2.5 voltage descents 0.1 - 0.15V while running heavy application.
The voltage descent of rev 0.4 is larger than that of rev 0.41.
I think this regulator must be changed to LM338T.

3.3Ｖ generation part1 (for chips)

The first step of 3.3V is provided by two LM1084 regulators(f, g).These voltages are joined to another regulator for memory via coupling with registors (R220, R221).
Output voltage at these regulators( f, g) drops still. (under 3.0V)
To get the reference voltager quickly, the circuit will be changed. This modification will give the best load regulation.
This voltage supplies to clock generator(p), HPT366(q), BXchip and PCI chip(r).

3.3Ｖ generation part2 (for memory)

If the first 3.3V decreases, the second 3.3 voltage(h) for memory decreases.