Chinese made DC-DC converters are not made properly, they are passive using simple opamp feedback to produce simple fixed voltage but not suitable for CPU use. Not what you expect from a proper VRM with a IC that has more features.

Second, someone need resources to design and make 2 or 3 phase VRM with quality smaller inductors for more quieter ripple and faster response for high watts like K6-2 500MHz and up.

Cheers,

Re: Gigabyte GA-586HX (Rev. 1.53)

Here is that other project I couldn’t recall.

Re: Wacky experiments: running a K6-2 on Socket 5 without an interposer

Another

https://resources.pcb.cadence.com/blog/2019-w … ing%20regulator.

Some good reading about vrms.

Re: Socket 5/7/SS7 VOLTAGE Interposer.

Re: SoCat-57 a Socket-5-to-7 Interposer design

Re: Fixing up a Soyo SY-5TF Socket 7 Motherboard

IBM Personal Computer 365 / M/T: 6589-17U / Voltage Regulator Module (VRM)

Re: Socket 5/7/SS7 (Voltage Interceptor) Tweaker. (Beta)

CalamityLime made this VRM module for the socket 5/7 tweaker devices. It turned out really well. Huge thanks!!! It should also work for 3.3v and socket 1/2/3 tweaker as well.

Edit see updated version later

You need to have ground via next to V0ut via to keep power paths short. This is called floating condition and is bad electrical design because ground is now long paths back to the power supply about 50 cm long. Wrong wrong, trouble trouble! 😀

Cheers,

pentiumspeed wrote on 2022-06-26, 16:03:

You need to have ground via next to V0ut via to keep power paths short. This is called floating condition and is bad electrical design because ground is now long paths back to the power supply about 50 cm long. Wrong wrong, trouble trouble! 😀

Cheers,

We removed it because the ground is rather annoyingly located under the heatsink clip on the interposer. I suppose running it to a motherboard mount would accomplish the same thing or just using the pad it has.

The long ground path increases the resistance of the wiring. Didn’t figure it was enough to be significant though. But I guess low voltage DC has a habit of low transmission distance. Of course the power supply is adjustable.

I can add the ground path back I suppose 😀

It is not the resistance at DC, it is the current noise that thinking processor makes will travel back via highest resistance to ground and causes floating "grounds" on the DC-DC converter to be also noisy and therefore resulting noise travel through the DC-DC converter VRM power circuit to the CPU through VRM's capacitors.

In other words, the ground now have more resistance losses and now vibrates the VRM and appears on the VRM's output. This is biggest reason the datasheet *always* specify shortest or massive ground and equally massive power planes. Remember less voltage, more massive the current transients are. That's nature of ohm law.

Cheers,

pentiumspeed wrote on 2022-06-26, 20:43:

It is not the resistance at DC, it is the current noise that thinking processor makes will travel back via highest resistance to ground and causes floating "grounds" on the DC-DC converter to be also noisy and therefore resulting noise travel through the DC-DC converter VRM power circuit to the CPU through VRM's capacitors.

In other words, the ground now have more resistance losses and now vibrates the VRM and appears on the VRM's output. This is biggest reason the datasheet *always* specify shortest or massive ground and equally massive power planes. Remember less voltage, more massive the current transients are. That's nature of ohm law.

Cheers,

ah, good point, I’ve delt with that in solar design. Makes total sense. Good points. Thanks!

I ran into this too with noise back in the high school on my project, was around 1993. My set up was 4 small boards of few TTL plus 64 bits SRAM (iirc 7489), yes 64 bits total, input and output is 4 bits wide, giving 16 lines of 4 multiple choices each, pre-stored by optical sensors programming the thing with a correct paper then, and slide each respondents paper with clocking dot by each line for 16 clocking dots to make thing step through it's memory comparing using TTL chips and give correct or incorrect answers) chips lashed up with magnet wires each wires about 2 inches between 4 boards. Barely worked and noisy.

If I had thought this through by designing properly, this would worked better but that was very valuable experience.

Cheers,

Sphere478 wrote on 2022-06-26, 21:01:

pentiumspeed wrote on 2022-06-26, 20:43:

It is not the resistance at DC, it is the current noise that thinking processor makes will travel back via highest resistance to ground and causes floating "grounds" on the DC-DC converter to be also noisy and therefore resulting noise travel through the DC-DC converter VRM power circuit to the CPU through VRM's capacitors.

In other words, the ground now have more resistance losses and now vibrates the VRM and appears on the VRM's output. This is biggest reason the datasheet *always* specify shortest or massive ground and equally massive power planes. Remember less voltage coupled with less resistance in sink devices means more massive the current transients are. That's nature of ohm law.

Cheers,

ah, good point, I’ve delt with that in solar design. Makes total sense. Good points. Thanks!

Latest calamitylime versions

Re: Socket 5/7/SS7 (Voltage Interposer) Tweaker. (Beta)

Re: Socket 5/7/SS7 (Voltage Interposer) Tweaker. (Beta)

A nice project by ChrisK Switching regulator module as replacement for hot linear regulators (mainly used on PCI VGA cards)