Can't overclock CPU and run 3000MHz Ram at the same time?

[just_dave]

39 minutes ago, fastcar123 said:

Not that I do or dont belive you, because I really do appreciate the help, but I actually would like to see that data.

Mostly for curiosity sake.

 

I went to collage for electrical engineering and know a fair bit about mosfets. Whatever I dont know I can look up or just ask for explanation

I am using maximum stress case for example, this means Ryzen 7 at 1.4V 4GHz, 300KHz Fsw, Highside duty cycle 0.1 and lowside duty cycle 0.9.

Used mosfets are Niko semiconductor PK616BA for highside and two Niko PK632BA for the lowside.

The lowside is absolutely fine, the PK632BAs have relatively low Rds(on).

The highside however, that is the problem.

The Rise/Fall times are awful, it has 24ns Rise and 23ns fall.

The Rds(on) of this mosfet is 7miliohms, that is also pretty high (but not such a problem as the Tr/Tf).

These times lead to big switching losses and the Rds(on) does account for quite a bit of the resistive losses.

Overall heat output for Tomahawk type VRM is approx. 20W.

For comparison, the Asus 4+2/4+4 has between 17 and 18W of heat loss.

Note: Asus uses really good voltage controllers out of their own depot (ASP line) and you can set phase control (phase-by-phase, all of them at the same time), LLC control (5 steps) and switching frequency. (default 200KHz, i would keep it there since there is no voltage stability issue and it keeps the efficiency up)

[fastcar123]

1 hour ago, dave_k said:

-snip-

Interesting. I dont know that I have ever seen an option to change the FSW in the bios. Is that set automatically?

 

Now i do know that the the Tr and Tf being at 24nS is so slow. This would probably explain why i have voltage spikes and dips that are a bit out of the comfort zone. if you look at my voltages oduring the tests, yhou will notice that even though i have my voltage set to 1.3625V it still spikes up to 1.39-1.4. This is why i never ever run anything a tthe max safe voltage.

I am curious though. Do you have the Tr/Tf data for the asus prime board?

 

I admit that i dont know what the rds is. I will look this up when i have an oppertunity.

But i will say that an FET its supposed to be a short of 0ohms when fully closed (assuming they are in a switch configuration). Because I dont know what this is, I feel that 7 mOhms is decent.

 

Is the 20W TDP for the tomahawk board really all that bad compaired to the more higher end Asus board that has 18W TDP?

This could be another situation of me just being ignorant but with the comparison in mind it doesnt seem too big of a deal. Now with that said, I know that overclocking anything pretty much throws TDP that was specced by a manufacture out the window. I would think with an overclock that high, its probably outputting much more heat than just 20W.

 

I really enjoyed my Asus 970 board when i was running FX processors because of the really finite voltage controls. My MSI board just lacks the ability to fine tune voltage the way that i want. I thought that it was just a ryzen thing but I'm starting to think not so much.

[just_dave]

20 minutes ago, fastcar123 said:

Interesting. I dont know that I have ever seen an option to change the FSW in the bios. Is that set automatically?

 

Now i do know that the the Tr and Tf being at 24nS is so slow. This would probably explain why i have voltage spikes and dips that are a bit out of the comfort zone. if you look at my voltages oduring the tests, yhou will notice that even though i have my voltage set to 1.3625V it still spikes up to 1.39-1.4. This is why i never ever run anything a tthe max safe voltage.

I am curious though. Do you have the Tr/Tf data for the asus prime board?

 

I admit that i dont know what the rds is. I will look this up when i have an oppertunity.

But i will say that an FET its supposed to be a short of 0ohms when fully closed (assuming they are in a switch configuration). Because I dont know what this is, I feel that 7 mOhms is decent.

 

Is the 20W TDP for the tomahawk board really all that bad compaired to the more higher end Asus board that has 18W TDP?

This could be another situation of me just being ignorant but with the comparison in mind it doesnt seem too big of a deal. Now with that said, I know that overclocking anything pretty much throws TDP that was specced by a manufacture out the window. I would think with an overclock that high, its probably outputting much more heat than just 20W.

 

I really enjoyed my Asus 970 board when i was running FX processors because of the really finite voltage controls. My MSI board just lacks the ability to fine tune voltage the way that i want. I thought that it was just a ryzen thing but I'm starting to think not so much.

Its not that awful that it would explode in a huge mushroom cloud, worse than the Tomahawk is just the B350M Gaming Pro 3 phase and Gigabyte 4+3.

The bad thing is that the highside gets painfully overloaded.

Asus has the loss spread all across the mosfets.

Asus 4+2/4+4 uses mosfets by Onsemiconductor. One 4C09B for HS and two 4C06B for LS.

The 4C09B has 32ns Tr but only 6ns Tf. This means that overall switching time is nearly 10ns smaller than with the PK616BA.

Rds(on) is drain-to-source resistance, this mean resistance between source (of power) to drain (load) when the FET is fully on. Rds(on) on Asus highside is "just" 5.8 mOhms

The switching frequency on B350-F Strix by default is 200KHz.

From my overclocks and IR and sensor measurments, it runs the coolest and is on par with MSI B350 Carbon/Krait, but those dont have such advanced BIOS tuning.

Dont forget on the mosfet chain of death

lots of heat > high temp > lower efficiency > higher heat output > higher temp > lower efficiency

[fastcar123]

3 hours ago, dave_k said:

-snip-

So this is not an issue that affects all MSI boards. I was thinking it was all boards from MSI.

I may be misniterpreting here, but it sounds like comparitvly speaking ( aside from faster MOSFETS), the MSI board is not too terrible compaired to the Asus board. This leads me to think that maybe the b350 asus board would not be too bad.

The minor differeances are really not all that bad.

[just_dave]

3 minutes ago, fastcar123 said:

So this is not an issue that affects all MSI boards. I was thinking it was all boards from MSI.

I may be misniterpreting here, but it sounds like comparitvly speaking ( aside from faster MOSFETS), the MSI board is not too terrible compaired to the Asus board. This leads me to think that maybe the b350 asus board would not be too bad.

The minor differeances are really not all that bad.

They are minor on the paper.

Higher in reality.

Guy who has tomahawk and ran even lower OC than i did had over 65°C on the VRM.

14°C higher than me.

Asus has better VRM in all aspects than MSI Small 4+2.

More efficient

better cooled

more controllable.

Believe it or not but the difference is significant

MSI B350 Carbon/Krait use twice more the crappy highside, so the heat loss is spread between few mosfets and it runs as cool as Asus but still lacks the better control.