P5B Deluxe OVP mod

[bartx]

Need help with ovp mod on this board. Actually 1,92-1,95V is max, I need more to push my Celly further At first pic:



Datasheet: http://www.dzjsw.com/jcdl/a/ADP3198.pdf

ADP3198 is the same IC as in P5K "Vanilla" board, link to topic:
http://www.xtremesystems.org/forums/...=145583&page=3

This resistor, which I marked is directly connected with 11 pin and with ground. So, my question is, if I'll solder parallel VR (I think that 1MΩ should be fine), max possible voltage should rise? Pencil mod also should be possible I guess.

[[XC] Oj101]

If that's correct, it should work with the Commando too, right?

[ceemic]

Yes, it's the same on Commando. I soldered there 1M VR, but I haven't had time to test it
Tho, one dude said that maybe we need to INCREASE resistance, not lowering it with VR. So that little resistor has to be changed to bigger one; or just removed and let a VR to do the job...
http://ceemic.pri.ee/hardware/comman...ando_mod06.jpg

[bartx]

This resistor must be replaced by bigger one, because I talked with one man, who is good at electronics Actually I haven't anything to replace it

[ceemic]

bartx, You can use just a VR. No resistor needed. 1M VR should work; just trim it down to ~300kohm and see if there was allready any gain. Then start increasing it more...

[celemine1Gig]

I just had a look at the datasheet. The resistor your marked should be Rlim, connected between Ilimit and GND.
The formula says:
Rlim = Vcl / (Alim * Ilimit)
--> Ilimit * Alim * Rlim = Vcl
--> Ilimit = Vcl / (Alim * Rlim)

That means that if you want Ilimit to increase, you have to decrease Rlim. And that means the first idea of connecting a trimmer in parallel should have been correct. I don't know why you think that you should increase Rlim. But maybe I'm missing something.

[bartx]

Just above this formula: "Thus, increasing Rlim now increases the current limit."

[ceemic]

celemine1Gig, tnx!

I'll test my mod this saturday and report back.

[celemine1Gig]

Just above this formula: "Thus, increasing Rlim now increases the current limit."

Honestly, this doesn't really make sense to me. Either the formula is wrong, or the sentence does not really make sense this way. Could be that "increasing" should actually mean decreasing.
I don't know, if I'm missing something, but it seems strange to me.

Or I don't know how to use maths. Who knows?

[bartx]

My board probably is going for sale, so I won't have any occasion to test this mod. I had serious issues with booting under cold - even on dice. I don't know what's wrong, every time when I'm benching on this board I've the same problem But it would be nice, if someone can test that

[loopy83]

Are there any news about this mod?

I tested the following mod, but it doent work! board died after 2-3s....


I think, after reading the datasheet we have to increase the resistor. So remove the existing one and replace it with a 500k VR....

to check, if its work, adjust the vr to the original value (arround 260-270k, measured on the removed res) and then adjust it to a higher level. when OVP will increase, its fine. when its lower, then we have to adjust the vr to the other direction

I will test it in the next weeks, hope to get good news and no other broken P5B

[blind_ripper]

thx for the news loopy .

[celemine1Gig]

BTW, where did you get the mod, posted above, from?

[loopy83]

out of this thread:
http://www.xtremesystems.org/forums/...t=ovp+commando

I also had a look to the datasheet and this kind of mod seems to be right. The goal is to increase the voltage on the ILIMIT Pin. So its logic to connect this pin to a higher potential over a vr.... but i tried this mod and my board died.

At this time I didnt know, that the ILIMIT pin provides a constant current, so the easier way is to modify the RLIM....

I will try to remove RLIM with arround 270k and replace ist with a 500k vr, adjusted to the same value for first test.... so both directions (higher RLIM and lower RLIM) are possible. In my opinion RLIM > 270k will do the job. When you are right (with decrese RLIM) , I am able to do that, too.

Maybe someone else can verify this kind of mod.... if not, I will try it in a fews days, because we have a big session on 09/06/26 and I want to test some 347 and 352. So a OVP mod is required for more then 1,9xV.

[persivore]

The problem limiting the voltage on the ADP3198 is not the current limit set by ILIMIT pin 11, it is the over-voltage protection generated internally in the device by comparing the Vcore (sensed on CSREF pin 15) to the VID+150mV. You can increase the Vcore over the OVP threshold by pulling the ground reference used by the VID DAC (FBRTN, pin 3) to a higher voltage than ground, effectively adding this voltage to the Vcore supply. There is usually a resistor between FBRTN and ground (on most boards I've seen using the ADP3198 anyway), so you can just connect a suitably sized resistor between FBRTN and a stable voltage supply. The 3.3v voltage rail is usually used, but it would be much better to use a low noise voltage reference if possible, so that you don't introduce unwanted noise into the Vcore supply.

[loopy83]

A very interesting point... did you test it on a similar board?

The overvoltage threshold is 250-350mV... the max VCore awailable in Bios is 1,7V... so all boards run into OVP arround 2V.

And you think, there is a pulldown resistor beetween FBRTN and GND?
But isnt it a bit risky to increase the ground voltage over 0V?
Because all other devices on the Board also use this reference 0V, so they can't work solid.

The only thing i could imagine:
The ADP3198 uses a seperate ground and is single-point connected to GND, to keep noise back from the GND plane. This is most realized with a 0Ohm resistor. So we could remove this resistor and connect a stable power supply arround 50-100mV to the GND of the ADP3198 chip. This should give a a boost arround 50-100mV. But i believe there is no single-point connection in this case.

O how could we prevent, to increase the GND level of the whole system/board?

Can we cut the track to pin 3 and solder instead of this track a stable 100mV power supply?

Thank you very much!

I hope you understand my bad english

PS: so the idea is, to solder a 50k VR from 3,3V to the ADP3198, not to pin 11 (as done before), but to pin 3... right?

[persivore]

I've used this method to overcome OVP on a number of boards (commando's, P5B vanilla's, P5K's) with no problems. I wouldn't expect there to be any problem as long as the voltage between GND and FBRTN is kept reasonably small (ie a few hundred mV). Looking at the block diagram of the ADP3198 on the first page of the datasheet, FBRTN is only used as the ground reference for the VID DAC, so hopefully it shouldn't effect anything else. I would guess that FBRTN is also used as the ground reference for the OVP/UVP compatators too tho, although its not shown.

On all the implementations of the ADP3198 I've come across, FBRTN has been connected to GND via small value resistor (around 22Ω IIRC), so there's no need to cut traces or remove any resistors etc. All you have to do is connect a suitably sized resistor between a stable voltage supply and FBRTN, creating a voltage divider with the resistor already on the board between FBRTN and GND. All the mod does is to raise the voltage of FBRTN slightly above GND, increasing the output of the VID DAC reletive to ground, increasing Vcore without triggering the OVP since the voltage between FBRTN and the VID DAC's output is still the same as that set in BIOS.

The system ground remains unchanged since you're only raising the voltage of FBRTN used for remote sensing of the CPU's ground supply, not ground itself.

Conencting a variable resistor between pin 3 (FBRTN) and +3.3v should be fine, as long as the 3.3v rail isn't too noisy. I usually use a variable resistor around 500Ω to 1K, but you would need to check the resistance between FBRTN and ground and choose a suitable variable resistor value based on that.

A very interesting point... did you test it on a similar board?

The overvoltage threshold is 250-350mV... the max VCore awailable in Bios is 1,7V... so all boards run into OVP arround 2V.

And you think, there is a pulldown resistor beetween FBRTN and GND?
But isnt it a bit risky to increase the ground voltage over 0V?
Because all other devices on the Board also use this reference 0V, so they can't work solid.

The only thing i could imagine:
The ADP3198 uses a seperate ground and is single-point connected to GND, to keep noise back from the GND plane. This is most realized with a 0Ohm resistor. So we could remove this resistor and connect a stable power supply arround 50-100mV to the GND of the ADP3198 chip. This should give a a boost arround 50-100mV. But i believe there is no single-point connection in this case.

O how could we prevent, to increase the GND level of the whole system/board?

Can we cut the track to pin 3 and solder instead of this track a stable 100mV power supply?

Thank you very much!

I hope you understand my bad english

PS: so the idea is, to solder a 50k VR from 3,3V to the ADP3198, not to pin 11 (as done before), but to pin 3... right?

[loopy83]

Ah I see... so I will measure the resistance beetween FBRTN and GND and then calculate a matching devider for my power supply.

I think I will use the hipro maximizer for this. Its a very stable and low noisy power supply and I can adjust the voltage of this power supply, too.

It sounds very good to me... i also checked the datasheet, so I understand what you meen

And when you tested this mod on commando's, P5Bs, I thin it would be fine.

I will be back with my calculations and maybe i test this Mod in the next days.

I will use our cascade to test it....

Thank you very much!

[bartx]

Where OVP hits on Commando? This board has 1.85v in bios, so ~2.1V should be possible I think if you say that treshold is 250-300 mV.

[loopy83]

OVP on my last commando was 2,02V and on my last P5B 1,95V.... so i think its ok.

Yesterday I measured the resistance beetween FBRTN and GND on this cap:


=> 20,9Ohm


When I use the Hipro maximizer, I can adjust the voltage from 1,8 to 4,2V.
So i dont need a VR resistor, to adjust the voltage on FBRTN.

Here is the mod. I think it works for OVP mod


So I can adjust the voltage on FBRTN pin from 72mV to 168mV. This should give me a boost from arround 150mV.

OVP at moment is 1,95V. With this Mod I am able to adjust to 2,1V.
I hope with -180°C on the board, this is ok for all other devices connected to VCore.

For first test I will measure the voltage on the FBRTN pin and adjust it with a VR to this level. When this works , I increase the Voltage slowly to 80mV and test, if the OVP is 80mV higher (arround 2V instead of 1,95V). If this is ok too, i use the maximizer to adjust the Overvoltage in the future. It is much easier to change the value with dipswitches

I Hope i can test it on weekend

PS: Be careful, the absolute max voltages on FBRTN pin is +-300mV !!!!

@persivore: is this the way you modded your commandos and P5Bs too? which power supply did you use? the 3,3V rail or another one?

[celemine1Gig]

BTW, what you circled in red is a capacitor and not a resistor.Might be connected in parallel to the real resistor(s), perhaps in order to get unwanted noise drained to GND (not sure though). And as long as FBRTN is only used for digital signals (VID etc.), like persivore said, it should be fine with a little voltage.

[loopy83]

Ups sorry.... I use the germen/europe symbol for a resistor.

Here in Germany/Europe the rectangle is a resistor. But you mean this USA/Japan symbol to be a resistor, right?



Sorry for that carelessness.... because XS is an international forum

[celemine1Gig]

I meant the red circle in the macro shot above. You circled a capacitor and wrote that it was the resistor between FBRTN and GND. That's a bit confusing, because someone could start to think that smd resistors look like that.

[loopy83]

Ah ok, I got it... thats my mistake
As shown in the datasheet, there is also a cap connected beetween GND and pin3. I could not find another component connected to pin3.

But i measured 20,9 Ohm... if there is only a cap, the resistance should be >1MOhm.

So i am a bit confused at moment... there must be a resistor to GND... but i cant find it.

But for the mod it isnt important, as long as the resistance beetween GND and Pin3 exists.
So my calculations are right and the mod should work... never mind, where the resistor is placed.

[persivore]

That looks fine. I've just used a variable resistor between +3.3v and FBRTN in the past.

There is usually a capacitor in parallel with the resistor between FBRTN and ground, which looks like what you've found.