TheRealKiwi Posted September 27, 2021 Posted September 27, 2021 Nice done Mods, can you please describe how you made the vdimm mod? And how much voltage you can generate? Quote
ludek Posted September 27, 2021 Posted September 27, 2021 90% volt mods is to connect a resistor between FB pin (feedback) and ground. Sometimes there is a linear voltage regulator op-amp based, so you try whichever in- pin (there are 'in+', 'in-' and 'out' pins) to ground resistor. For example many socket A AsRock motherboards had LM358 as the part of NB voltage. Very often configuration is 3V3 rail -> mosfet (op-amp driven), so max VDIMM should be at least 3,1 Volt (resistance etc.). Quote
TheRealKiwi Posted September 27, 2021 Author Posted September 27, 2021 thanks for your very quick response! I know feedback mods, i indentified the LM324 as vDimm Controller but was confused about the Datasheet because there is no FB-Pin, so thanks for your description. I think i found a post of Lexer https://www.overclock.net/threads/the-socket-939-appreciation-club-and-knowledgebase-official.293448/page-1436#post-26399454 Am I right, that he used Pin 9 (Input 3-) to Pin 11 (GND)? Just for understanding: - Should it work to another GND-Point too? - the other Input/Output curcuits (input 1/2/4) can be other voltages (vDD or something) or not used? - if i modify that Input voltage to 5V instead of 3V3 Rail, can it work higher than 3,1V or i will fire it up? 3,0V is enough for my TCCD, only for research Quote
ludek Posted September 28, 2021 Posted September 28, 2021 Am I right, that he used Pin 9 (Input 3-) to Pin 11 (GND)? ~Looks like this, considering "ST" logo and the middle GND pin. - Should it work to another GND-Point too? ~Yes, but the closest is best (there is sometimes GND, Power-GND, Analog GND, so to be shure I recommend to connect to 2cm/1inch "as close as possible") - the other Input/Output curcuits (input 1/2/4) can be other voltages (vDD or something) or not used? ~yes, I remember AsRock had AGP, vDimmRef, vNB, vDimm on one quad-op amp like LM324, impressive - if i modify that Input voltage to 5V instead of 3V3 Rail, can it work higher than 3,1V or i will fire it up? 3,0V is enough for my TCCD, only for research ~in this configuration almost always the power componnent is a power SMD transistor. If your mems eat 1Amp or 2 amps, and vMem is 2.5, from 3V3 rail, the power dissapated on transistor is 3.3-2.5 over 1 Amp is (I=U/R so 3.3-2.5 is 0.8 x 1 is 0.8 W, it's not huge, but if you run full load, the current might jump to few amps) So 1A from 3.3V is 0.8W and from 5V is 5-2.5 x 1 = 2.5 W. This makes huge power losses and hot transistor. Quote
TheRealKiwi Posted September 29, 2021 Author Posted September 29, 2021 i did this mod as seen in the picture (before you confirmed that) and works great, thanks! On 9/28/2021 at 8:13 AM, ludek said: yes, I remember AsRock had AGP, vDimmRef, vNB, vDimm on one quad-op amp like LM324, impressive i asked because i only can go up to ~292Mhz BusSpeed, maybe need to try another CPU, because other guys get 320+ without vNB Mod. On 9/28/2021 at 8:13 AM, ludek said: in this configuration almost always the power componnent is a power SMD transistor. If your mems eat 1Amp or 2 amps, and vMem is 2.5, from 3V3 rail, the power dissapated on transistor is 3.3-2.5 over 1 Amp is (I=U/R so 3.3-2.5 is 0.8 x 1 is 0.8 W, it's not huge, but if you run full load, the current might jump to few amps) So 1A from 3.3V is 0.8W and from 5V is 5-2.5 x 1 = 2.5 W. This makes huge power losses and hot transistor. mhh i think i dont get it fully. the differenz from input and Output voltage multiply by Output current = power dissapated of transistor? then for example, a RAM about 2,5V and 1A @Stock: vDimm 2,5V/1A=2,5Ohm @5V Rail = (5V-2,5V)*1A = 2,5W dissipation as you wrote but if i need 3.5V for same RAM Typ vDimm 3,5V/2,5Ohm=1,4A @5V Rail = (5V-3,5V)*1,4A = 2,1W dissipation? with higher Voltage and higher Current less dissipation, i think there is a mistake in calculation. Quote
ludek Posted September 29, 2021 Posted September 29, 2021 P=u*u/r or p=i*i*r, my mistake! So its even worse because voltage diff is squared. Ehh electronic engineer doesnt remember ohms law... ? Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.