|
Post by somerandomtechyboi on Oct 23, 2021 23:53:18 GMT -5
inspect the pins on the ram slot before you do that. Use a magnifier. sometimes a pin will get pushed in and doesn't make contact. if that's the case, sometimes you can use a needle or fine dental pick to pull it out. Vin Ill keep that in mind Now the problem is a damaged nb, not really unfixable if i buy some basic tools that will allow me to swap out nb chips, i do have a dead p5q pro and the nb seems ok so ill salvage an nb chip off that mb Currently im looking at a cheap heatgun, some solder paste in a syringe (no stencil so i need a very small needle tip, maybe 0.3mm), some flux, tweezers, and some aluminum foil. i dont think ill need to buy solder wick cause i have some cables lying around to coat in flux I can prob convice mom to buy since i have ~5 broken mbs that can be fixed by swapping chips around, currently im willing to fix obv my ud3p - g31m es2c - p5kpl am se (2) Ud3p has damaged nb G31m es2c has a shorted sb (i think internally though ill try reballing before outright swapping it out) One of the p5kpl am se has a damaged sb and that an error on my part cause i was trying to take off the heatsink, stupid tim was hard as a rock Another has a broken socket (snapped pins) Another board i can fix is my p5qpl am but im not willing to do that cause trash fsb overclocking, maybe ill fix it up one day and put it as a display mb since it is my first mb that i ever bought but not now
|
|
|
Post by somerandomtechyboi on Nov 11, 2021 22:21:01 GMT -5
Been awhile and havent gotten the equipment to fix my ud3p just yet, though i was searching around for some ddr2 1066 cl5 rams so i can have a ram kit that ocs higher than my team rams so i can reach ludicrous fsbs once the ud3p is fixed but when i searched for transcend rams i was looking for axeram but ive only found 800mhz ones, though i think i may have found a hidden gem, some low profile transcend ddr2 800mhz sticks (567971-2757) but the dies are elpida dies (E1108AF8G), ive heard that elpida dies clock pretty well so what can i expect from these dimms?
I atleast want them to beat my zepplin rams (1086mhz 2.4v)
|
|
|
Post by Mr.Scott on Nov 12, 2021 18:33:22 GMT -5
The Axeram 1200's are good one's to try to find. Very tough these days. As far as DDR2 goes, you're wasting your time on anything but Micron D9 IC's. They rule DDR2. Crucial Ballistix is a safe bet. 
|
|
|
Post by somerandomtechyboi on Nov 13, 2021 6:54:46 GMT -5
The Axeram 1200's are good one's to try to find. Very tough these days. As far as DDR2 goes, you're wasting your time on anything but Micron D9 IC's. They rule DDR2. Crucial Ballistix is a safe bet.  Those are pretty much unavailable, 1066 is already a pain in the ass to find, 1200 is another level of hard to find I can find micron d9 in corsair xms rams but the cheap ones are 800mhz, will they stack well against my team extreem dark ddr2 rams? Cause my team extreem dark do 1250 max, 1270 if i loosen the timings to 7-15-15 Also how do those elpida rams oc anyways?
|
|
|
Post by Mr.Scott on Nov 13, 2021 9:37:18 GMT -5
Test DDR2 at cl4 or 5. Good DDR2 800 should do 1000 or better @ 2.3v. Good 1066 should do at least 1200. The Elpida will probably not do that. Test your Team Extreme at the tight CL also.
|
|
|
Post by somerandomtechyboi on Nov 14, 2021 9:11:55 GMT -5
Test DDR2 at cl4 or 5. Good DDR2 800 should do 1000 or better @ 2.3v. Good 1066 should do at least 1200. The Elpida will probably not do that. Test your Team Extreme at the tight CL also. My team rams cap at around 1080 at 5-5-5 but at 6-6-6 cap out at 1250, havent tested 4-4-4 yet but itll prob die. Stable max is ~1170 but i run 1150 6-6-6-12 2.1v, seems like ram volt doesnt help much freq wise but seems to help timings Maybe for testbenching ill buy some 5-5-5 1066 rams instead of my current 6-6-6 1066 extreem dark, or maybe ill just be a f ing cheapskate and tec cool the ram xD Well only 5-5-5 1066 kit ive found at a decent price is some gskill 1066 and dominator 1066 both w 5-5-5-15 timings so they should overclock pretty well Btw how do those corsair xms sticks overclock? Cause they are 800mhz 4-4-4-12 and they are micron d9 so are they better than my team extreem rams or same ocability?
|
|
|
Post by Mr.Scott on Nov 14, 2021 10:30:26 GMT -5
Depends on how you want to compare. Overclockability, or performance.
4-4-4-12 @ 1000 will perform better than 6-6-6-x @ 1250
Test with SuperPi.
|
|
|
Post by somerandomtechyboi on Nov 15, 2021 3:17:33 GMT -5
Depends on how you want to compare. Overclockability, or performance. 4-4-4-12 @ 1000 will perform better than 6-6-6-x @ 1250 Test with SuperPi. Id rather have the overclockability cause if im gonna use a board like the ud3p then how the hell am i gonna reach 700fsb+ with rams that dont oc well? Though for a garb board that doesnt do high fsb then yea ocability basically doesnt matter at that point and best go for tight timings Btw ive been thinking about frankensteining some boards once i get the tools i need to repair my ud3p, would it be possible to put a p45 nb on a p41 mobo? Cause most of the pcb designs from p41 to p45 seem to be quite similar (looking at the giga boards) so maybe if i can put a p45 nb on a p41 mobo i could get a ddr3 p45 board for cheap and not pay the ludicrous 60$ at the lowest for one of those mbs, heck i might be able to go for those boards with only 2 slots of ddr3 so closer to the mb = more ocability so garbage board to a board that goes vroom vroom xD Not to mention the other benifits like being able to run 8 or 16gb ddr3, sure 4gb ddr2 exists but they are garb at overclocking Though the roadblock that ive found is that p45 needs ich10 but p41 comes with ich7, would it be possible to just drop in an ich10 in place of an ich7? Cause again with the board layout similarities across diff chipsets, another one would be obtaining a bios chip thats compatible with a p45 bios but my ep45 has a dual bios so i can just yoink one of the chips anyways Guess ill take off parts like the nb and sb off my fully dead mobos and compare the high end stuff like p45 and ich10 to the lower end stuff like p41 and ich7 and see if they are compatible, or ill just figure out a way to run p45 with an ich7 xD If i can find em ill prob prefer fankensteining the 33 or 43 chipsets since they run ich9 or 10 anyways so no hassle with swapping sb, but giga p43 are not available, i have found a deal on one in a non tested computer but it seems to be a ddr2 board and not the more desirable ddr3 board, maybe ill get it if my p45 goes bust and doesnt work Though ill most likely upgrade g/p41 boards to g/p31 cause 41 chipsets are crippled and have no fsb capability whatsoever unlike the 31 chipsets which whoop their asses and run as high as they can
|
|
|
Post by Macsbeach98 on Nov 20, 2021 23:13:02 GMT -5
Hi and welcome to the forum I have been reading your posts since you showed up unfortunately real life has taken over my hobby for now and I have not been real active. If you think you are going to change NB and SB chips successfully with a hot air tool and you are not going to use a stencil and proper sized solder balls what you are going to put dots of solder on the lands with a soldering iron. You are going to fail every time you try. Take it from me I have tried all that. I use a ACHI IR6500 rework station now the procedure goes like this. Put the chip into the jig with the stencil and fill it with balls heat it with the hot air tool but not applying to much heat or it will crystalise the solder then after cleaning up the lands on the board position the chip on the board accurately when the rework station starts it heats the board from the bottom to around 160c takes about 5 mins before the top heater comes on then the top heater takes the temp up in 10 degree intervals to 220c if using leaded balls. Heat has to be applied evenly and accurately and the board has to be hot too you are not going to do that with a hot air tool even with the rework station you are not going to get 100% success very far from it. Reworking chips is a lot harder than the videos you have seen on youtube. Putting ICH10 on in place of ICH7 you will need the datasheets to check all the lands are signal compatible you will also need a modified bios to suit the chip. I doubt they are going to be compatible ICH9 to ICH10 might be compatible but ICH9 works pretty good as is. My advice would be to submit your ref frequency with the G31 it will be a record also run all the other benches you can for it and see how you go. There is great satisfaction from fixing boards but there is also a lot of satisfaction running the benches and being able to score highly against other same boards. P45 X48 is where its at forget the other lesser chipsets there was a couple of P35 boards that ran good but not many. You seem enthusiastic and your young you could go well at this the more you play with the more experience you will get.
|
|
|
Post by somerandomtechyboi on Nov 25, 2021 4:51:47 GMT -5
Hi and welcome to the forum I have been reading your posts since you showed up unfortunately real life has taken over my hobby for now and I have not been real active. If you think you are going to change NB and SB chips successfully with a hot air tool and you are not going to use a stencil and proper sized solder balls what you are going to put dots of solder on the lands with a soldering iron. You are going to fail every time you try. Take it from me I have tried all that. I use a ACHI IR6500 rework station now the procedure goes like this. Put the chip into the jig with the stencil and fill it with balls heat it with the hot air tool but not applying to much heat or it will crystalise the solder then after cleaning up the lands on the board position the chip on the board accurately when the rework station starts it heats the board from the bottom to around 160c takes about 5 mins before the top heater comes on then the top heater takes the temp up in 10 degree intervals to 220c if using leaded balls. Heat has to be applied evenly and accurately and the board has to be hot too you are not going to do that with a hot air tool even with the rework station you are not going to get 100% success very far from it. Reworking chips is a lot harder than the videos you have seen on youtube. Putting ICH10 on in place of ICH7 you will need the datasheets to check all the lands are signal compatible you will also need a modified bios to suit the chip. I doubt they are going to be compatible ICH9 to ICH10 might be compatible but ICH9 works pretty good as is. My advice would be to submit your ref frequency with the G31 it will be a record also run all the other benches you can for it and see how you go. There is great satisfaction from fixing boards but there is also a lot of satisfaction running the benches and being able to score highly against other same boards. P45 X48 is where its at forget the other lesser chipsets there was a couple of P35 boards that ran good but not many. You seem enthusiastic and your young you could go well at this the more you play with the more experience you will get. Yea for being some idiot 14 year old that wants to start a small computer buisness to support my family abit, has a tiny budget, and has a hobby for f ing around with xoc, hardware modding, etc. Ive gone pretty deep into the rabbit hole for ocing and hardware modding, only reasons i got this far is budget constraints and sheer determination xD I cant register a valid.x86 account for some reason cause it doesnt send me the activation emal though i think i have the validation file somewhere, so thats why i havent submitted it to hwbot For swapping bga chips wise ive decided to go with a 1200w heatgun, universal stencil (multiple of these with diff sizes), 138c solder paste as a substitute to solder balls cause they seem to work looking at yt vids, some solder wick, flux, and kepton tape. I have aluminum foil, a solder iron laying, and some solder laying around, anything else missing? Also ive been looking at ways to make my p5q better, seems like ill want to flash a p5qd turbo bios on it instead of kets modded bios (p5q pro turbo), while looking at some hwbot freq records ive stumbled upon someone that has modded some caps onto the nb power delivery hwbot.org/submission/2173328_smetankin_reference_frequency_p5q_pro_614.9_mhzSeems like they put a 3300uf capacitor on the lower solid capacitor marked 863 and some smalled i cant really tell but looks to be 4.3 or 43uf capacitor on the solid capacitor marked 87P From my understanding these cap mods are done to stabilize voltage or something, idk much about electronics but what would the effect of an extra/larger cap be in terms of voltages/power delivery? Would it make power delivery more stable aka less vdroop? Or something else? Idk what the hell the markings on the solid capacitor 863 or 87P mean so im guessing its just capacitance or whatever the hell its called, theres prob an equation for sht like this right? Or do you just yolo it and shove whatever capacitor as a cap mod? My source of capacitors are from dead psus Tbh this p5q has been quite the sad excuse for a p45 mobo, modded the bios and it still only does 550fsb max, if a g31 competes with a p45 you know that p45 board is a sad excuse For the cap mod ill prob just use my full dead giga p41 as a guinea pig so if i f up and the cap explodes or something ill atleast not ruin the nb of my dead p5q pro, or i just yolo it
|
|
|
Post by Macsbeach98 on Nov 27, 2021 3:44:59 GMT -5
Yea for being some idiot 14 year old that wants to start a small computer buisness to support my family abit, has a tiny budget, and has a hobby for f ing around with xoc, hardware modding, etc. Ive gone pretty deep into the rabbit hole for ocing and hardware modding, only reasons i got this far is budget constraints and sheer determination xD I cant register a valid.x86 account for some reason cause it doesnt send me the activation emal though i think i have the validation file somewhere, so thats why i havent submitted it to hwbot You dont have to register when you submit to CPU-Z validator when you put your file in and it validates it save the address thats in the address bar that is what you use when you submit to HWBot.
For swapping bga chips wise ive decided to go with a 1200w heatgun, universal stencil (multiple of these with diff sizes), 138c solder paste as a substitute to solder balls cause they seem to work looking at yt vids, some solder wick, flux, and kepton tape. I have aluminum foil, a solder iron laying, and some solder laying around, anything else missing? See how you go I dont have much confidence you will be successful but experience is the best teacher any questions you have just ask.Also ive been looking at ways to make my p5q better, seems like ill want to flash a p5qd turbo bios on it instead of kets modded bios (p5q pro turbo), while looking at some hwbot freq records ive stumbled upon someone that has modded some caps onto the nb power delivery hwbot.org/submission/2173328_smetankin_reference_frequency_p5q_pro_614.9_mhzSeems like they put a 3300uf capacitor on the lower solid capacitor marked 863 and some smalled i cant really tell but looks to be 4.3 or 43uf capacitor on the solid capacitor marked 87P Ok from the looks of the choke on it it has a single phase buck controlled regulator there would be a small chip controlling the transistors its that chip that you would use for NB voltmods. The job of the regulator is maintain a constant smooth voltage under varying loads. The output of the regulator without any capacitors would be a sawtooth waveform as it pulses its output on and off the job of the capacitor is to smooth the sawtooth to be a smooth DC voltage. Adding more capacitance can sometimes help but if there is enough capacitance it wont do anything. One thing I would not have put a 3300 6.3v electro on there a solid cap of the same value as the original would give double the capacitance would of been a much better choice the reason being solid caps have a low ESR (effective series resistance) solids are a lot faster reacting than electros especially big electros.From my understanding these cap mods are done to stabilize voltage or something, idk much about electronics but what would the effect of an extra/larger cap be in terms of voltages/power delivery? Would it make power delivery more stable aka less vdroop? Or something else? VDroop is an intel and AMD specification and is supposed to be there to stop voltage spikes over what they think is unsafe some boards have a lot more than they are supposed to though extra caps wont stop vdroop of course for extreme overclocking we dont want droop at all. Idk what the hell the markings on the solid capacitor 863 or 87P mean so im guessing its just capacitance or whatever the hell its called, theres prob an equation for sht like this right? Or do you just yolo it and shove whatever capacitor as a cap mod? My source of capacitors are from dead psus 863 the first 2 digits are the amount the third digit is the amount of zeros after the amount so its 86000nF or 86uF 87P the P just indicates uF so its 87uF.Tbh this p5q has been quite the sad excuse for a p45 mobo, modded the bios and it still only does 550fsb max, if a g31 competes with a p45 you know that p45 board is a sad excuse. Well I cant say weather the board is sad or not it could be. We had a friendly ref freq comp here a couple of yrs ago I was using a EP45T-UD3R/E8500 and 560mhz was pulling me up one bios setting done it for me NB delay and I was straight up to 642mhz.warp9-systems.proboards.com/thread/677/warp9-hold-line-bench-offFor the cap mod ill prob just use my full dead giga p41 as a guinea pig so if i f up and the cap explodes or something ill atleast not ruin the nb of my dead p5q pro, or i just yolo it Pull all the solid caps off it to use in other projects
|
|
|
Post by somerandomtechyboi on Nov 27, 2021 8:46:12 GMT -5
Yea for being some idiot 14 year old that wants to start a small computer buisness to support my family abit, has a tiny budget, and has a hobby for f ing around with xoc, hardware modding, etc. Ive gone pretty deep into the rabbit hole for ocing and hardware modding, only reasons i got this far is budget constraints and sheer determination xD I cant register a valid.x86 account for some reason cause it doesnt send me the activation emal though i think i have the validation file somewhere, so thats why i havent submitted it to hwbot You dont have to register when you submit to CPU-Z validator when you put your file in and it validates it save the address thats in the address bar that is what you use when you submit to HWBot.
For swapping bga chips wise ive decided to go with a 1200w heatgun, universal stencil (multiple of these with diff sizes), 138c solder paste as a substitute to solder balls cause they seem to work looking at yt vids, some solder wick, flux, and kepton tape. I have aluminum foil, a solder iron laying, and some solder laying around, anything else missing? See how you go I dont have much confidence you will be successful but experience is the best teacher any questions you have just ask.Also ive been looking at ways to make my p5q better, seems like ill want to flash a p5qd turbo bios on it instead of kets modded bios (p5q pro turbo), while looking at some hwbot freq records ive stumbled upon someone that has modded some caps onto the nb power delivery hwbot.org/submission/2173328_smetankin_reference_frequency_p5q_pro_614.9_mhzSeems like they put a 3300uf capacitor on the lower solid capacitor marked 863 and some smalled i cant really tell but looks to be 4.3 or 43uf capacitor on the solid capacitor marked 87P Ok from the looks of the choke on it it has a single phase buck controlled regulator there would be a small chip controlling the transistors its that chip that you would use for NB voltmods. The job of the regulator is maintain a constant smooth voltage under varying loads. The output of the regulator without any capacitors would be a sawtooth waveform as it pulses its output on and off the job of the capacitor is to smooth the sawtooth to be a smooth DC voltage. Adding more capacitance can sometimes help but if there is enough capacitance it wont do anything. One thing I would not have put a 3300 6.3v electro on there a solid cap of the same value as the original would give double the capacitance would of been a much better choice the reason being solid caps have a low ESR (effective series resistance) solids are a lot faster reacting than electros especially big electros.From my understanding these cap mods are done to stabilize voltage or something, idk much about electronics but what would the effect of an extra/larger cap be in terms of voltages/power delivery? Would it make power delivery more stable aka less vdroop? Or something else? VDroop is an intel and AMD specification and is supposed to be there to stop voltage spikes over what they think is unsafe some boards have a lot more than they are supposed to though extra caps wont stop vdroop of course for extreme overclocking we dont want droop at all. Idk what the hell the markings on the solid capacitor 863 or 87P mean so im guessing its just capacitance or whatever the hell its called, theres prob an equation for sht like this right? Or do you just yolo it and shove whatever capacitor as a cap mod? My source of capacitors are from dead psus 863 the first 2 digits are the amount the third digit is the amount of zeros after the amount so its 86000nF or 86uF 87P the P just indicates uF so its 87uF.Tbh this p5q has been quite the sad excuse for a p45 mobo, modded the bios and it still only does 550fsb max, if a g31 competes with a p45 you know that p45 board is a sad excuse. Well I cant say weather the board is sad or not it could be. We had a friendly ref freq comp here a couple of yrs ago I was using a EP45T-UD3R/E8500 and 560mhz was pulling me up one bios setting done it for me NB delay and I was straight up to 642mhz.warp9-systems.proboards.com/thread/677/warp9-hold-line-bench-offFor the cap mod ill prob just use my full dead giga p41 as a guinea pig so if i f up and the cap explodes or something ill atleast not ruin the nb of my dead p5q pro, or i just yolo it Pull all the solid caps off it to use in other projectsIve swapped bios to p5qd and fsb max initially was 570, modded that bios by swapping ebb with one from an MIIF and doesnt seem to have helped much, highest i have now is 585fsb what nb and cpu delay setting did you use anyways? Cause giga p45 have very similar layouts to these p5q boards, heck bios will prob work if i modded it cause it stops just before it posts (code 26, post iirc is code 52 right after 26), i use a 1:2 ratio either nb 200 cpu 400 or nb 400 cpu 800 which seems to work fine I dont really see a point in nb voltmodding other than adjusting nb on the fly but again these boards are trash for benching cause after 570 ill have to reset and boot to the next setting instead of just applying it so yea board hates high fsbs Welp thanks for the crap ton of info but i gotta finish a bunch of school assignments atm so maybe after monday or something ill be able to start f ing around with hardware modding
|
|
|
Post by Macsbeach98 on Nov 29, 2021 4:19:28 GMT -5
NB delay 7us that was about it.
|
|
|
Post by somerandomtechyboi on Dec 1, 2021 7:13:02 GMT -5
NB delay 7us that was about it. Welp my board only does 1.2ns max cause for the nb and cpu delay its measured in ps So far ive found that 800nb and 1000cpu seems to be good on high oc but the board is just not good for high oc tbh, very underwhelming and quite pathetic, well hope that ud3p gets fixed soon cause i dont wanna deal with this pos anymore
|
|
|
Post by somerandomtechyboi on Dec 20, 2021 10:50:35 GMT -5
Been awhile and now i wanna look at voltmodding my g31m s2c, i was comparing my "dead" p5kpl am with a "dead" g31m es2c (same thing but 2 bios chips) and looking at the nb + sb power delivery this is what ive found
The p5kpl am has its nb + sb vrms close to the nb and it does have a volt controller on it (rt8105), the sb seems to be fed by a mosfet and cap and im not sure if the as324m-e1 is a volt controller or not
The giga g31 has a completely bare nb that looks to be powered by a mosfet and cap with no volt controller and the nb + sb vrms are close to the sb and it does have a volt controller on the sb (rt8105), sb volt control is something ill prob need for higher than 542fsb since it just gives a pagefile error above 542, even 542.5 still yeilds that error, nb volt controll is prob also something i need to get even higher fsbs but mainly in max stable fsb cause above 880mhz ram and it just becomes unstable even on sticks like extreem dark 1066
Now how do i go about raising the nb voltage? It doesnt appear to have a volt controller so im guessing that ill have to use 3.3v rail, divide it with a variable resistor, and hook it up somewhere, idk what values i should use for voltage dividing, and im guessing that ill have to hook it up to the mosfet after desoldering it, do i need to make a proper e power for this instead of just directly hooking it up to a voltage rail? Cause nb does consume abit of amperage
|
|
|
Post by Macsbeach98 on Dec 24, 2021 1:54:47 GMT -5
Ok we will start with the RT8105 it is easy here is the pinout.  Pin 6 is the feedback pin measure the resistance to ground then multiply the value by 20 that is the value of the trimmer you will need set to max resistance and soldered onto pin 6 and the other side to ground lower the resistance of the trimmer to raise voltage while measuring the output of the mosfets with a meter. AS324M-E1 Is a Quad operation amplifier basically its a linear regulator that can control upto 4 circuits whereas the RT8105 is a PWM regulator it is adjustable too seeing as it has 4 circuits you are only worried about 1 circuit. Below is a pic of the pinout and basic circuit for one circuit. in the pic below on the right you want to know what pin goes to the Gate of the mosfet use your multimeter to work out which pin goes to the mosfet using the left of the diagram work out which pin it is and it will tell you which pin is the negative input thats the pin you want to attach your trimmer to being a linear regulator I would probably start at 50 times over the value as adjustments to linear regs tend to be more vicious than PWM controllers. The download links for both controllers are at at the bottom of this post. If you have any questions ask. pdf1.alldatasheet.com/datasheet-pdf/download/176222/BCDSEMI/AS324M-E1.htmldatasheetspdf.com/datasheet/RT8105.html
|
|
|
Post by somerandomtechyboi on Dec 25, 2021 8:59:36 GMT -5
Ok we will start with the RT8105 it is easy here is the pinout.  Pin 6 is the feedback pin measure the resistance to ground then multiply the value by 20 that is the value of the trimmer you will need set to max resistance and soldered onto pin 6 and the other side to ground lower the resistance of the trimmer to raise voltage while measuring the output of the mosfets with a meter. AS324M-E1 Is a Quad operation amplifier basically its a linear regulator that can control upto 4 circuits whereas the RT8105 is a PWM regulator it is adjustable too seeing as it has 4 circuits you are only worried about 1 circuit. Below is a pic of the pinout and basic circuit for one circuit. in the pic below on the right you want to know what pin goes to the Gate of the mosfet use your multimeter to work out which pin goes to the mosfet using the left of the diagram work out which pin it is and it will tell you which pin is the negative input thats the pin you want to attach your trimmer to being a linear regulator I would probably start at 50 times over the value as adjustments to linear regs tend to be more vicious than PWM controllers. The download links for both controllers are at at the bottom of this post. If you have any questions ask. pdf1.alldatasheet.com/datasheet-pdf/download/176222/BCDSEMI/AS324M-E1.htmldatasheetspdf.com/datasheet/RT8105.htmlHow do i measure resistance to ground though? And what about the nb side of my giga g31? Cause it doesnt seem to have any volt controllers, literally just a mosfet and a cap
|
|
|
Post by Macsbeach98 on Dec 27, 2021 2:04:46 GMT -5
With a multimeter from the feedback pin to ground. Your G31 has to have a NB voltage controller the gate pin probably disappears into the board but comes up somewhere without one a mosfet cant regulate the voltage according to load.
|
|
|
Post by somerandomtechyboi on Dec 27, 2021 23:52:41 GMT -5
With a multimeter from the feedback pin to ground. Your G31 has to have a NB voltage controller the gate pin probably disappears into the board but comes up somewhere without one a mosfet cant regulate the voltage according to load. Does that mean if i raise sb volt im also raising nb volt?
|
|
|
Post by Macsbeach98 on Dec 28, 2021 0:22:29 GMT -5
If they are fed from the same controller it does.
|
|
|
Post by somerandomtechyboi on Dec 29, 2021 8:06:44 GMT -5
If they are fed from the same controller it does. Well crap, that prob means they are running the same voltage, could too high sb volt make oc unstable?
|
|
|
Post by Macsbeach98 on Dec 30, 2021 2:58:41 GMT -5
If it doesnt like high voltage yes it will be unstable
|
|
|
Post by somerandomtechyboi on Dec 30, 2021 10:51:37 GMT -5
If it doesnt like high voltage yes it will be unstable Well sht Any way i can decouple sb and nb like hooking up an epower to the nb or something? Or maybe i could reduce sb input voltage by swapping a resistor or something
|
|
|
Post by Macsbeach98 on Dec 31, 2021 3:25:18 GMT -5
Has the board got a separate voltage adjustment for both the NB and SB?
|
|
|
Post by somerandomtechyboi on Jan 2, 2022 19:53:42 GMT -5
Has the board got a separate voltage adjustment for both the NB and SB? Nope but looking at its p31 brother (p31 ds3l) that seems to also have a bare nb but theres a vrm under the 1st pcie slot which i assume is prob for pcie power delivery and on the lowest part of the cpu vrm there seems to be an extra volt controller, maybe its connected to the nb On the g31 its literally just the sb vrm and thats it
|
|
|
Post by somerandomtechyboi on Jan 22, 2022 9:14:37 GMT -5
Ok ive been searching for some micron d9 (d9gmh or d9gkx) and ive found some corsair and gskill kits. Gskill is most likely d9 since theres a 1066 5-5-5-15 2x1 kit runs at 2.1v and a lone 800 4-4-4-12 kit that runs 2v, Im particularly interested in the corsair ones cause of price, though most of them are xms2 5-5-5-18 rams that run at 1.8 or 1.9v, are these micron d9?
I also wanna know if they are d9gmh or d9gkx since i may be able to fix my ud3p and run ludicrous fsb, if not then i guess i could run 1200+ if its d9gkx, for all i know d9gmh prefers to have lower timings but seems to not clock as high so im unsure of d9gmh max ram speed capabilites
The 2x1 gskill 1066 5-5-5-15 kit is F2-8500CL5D-2GBPK The lone gskill 800 4-4-4-12 kit is F2-6400CL4D-2GBPK All the corsair kits are CM2X1024-6400 but i have found ver 2.8, 4.8, and 7.5
|
|
|
Post by Mr.Scott on Jan 22, 2022 10:36:53 GMT -5
None of your Corsair kits are D9. Your 800/6400 G.Skill kit is not D9. Your G.Skill 1066/8500 kit has a high probability of being D9's, but it not a given.
|
|
|
Post by Bones on Jan 22, 2022 14:43:44 GMT -5
You'll also find there are many variants of D9's out there to find, D9GMH or D9GKX are just two examples of it. Personally I have a few sets. I know one is a set of D9JN's and the other is a set of D9JMN's, both being good sets of sticks and are an OCZ make and model of RAM. I'll also call ground in to give his perspective on the subject, he knows alot about it and his advice is well worth seeking.
|
|
|
Post by osmiumoc on Jan 22, 2022 21:53:08 GMT -5
Hi, just found your thread here. I'm the guy who posted the HP part numbers for Micron D9s in another thread. I had a lot of luck and found an electronics recycler who sold me 18 sticks of D9GKX for cheap. First of all, maybe you found it already but this list was very useful for me: web.archive.org/web/20111228163948/http://ramlist.i4memory.com/ddr2/Here is my post about OEM part numbers: warp9-systems.proboards.com/post/40146/threadFinding good DDR2 RAM today can be a test of patience, but from my experience sooner or later something will show up for cheap. Just keep looking. For high FSB numbers without the trouble of DDR2 limits, a DDR3 socket 775 board might be nice. I decided against that, because for some reason I love overclocking with DDR2. Socket 775 + DDR2 is my favorite platform. As for your board modifications, I can not give you any advice on those. I have not yet done any serious volt mods or epowers. I started out on 775 with a P5Q-E a while back, all I can say is that the P45 chipset is almost never the limiting factor for FSB clocks. You don't need to give P45 very high voltages. The limits I encountered were usually related to my DDR2 sticks not clocking very high, CPU hitting its limit (VTT wall at some FSB) or lacking bios options for fine tuning GTLs and/or clock skews. Almost all my 775 boards are P45, so I can only give some feedback on this chipset. I only have one X48 and one nForce 780i left. The nForce is pretty bad for high OC.
|
|
|
Post by Macsbeach98 on Jan 22, 2022 23:07:00 GMT -5
I just done a search on fleabey and there is a heap of ballistix on there. Do a worldwide search for "Crucial Ballistix DDR2" You want 16FD3 or 16FD5 on the small rectangle sticker like in the photo below. Some of them are cheap too I have got 4 sets of FD3 that will bench at least 660 5-5-5-15 2.35v Dual Channel 
|
|