If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#1
|
|||
|
|||
vcore mod a7n8x-x
Paul wrote:
The datasheet for the ST Micro L6917BD Voltage Regulator is he http://www.st.com/stonline/books/pdf/docs/8162.pdf The mod is connecting to pin 9 of the chip, which is the FB terminal. Pin 9 is on the side with the indented "dot" in the top of the package - the dot marks pin 1. By adding or subtracting current from the summing junction FB, it is possible to change the output voltage. (Note: This method isn't available on all voltage regulators - it just happens that the connectivity of this chip exposes the necessary resource for you to play with.) By connecting the variable resistor from pin 9 to ground, current is pulled from the summing junction, and to replace it, the chip increases its output voltage. The output voltage is tied to FB via a resistor as well. So, the mod pulls on the junction and the output voltage rises in response (this is closed loop feedback). Think I understand most of this, it hits ground and therefore the chip increases output voltage, as there is some 'leaking' the chip 'thinks' it needs to give more voltage to keep it stable,.. (Doesn't matter if I'm wrong on this one,.. the result is what it's all about right?) In this case, the maximumoc web page states to take a 25000 ohm variable resistor and connect it from pin 9 to ground. My advice to you, is to use two resistors - a fixed one and a variable one - connected in series. The reason for this, is it is too tempting to keep turning the knob on the variable resistor. If you do that without a protection resistor, the voltage could be raised too high and kaboom! | L6917BD | | dot | |*_1_2_3_4_5_6_7_8_9_______| | Fixed Resistor | + | Variable (use either one of the end terminals Resistor and the center one) | Ground By looking at page 30 of the ST micro datasheet, you can see that the resistors used in the feedback network- the voltage divider network and the shaping network, are using fairly small resistance values. This suggests to me that the 25K variable pot is too big - what will happen is, the "interesting" area of pot rotation will be so small, it will be easy to overshoot the mark when dialing it. The voltage change you desire will be over a small portion of the adjustment range. To figure out what the value should be, I would start with a selection of fixed resistors. Say 20K, 10K, 5K, and so on. First connect the 20K resistor from pin 9 to ground. Using a voltmeter, measure Vcore when the resistor is connected and then disconnected (in the same session). As I have no experience whatsoever with electronics, where should I stick this voltmeter to measure the vcore. Is there another way like bios or mbm5 to see the difference? Also I cannot hold the resistor in my hand,.. what should I use to connect and disconect the resistor, I'm pretty bad eating chinese with wooden sticks :-D Do I use a rubber thingy or is there something else to keep it at it's place and (re)move it again? With the 20K resistor, the change in voltage should be quite small. Next try the 10K resistor, and see how much the voltage changes. (Note: Keep your fingers off the contacts when connecting the resistor, because your body resistance will screw up the experiment.) What should happen, is the increase in voltage seen with the 10K resistor should be twice the amount seen with the 20K resistor. Based on this, you should be able to extrapolate how much resistance will be needed to reach your desired voltage boost. That value of resistance will become the "Fixed Resistor" in the picture I drew above. Say for example: 1) No resistor from pin 9 to ground 1.65V measured 2) 20K resistor from pin 9 to ground 1.75V measured (delta = 0.1V) 3) 10K resistor from pin 9 to ground 1.85V measured (delta = 0.2V) To get to a boost of 0.4 volts, then it is easy to see that a 5K ohm resistor would do that, so your fixed resistor becomes 5K ohms (and 5.1K ohms is the nearest value at the store). Got it,.. The variable resistor will be larger than thix fixed resistance. If the minimum boost is to be 25% of the maximum boost, then the variable resistor will be 3 times the value of the fixed resistor. By including the fixed resistor in series, you are guaranteeing that even if the variable pot is turned until it reads zero ohms, the fixed resistor limits the boost to 0.4V or whatever you designed it to be. Ok,.. So I can up my voltage with the variable resistor? Say, after testing the fixed resistors I've found out I need the 5k ohms as in your example. The 'Minimum voltage I can set vcore now has changed from 1.65v to 2.05v (1.65+delta=0.4v) correct? Now the variable resistor comes in and 'limits' (?) the vcore to a extra maximum of 0.4v? ehm,.. again,.. In bios my vcore reads 1.65v but will really be 2.05? 1.85v will give an output of 2.25v vcore? Ok, think I'm getting there,.. as the variable resistor is between ground and the fixed resistance I can 'lower' the extra vcore coming from the mod,.. Because as you said if I turn the variable off (zero ohm) I wil never get over 0.4v extra vcore, on the other hand is it true that when I turn it on full it won't give me the extra voltage and my 1.65v will give me the original output of 1.65v again? Is my interpretation correct here? Note that, once you touch the board with a soldering iron, the warranty is voided. What would you choose :-D Miss all the exitement and go for the dull waranty? Also note, that the other mod, the chipset voltage mod, is pretty dangerous as well. At least one of the first people to try this, borked his board (I think I read about some of that over on nforcershq). And, to the best of my knowledge, no one really knows what the Nvidia stated voltage spec is for this voltage. (So, the degree of danger to the chipset is unknown, unless someone knows what the nominal value is supposed to be). Not there yet,.. first this one,.. As I told you before I'm a complete NOOB on this stuff so lets get some things clear for me,.. (ehh,.. don't lauch) A resistor is a little iron thread with a dead dried worm in the middle, Right? Am I correct when I say, I need 3 fixed ones, 20K, 10K, 5K? What kind of variable resistance should i get? Where can I get these fixed and variable resistensis, and how will I be absoluely sure I have the right ones,.. Have fun, Paul Thx Paul -- Ugh! The notion that south is down is a European deceet. |
#2
|
|||
|
|||
In article K1kOb.212688$_x2.426327@zonnet-reader-1, "Moods"
wrote: Paul wrote: The datasheet for the ST Micro L6917BD Voltage Regulator is he http://www.st.com/stonline/books/pdf/docs/8162.pdf The mod is connecting to pin 9 of the chip, which is the FB terminal. Pin 9 is on the side with the indented "dot" in the top of the package - the dot marks pin 1. By adding or subtracting current from the summing junction FB, it is possible to change the output voltage. (Note: This method isn't available on all voltage regulators - it just happens that the connectivity of this chip exposes the necessary resource for you to play with.) By connecting the variable resistor from pin 9 to ground, current is pulled from the summing junction, and to replace it, the chip increases its output voltage. The output voltage is tied to FB via a resistor as well. So, the mod pulls on the junction and the output voltage rises in response (this is closed loop feedback). Think I understand most of this, it hits ground and therefore the chip increases output voltage, as there is some 'leaking' the chip 'thinks' it needs to give more voltage to keep it stable,.. (Doesn't matter if I'm wrong on this one,.. the result is what it's all about right?) More or less. This is a circuit with feedback, so one trend is used to counteract another. In this case, the maximumoc web page states to take a 25000 ohm variable resistor and connect it from pin 9 to ground. My advice to you, is to use two resistors - a fixed one and a variable one - connected in series. The reason for this, is it is too tempting to keep turning the knob on the variable resistor. If you do that without a protection resistor, the voltage could be raised too high and kaboom! | L6917BD | | dot | |*_1_2_3_4_5_6_7_8_9_______| | Fixed Resistor | + | Variable (use either one of the end terminals Resistor and the center one) | Ground By looking at page 30 of the ST micro datasheet, you can see that the resistors used in the feedback network- the voltage divider network and the shaping network, are using fairly small resistance values. This suggests to me that the 25K variable pot is too big - what will happen is, the "interesting" area of pot rotation will be so small, it will be easy to overshoot the mark when dialing it. The voltage change you desire will be over a small portion of the adjustment range. To figure out what the value should be, I would start with a selection of fixed resistors. Say 20K, 10K, 5K, and so on. First connect the 20K resistor from pin 9 to ground. Using a voltmeter, measure Vcore when the resistor is connected and then disconnected (in the same session). As I have no experience whatsoever with electronics, where should I stick this voltmeter to measure the vcore. Is there another way like bios or mbm5 to see the difference? Also I cannot hold the resistor in my hand,.. what should I use to connect and disconect the resistor, I'm pretty bad eating chinese with wooden sticks :-D Do I use a rubber thingy or is there something else to keep it at it's place and (re)move it again? Sure, if you don't have a voltmeter, use the BIOS Monitor or MBM5. The monitor chip on the motherboard is good enough for this purpose. In terms of how you make your connections, and what components you choose to connect, the possibilities are endless. I like the things I do to be vibration proof, and that is why I use a soldering iron. You will find some mods, where a pencil is used to make a conductive mark, or defroster repair (conductive) paint is used to make connections. This kind of approach is fine for modifying logic signals, but in this case, we are aiming for a specific resistance to ground, and if you use a "crappy" connection method, the Vcore voltage will vary all over the place. You would end up with an unstable system that way. Soldering a component is a metal to metal connection, and done properly, will last a long time. When I described my approach, I was trying to explain how the use of an additional resistor added safety to the method. You can put a 25K variable pot all by itself in the circuit, or you could use say, a 5.1K fixed resistor in line with a 20K variable pot. The difference is, if you accidently turn the 25K pot too far, it will go all the way down to 0 ohms, and the voltage you are trying to adjust will shoot up so high, the powered device will be fried. Now, many people will have done the mod, using nothing but the 25K variable pot - but my experience with these pots is, you are always tempted to turn them a little farther, and without a safety feature in your method (like the extra resistor in line), it is just too easy to ruin the board or the processor. With the 20K resistor, the change in voltage should be quite small. Next try the 10K resistor, and see how much the voltage changes. (Note: Keep your fingers off the contacts when connecting the resistor, because your body resistance will screw up the experiment.) What should happen, is the increase in voltage seen with the 10K resistor should be twice the amount seen with the 20K resistor. Based on this, you should be able to extrapolate how much resistance will be needed to reach your desired voltage boost. That value of resistance will become the "Fixed Resistor" in the picture I drew above. Say for example: 1) No resistor from pin 9 to ground 1.65V measured 2) 20K resistor from pin 9 to ground 1.75V measured (delta = 0.1V) 3) 10K resistor from pin 9 to ground 1.85V measured (delta = 0.2V) To get to a boost of 0.4 volts, then it is easy to see that a 5K ohm resistor would do that, so your fixed resistor becomes 5K ohms (and 5.1K ohms is the nearest value at the store). Got it,.. Again, I'm just illustrating a point. You can still use the single variable pot if you want - you would just have to be religious about how to set it. The best pots are the ones that allow you to visually determine how far they've been turned. The pots used as volume controls on stereo equipment, for example, are not necessarily linear, so that is why an ohmmeter is good to use, to calibrate position versus resistance. For example, say you knew that no less than 5K of resistance should be used - you could make a mark on the pot, to show how far it can be safely turned. There are also some trim pots that are 10 turn or 20 turn devices - since you cannot tell what the position of the wiper arm is at inside of these, they are very difficult to use without some kind of safety feature added to your circuit. The variable resistor will be larger than thix fixed resistance. If the minimum boost is to be 25% of the maximum boost, then the variable resistor will be 3 times the value of the fixed resistor. By including the fixed resistor in series, you are guaranteeing that even if the variable pot is turned until it reads zero ohms, the fixed resistor limits the boost to 0.4V or whatever you designed it to be. Ok,.. So I can up my voltage with the variable resistor? Say, after testing the fixed resistors I've found out I need the 5k ohms as in your example. The 'Minimum voltage I can set vcore now has changed from 1.65v to 2.05v (1.65+delta=0.4v) correct? Now the variable resistor comes in and 'limits' (?) the vcore to a extra maximum of 0.4v? ehm,.. again,.. In bios my vcore reads 1.65v but will really be 2.05? 1.85v will give an output of 2.25v vcore? Ok, think I'm getting there,.. as the variable resistor is between ground and the fixed resistance I can 'lower' the extra vcore coming from the mod,.. Because as you said if I turn the variable off (zero ohm) I wil never get over 0.4v extra vcore, on the other hand is it true that when I turn it on full it won't give me the extra voltage and my 1.65v will give me the original output of 1.65v again? Is my interpretation correct here? No, the purpose of the 5.1K (or whatever you determined by experimenting), is to set an _upper limit_. That is your safety feature. Say you download the datasheet for your processor from AMD, and it says "absolute max Vcore voltage" is 2.0 volts, then you want to work out what value of resistance will cause the Vcore circuit to make just slightly less than 2.0 volts. This separate resistor is in the circuit, in case you accidently or intentionally turn the variable resistor all the way to zero ohms. By adding the fixed resistor in series, the whole circuit will vary from say 5.1K ohms to 25.1K ohms, as the knob is turned all the way from one extreme to the other. When the knob is in the position where the total circuit resistance is being determined by the fixed 5.1K resistance, then you know the circuit cannot rise above your "safe" voltage. Again, in table form: No resistor (unmodified voltage) 1.65V measured 20K resistor from pin 9 to ground 1.75V measured (delta = 0.1V) 10K resistor from pin 9 to ground 1.85V measured (delta = 0.2V) 5.1K resistor from pin 9 to ground 2.05V measured (delta = 0.4V) 0 ohms from pin 9 to ground Kaboom - circuit shoots above max allowed So, we want to make sure we don't go below 5K ohms or whatever the projected value happens to be. By placing the resistor in series with the variable pot, we are making sure that no matter how the pot is dialled, there is _at least_ 5K ohms in the circuit. And, from our table, we then know the circuit cannot shoot above 2.05 volts. So, the purpose of my proposed experiments, was to determine just what that value of "safe" resistance might be. Even testing with just the 20K resistor will allow you to extrapolate what the safe value would be. My showing 10K and 5K in the table was to illustrate the linear relationship between the lowering of the resistance and the rise in voltage. Buying a multimeter doesn't have to be expensive. A big box hardware store near where I live, has multimeters (voltmeter/ohmmeter/ammeter all in one instrument) on sale after Christmas. Little ones are available for anywhere from $20 to $40. The tolerance on these meters is somewhere in the 1% to 3% range, so they are good enough for getting your circuit in the ballpark. Radio Shack also sells stuff like this, but the price might not be as reasonable. Using the ohmmeter on the multimeter, you can measure the resistance of your variable pot before soldering it into the circuit. You can even make a legend out of cardboard and affix it to the pot, to show the resistance value versus knob position. Note that, once you touch the board with a soldering iron, the warranty is voided. What would you choose :-D Miss all the exitement and go for the dull waranty? If you have spare computers like I do, then destroying a board isn't a problem. If this is your only computer, then don't do it. Restrict yourself to un-modded overclocking, until the day comes that you have enough equipment that destroying a board isn't the end of the world. Also note, that the other mod, the chipset voltage mod, is pretty dangerous as well. At least one of the first people to try this, borked his board (I think I read about some of that over on nforcershq). And, to the best of my knowledge, no one really knows what the Nvidia stated voltage spec is for this voltage. (So, the degree of danger to the chipset is unknown, unless someone knows what the nominal value is supposed to be). Not there yet,.. first this one,.. As I told you before I'm a complete NOOB on this stuff so lets get some things clear for me,.. (ehh,.. don't lauch) A resistor is a little iron thread with a dead dried worm in the middle, Right? Am I correct when I say, I need 3 fixed ones, 20K, 10K, 5K? What kind of variable resistance should i get? Where can I get these fixed and variable resistensis, and how will I be absoluely sure I have the right ones,.. This deal with the 20K, 10K, 5K resistors was to illustrate a point. That when you don't know all the details of a circuit, you can experiment and observe a cause and effect. Generally, I like to know a great deal about a circuit before I mod anything - my "batting average" is a lot better than when I started experimenting as a kid :-) Notice that in the case of the circuit that used a resistance between 1Meg ohm and 2Meg ohms, placing our 20K, 10K, 5K resistors in that circuit, would have killed it instantly. I picked the 20K value as a starting point, because the author of the maximumoc web page had already been successful by using a 25K ohm pot. For this method to be perfectly general, you would need a whole slew of resistors, stretching up to 10meg ohms or more. (You can buy something like this, in the form of a "resistor kit". A kit will contain a whole bunch of different resistor values in small quantities - I have a plastic storage tray with a couple thousand resistors in it, which came from kits like this.) Have fun, Paul Thx Paul For parts, there is Digikey.com (huge selection) and a place I used as a kid is Jameco.com (a bit smaller selection). Both places have downloadable catalogs in Acrobat PDF form. The last Digikey catalog I downloaded was 42MB (http://dkc3.digikey.com/pdf/T033/DigiKey.pdf) and the Jameco catalog is 17MB ftp://63.236.247.148/Current/Main/234catalog.pdf On PDF page 72 of the Jameco catalog, there are resistors (like 100 1/8 watt resistors for $0.99) or kits (2700 resistors for $29.95). If you know you will only need a few values, then getting individual bags might make sense, whereas if you want to plan for the future, maybe a kit makes more sense. In the circuits you will be working on, the power rating shouldn't be an issue - I like 1/8th watt resistors as they are pretty small and don't take up a lot of room. In circuits dissipating lots of power, you would need higher power parts. Page 73 has some variable pots. The ones with the metal adjustment shaft on them, are "volume control" type pots. The ones shown are "linear taper", so they won't have any strange ohms behavior as you turn them. The shaft makes these easy to turn, but they are rather large devices. You solder one lead to the center (wiper) pin and solder your second wire to either of the other two terminals. The resistance goes to zero, as the knob "position" approaches the terminal you've soldered the second wire to. The X201 is smaller and still provides visual feedback. In my work with those, they tend to turn too far, too easily, so again, a safety series resistor should be used with them. You will also notice there are some more expensive multi-turn pots on the page - as near as I can tell from the photos, you cannot tell much about their current setting - you have to count how many times they've been turned, and it is easy to lose your place. On some of the multi-turn pots, you cannot even tell when you've hit the end of the travel, as they have a clutch that lets the shaft go when you hit the end. Some of them click, while others are pretty silent while they slip. Soldering irons are on PDF page 146. I would get a Weller WP35 for $39.95 as a starter iron, as I believe the tip is grounded, and a grounded tip helps avoid ESD discharge or other disturbance to the circuits you work on (of course, the computer is unplugged from the wall while you are soldering!!!). Two pages over, there are some replacement tips for the soldering irons, and I like something smaller than the 1/8th inch chisel, as it is too large for this kind of work. There are other cheaper irons there too, if you want to buy a throw away iron. There is even a cheap soldering station, which gives you temperature control - that means the tip of the iron is thermostatically controlled, so just like the furnace in your home, it calls for heat only when needed. The stand alone soldering irons (including the WP35) freerun - there is a heating element that just keeps on heating them, no matter how hot the tip is getting. Rosin core solder is available on PDF page 149. A pound of solder would last a long time, so get one of the smaller "tubes" of solder for a couple of quick mods. For comments on the types, see: http://groups.google.com/groups?hl=e...nnrp1.deja.com You will also want to pick up a roll or two of "desolder braid". These are tiny woven copper braids, which, when pressed against a piece of solder and heated, cause the solder to be soaked up into the braid by capillary action. I like the thinnest I can find, like 0.025" or so. The thicker kinds are too hard to heat with the iron. Before using a piece, you flatten them out a bit more by pressing on them with a dull object - the flattening makes them soak up the solder a little better. For wire, don't buy your wire at Radio Shack, as all they sell now is unplated copper wire. Using nickel plated wire or better, makes soldering so much easier. On PDF page 108 there is some Kynar wire wrap wire, which is a thin plated wire - the thinness in this case is to ensure the wire flexes and doesn't break whatever you solder it to. As I indicated in a previous post, don't solder resistors or potentiometers directly to the board, as the mechanical stress can be enough to rip stuff off the motherboard - use a thin wire to connect to the motherboard. A short piece, stripped and soldered, gives stress relief. To strip the wire, you need a pair of wire strippers. (When I was a kid, I used to occasionally use my teeth for this :-) In a pinch, the model 78991CF on PDF page 154 can be used - these have a set screw that you adjust for each gauge of wire to be stripped. If you have only one roll of wire in your collection, this stripper would be fine. Note that there is a knack to stripping wire without nicking it - you will be very frustrated at first, because you'll need a lot of tries to get good stripped ends that won't break off later. Strippers are better than any knife or pliers for the stripping operation. I prefer the Ideal strippers (the ones just below the 78991CF strippers), but I notice they don't have model 45-125 listed - it strips 30 gauge wire, and you would need strippers that go down to 30 gauge, if you hope to strip the wire on the roll of wire wrap wire you buy. For big stuff, like the variable pot, you will need a way to mechanically fasten it to the motherboard. I have no advice to offer on this, because I would use a fixed resistor only for a job like this, and they can be held down with a Locktite product or in a pinch, a piece of electrical tape. When doing all this soldering, there will be a lot of soldered connections lying around. They can be covered with polyolefin heat shrink tubing. You slide a piece of this stuff over the wire before soldering. After the joint is made, slide the tubing over the joint, then apply heat to the tubing, to make it contract. You only have to contract it enough, to prevent it from sliding away from the area you need the insulation. The tubing is on PDF page 106 in the catalog and comes in one yard lengths. Buy several different diameters, as it is difficult to predict what size will be needed to cover your solder joints. It shrinks to around half its diameter when fully shrunk. (Note - don't shrink it by using the _tip_ of your soldering iron - this will foul and pollute the tip. Instead, use the _body_ of the iron below the tip. It'll stink, but better the body gets dirty than the tip. You need to keep the tip clean for good soldering. If the iron is hot enough, holding it underneath the tubing, without touching the tubing, might shrink it enough so it doesn't slip off.) So, here is what your "volt modders" kit will consist of: 1) Fixed resistors. For most work, 1/4 or 1/8W resistors are fine. For work like this, I'd probably use the 1/8 Watt type. 2) Variable pot. One turn pots are easier to see what they are set to. Ten turn pots allow more precise settings, but you cannot see what the current setting is, without dialling them down to the end of their travel. 3) Rosin core solder (small tube should be enough; 1lb lasts a lifetime) 4) Soldering iron (35watt, smaller tip is better) 5) Desoldering braid (or a "solder sucker" pump action device) 6) Wire (preferably tin, nickel, or silver plated, for easy soldering) 7) Wire strippers (preferably fixed gauge like 22-24-26-28-30 ga. in addition to a pair of the cheap adjustable version to handle other gauges of wire) 8) Polyolefin heat shrink tubing, to insulate solder joints from mobo. 9) Multimeter, with enough ranges to measure the kinds of resistors you plan on working with. Should have a high input impedance for measuring voltages (like higher than 10 megohms). No need to spend a lot of money to get a "modder" grade instrument. Try a hardware store, as sometimes they have after Christmas bargains on multimeters, whereas RadioShack pricing will remain fixed. Note - I haven't done business with Jameco in many years, so check resellerratings.com or Google or the like, before you buy. This is how I would do the mod, as safely as I know how to. Heat shrink tubing goes over the solder joints with the "+" sign on them. Potentiometer should be mechanically secured somehow, without stressing delicate motherboard components. The potentiometer terminals can be covered with electrical tape, after they've been soldered. Note that electrical tape leaves residue which can pollute a solder joint if you need to redo it later - clean it off with a solvent. chip---wire--+--resistor--+--wire----- ---wire---ground | | (no wire on third terminal) --- | | | potentiometer Keep all wires short. Long wires work like an antenna. Using wire relieves stress on the connections to the motherboard. If you wanted, you could solder one end of the resistor directly to the potentiometer, so only two wires are used. If you have any more questions, ask them _before_ you proceed :-) Plenty of people in this newsgroup know how to solder. HTH, Paul |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Vcore on the MSI Neo2 Platinum | David Johnstone | Overclocking AMD Processors | 2 | January 7th 05 11:50 AM |
Vcore measurement is low - Is it a problem? | Email Invalid | Overclocking | 2 | June 29th 04 11:29 PM |
Vcore measurement is low - Is it a problem? | Email Invalid | Homebuilt PC's | 2 | June 29th 04 11:29 PM |
Asus A7v8x-x with major vcore problems | 23_toemek | Asus Motherboards | 2 | December 15th 03 02:02 AM |
Req - A7N8X Deluxe Bios Help | jibby | Asus Motherboards | 0 | July 23rd 03 05:17 AM |