PS voltages
I recently installed MBM5. And at Ageek's suggestion( :thumbup Thank you), I started a .txt log to keep an eye on my machine.
After about 10 hours of logging, couple of things caught my eye;
Readings are from a Thermaltake 420 watt
the +12 runs 11.92
the -12 runs -11.86
the +5 runs 5.03
the -5 runs -4.99 to -5.04
The +/-12v. rails almost never change. some changes in the +5 rail, and lots in the -5 rail.
I realise these will change as the load goes up and down. But how much is acceptable? Should the 12v. rails always be low?
Can I adjust the rails from inside the PS or do I need to? How easy is it to blow it up, trying?
Thank you;
Dragstk
After about 10 hours of logging, couple of things caught my eye;
Readings are from a Thermaltake 420 watt
the +12 runs 11.92
the -12 runs -11.86
the +5 runs 5.03
the -5 runs -4.99 to -5.04
The +/-12v. rails almost never change. some changes in the +5 rail, and lots in the -5 rail.
I realise these will change as the load goes up and down. But how much is acceptable? Should the 12v. rails always be low?
Can I adjust the rails from inside the PS or do I need to? How easy is it to blow it up, trying?
Thank you;
Dragstk
0
Comments
Thanks Geeky1
Both 12 and 5 are fine, ditto the inverted 12 and 5 (negative voltages). If a test of back-probing one of the unused HD connectors (while computer is runnig is close to what MBM5 shows, then you can declare 12 and 5 fine for legs.
I would try a backprobe this way. Take a connector for HD that is not in use, with computer on, connect the RED (HOT) probe to the back of the yellow (shove probe in next to metal inside socket, from WIRE side. On the Black half of leg, you can use the Black lead next to the 5 volt RED (not the RED wire, the black one next to it. Now let your VOM or digital multimeter cycle to stable with the machine under load-- running, folding, etc. That gets you 12 volt backprobe reading.
For 5 volt, use RED lead to RED wire metal inside connector, and Black to other black wire. Let it settle, you have a 5 volt backprobed reading.
Ok, milspec, tightest you can get, is about 5% flux for a normal milspec PSU:
so, .05X12 =.60 VCD variance acceptable for 12 volt line.
Range of 11.4 VDC to 12.6 VDC is acceptable for MILSPEC-- very, very good PSU.
.05x5 is .25 Volts variance acceptable for 5 volt line.
Range: 4.75 VDC to 5.25 VDC for MILSPEC PSU.
But, average PSU is not MILSPEC grade, it is consumer grade. Normally, a 7% range there is what I look for. So, ranges bacome:
.07X12= .84 VDC variance
So, 11.16 VDC to 12.84 VDC is consumer grade, inexpensive PSU variance that mfr will accept very willingly.
.07X5=.35 VDC.
So, 4.65 VDC to 5.35 VDC is acceptable for a not very good PSU.
Negative voltages are inverts of these, put minus sign in front for range that are considered good.
Your PSU has VERY good output from MBM 5, see if the backprobes match that.
John D.
Learning the why, along with the what to do, makes the understanding easier.
Thanks again;
Dragstk
Um, err.... In a modern system, the negative voltages are gotten with reverse polarity. They are still needed, how do you zero a high bit in RAM except by dropping the voltage??? How do you do this??? You OFFSET a positive 2.7 or so with a polarity inverted 2.7 (comes out to negative 2.7). The so-called negative voltages do NOT come out of the PSU, in the sense of what it supplies, that is right, but they ARE used with negative polarity circuitry in the circuits. They have to exist in some degree, as the circuits are so big and parts so far from ground (in terms of the tiny components at transistor level that current has to flow through to allow complex and large chunk logic) that without them present to compensate parts of the circuit would have be very resistive, wasting power and turning it into heat. To do ALL this by resistance would be prohibitive.
What you are saying is basically in regard to PSU, the PSUs do not have true inverters in them to supply discrete negative voltage lines in most cases. However, voltage levels are still inverted in other parts of the computer, in the circuitry. The place they get inverted changed, they still get inverted with polarity of circuit changes INSIDE the circuit, to compensate for bad ground areas. NO SUCH THING as a universally ground drop circuit, we cannot make components that perfectly yet.
Ideally, the negative voltages should mathematically be equal to or LESS THAN, the positives. For instance, a 12.5+ and a -13.01 is acceptable. Machine thinks of anything as less than 0 as 0. Pure positive integer, nothing less than zero. And, zero is a valid value. But the negative can pull a bit value to zero where needed and a ground is not perfect.
I will give you a simple example:
Say you want a fan at about half speed-- you can resitance down the input voltage, with a thermistor or rheostat, or you can set up a voltage offset connection from 12 to 5 volts instead of 12 to 0 (ground) and get 7 volts through fan and lower speed and quieter fan. Internals of computer are still analog\digital blends overall. IC internals are not perfectly pure digital, they have compensation designed into them for imperfections in mfring and design, or workarounds for design errors in many cases.
To replace a one with a zero in a RAM bit inside the chip, you have to pull down the voltage. To do so fast, you use inverted voltages internal to circuit. BUT, it is polarity inversion and inverters are not longer needed EXPLICITLY as visible things in PSUs.
For PSU output legs, the negative can be discounted, for circuitry internals, NOT as there is not any such thing as a totally perfect ground in reality. The negative voltage presences can be used to eval ground and reveal compensation needed for imperfect ground. But it IS done inside the circuitry and not in the PSU for the most part. Otherwise, the boxes would be one heck of a lot hotter than they are if it were done totally with resistance. I speak to what is actually being done, not to theory, here. Inside circuit, you CAN have a negative voltage in some places at some tiny finite times. Theory meets reality again.
MARC, I think you understand this, I simply think that saying "discount negative voltages" is not totally true. And have empirically confirmed that fact.
John D.