I need to test by substitution

Long shot, but some motherboards need a BIOS update before they will boot CPUs that were released after the motherboard. You had to borrow a previous-gen CPU to do the update. What motherboard do you have?

Which is another reason for:

Not local nor have suitable bits, I'm afraid. Ask a local computer shop?

Theo

Shops which do builds for customers, should have the staff to accommodate a test request. A reasonable charge, is one hour labour. Ask in advance what they charge, so there are no surprises. They can bring up the BIOS, do a quick memtest, to prove most bits of the thing are working. The modern UEFI also allows a little keyboard/mouse testing. Even memtest takes keyboard input. And UEFI uses the mouse.

There were reports many years ago, that a mom&pop store would charge $25 for a basic test. And it's easier today, since you don't need a video card to get most new stuff working. There may be a GPU inside the CPU for the purpose. That is one less component to worry about in a basic test.

This is not like the two hundred they charge, where they "insist on installing an OS" as a tidy earner for themselves. What you're asking for is just a hardware test. And they will have seen this request before.

Even when I returned my first hard drive ever, that had infant mortality (motor would not spin), the staffer tested it right in front of me and agreed it was defective. Part of the reason they examine returned items, is in case of fraud (the returned item isn't really the item it appears to be).

Diagnostics Service Labour Charges System Checkup. (LADTDI0001) Free System Diagnostic. (LADTDI0002) $49.99 <=== External/ Internal HDD or SSD Diagnostic. (LADTDI0003) $14.99

System Build. (LADTSY0001) $59.99 System Build with Operating System. (LADTSY0002) $79.99

The latter two, assume components purchased at the store. So that's an assembly fee. And with Windows 10 installing its own drivers, there is less work today to installing an OS. The fee for the boxed copy of Win10, would be a lot higher than the fee to install it.

There are some stores, I would not even check the web site :-) They're that shady.

Paul

Are there any leds illuminated on the motherboard when you switch on, or is it completely dead..?

Is the on/off button connected to the correct header..?

First, test the PSU completely unplugged from the system.

On the 24 (or 20) pin connector, short the green wire to any black wire, it should start up and stay running.

IS there any memory on the mobo and is it the right type of memory?

And also check whether the Mobo needs extra power leads plugged in too as well as the standard ATX 24 pin powwer plug...

Also some graphics cards also need power plugged into the card itself, usually 4 pin or 6 pin or 8 pin with black and yellow wires.

neither will start without teh supplementary power connected.

ALso is there a series of beeps or a series of flashing lights? this can be a code to tell you what failed at POST (Power On Self test) and the mobo and bios manufacturer should have a look up table somewhere so you can deduce what the sequence of beeps/light flashes is trying to tell you.

Does anything come to life.

Do PCs still have (BIOS) beeps to indicate a fault prior to booting?

Example

Good point. Make sure all the power inlet connectors on the mobo and GPU are wired to the PSU.

Some mobos have 7-segment LEDs displaying a two digit POST code. You can also get add-in cards that show it:

Theo

And another one, does the switch on the case actually work or is the wrong button on the case being pressed or is the wrong pair of wires being plugged onto the mobo's pins?

I've used a simple jumper to short the PW-SW pair of pins as a substitution in case the case switch was broken.

S.

Been a long time since I last bought components from these guys, but they may be helpful or know one of their customers that can help. They are in Peerglow Industrial Estate, not far from Watford Football stadium.

Unplugged, without any load on the PSU, many/most/all will simply shut themselves back down.

There's a 500W Seasonic ATX PSU sitting on my bench that begs to differ, all rails stable at correct voltages ...

There was a brief interval, where ATX PSU had two rows of numbers. One row was "minimum load" for regulation purposes. The other row (the normal information) was max output limits. And that documented the anomalous behavior of supply design at the time.

Supplies now, just have the normal information. One row of numbers (for max), means the "min load" is zero.

And, they remain running when PS_ON# is asserted as a logic level. Ground PS_ON# and the supply comes on as long as it is grounded.

The era of flaky implementation has pretty well ended. Lots of double forward conversion designs, and high efficiency.

Also gone, are the extra-long ATX supplies that had four transformers instead of one, and those could not safely be used without a wiring diagram showing which rails were on which lead-sets. Generally, the supplies use one giant 12V transformer, and for forward conversion, a separate board takes +12V in and makes +5V and +3.3V from it. And the total output of the 5V/3.3V board is around 100W or a bit more.

Some of the PSU designs, had a 5W or 10W load inside, to serve as a stabilizer. And that's how they could run "open circuit", by not actually being open circuit at the time.

The voltage regulation on ATX supplies is not push-pull. The device is push-only. This means, theoretically, the switching converter on the primary side, would have to stop, or set the PWM width to zero, to stop charging up the buffer caps on the output.

To characterize exactly what the PSU is doing no load, you could use your Kill-a-Watt meter while carrying out your load tests. That might show whether it had a dummy load, a switchable dummy load, or whatever.

Another way to decide what it is doing, is use a pickup coil and sample for mag field. And see if you can see a signal say from 100KHz or higher on the pickup coil. As evidence the switcher was running.

But generally, the supervisor has to keep running, even if the switching action on the main section was being modulated. And the supervisor has had more than one implementation, linear on some, SMPS on others.

This design (yes, it's really old), you can see it is using dummy loads.

The 3.3V also has some "gadget" connected to the output, which almost looks like a shunt regulator.

Paul

The last status report, was no PS_ON# assertion was evident (no fans spinning). The board and processor have been tested, so that means someone got the board to drive PS_ON# and make the PSU come up.

A momentary contact on PWR SW header pin and GND, should be enough to start it running.

There are two motherboards with close to the same name, one has the (now more or less standard) four white LEDs for POST sequence, the other has no LEDs. Not even the traditional Asus Green LED on +5VSB is being used any more (the green LED goes out, indicating it is safe to add or remove components from the motherboard).

But the white LEDs don't do anything, unless you use the PWR SW pin to kick off the sequence.

Stripping all the LEDs off a motherboard, is pretty cheesy. Especially in this case, considering the retail price of the board.

So for this debug, "fans spinning" is the next stop. The BIOS version should be OK (the initial BIOS release is supposed to work with the processor).

Some Asus boards, the 2x4 PWR_SW, RST_SW, and two LEDs thing, there is a header in the box, allowing you to fit the front panel wires, then seat the header onto the mating header for FP on the motherboard. That's good for people having trouble seeing the legends on that interface. That thing normally comes in a separate plastic bag. Again, at the price of this motherboard, one of those (as it's Asus) should be included. There could well be a patent on the thing for all I know.

Paul

IIRC you need a load on the 5V before the other rails will come up. That's been the recieved wisdom from those who wanted to make 12v power supplies out of them...

What, like this?

I do have one of these, but it is temporarily locationally challenged at the moment. From the pictures, it does look as if the only load is the LED, but that must be sufficient, as mine works.

A year ago I was caught out by a new motherboard that needed 16 pins for the cpu when my existing psu had fewer, 12 I think. When I connected them I assumed (wrongly, it turned out) that the system would run but at a lower speed. But when it failed to power up I forgot about the pin shortage on the cpu. Oops. One new psu with the right number of pins for the cpu and everything was ok.

I actually made a visit to them a couple of months ago but they didn't stock what I wanted for this build.

Alan