0
\$\begingroup\$

I am currently trying to solve an issue with a vintage PC power supply (Schematic here: http://techmattmillman.s3.amazonaws.com/wp-content/uploads/2017/01/5150_PSU.png)

At idle, there is about 5A load on the 5V rail, and pretty much no load on the +12V rail, during this time I can see with my scope that the power supply runs in DCM mode, resulting in it making an annoying hissing noise.

But then if I put a small load (say 200mA) on the 12V rail, it switches to CCM mode and becomes inaudible.

Here's the tricky part: If I instead put that extra dummy load on the 5V rail, but at 500mA, this does not have the same effect. The supply remains in DCM mode and gets even noisier! (I tried a whole range of loads: 100mA to 3A, they all have the same effect)

What on earth is going on here! What is the difference between loading the +12V rail and the +5V rail from a DCM/CCM point of view?

Why does the load have to be on the 12V rail to stabilise the supply?

\$\endgroup\$
12
  • \$\begingroup\$ Strange. Do you know the leakage inductance of each winding? Is the 5 V "loose"? How is 12 V cross regulation with respect to 5 V. \$\endgroup\$
    – winny
    Commented Feb 6, 2017 at 11:21
  • \$\begingroup\$ Definitely do not know leakage inductance. Do know that that 12V ouput floats at 15-18V, and is regulated back to 12V with a large linear regulator though. \$\endgroup\$
    – Matthew Millman
    Commented Feb 6, 2017 at 11:27
  • \$\begingroup\$ If you load the 12 V just slightly, is it still 15-18 V? If yes, then you have weak cross regulation/high leakage inductance. It's still strange though that 3 A on the 12 V does not cut it. Can you put an oscilloscope on it? \$\endgroup\$
    – winny
    Commented Feb 6, 2017 at 11:49
  • \$\begingroup\$ Should clarify, that 100mA to 3A was a dummy load on the 5V. 12V is limited to about 1.5A, and it does not take very much load to get it into CCM. What I have observed is that there needs to be a ratio of load between 5V and 12V to remain in CCM. As soon as the load on 5V gets proportionally larger than 12V load we go back to DCM. The pre-regulator voltage on the 12V does drop, i.e. 1A will drop it to 15V, but upping the load on the 5V to 8A will pull this up again, to 16-17V \$\endgroup\$
    – Matthew Millman
    Commented Feb 6, 2017 at 11:58
  • \$\begingroup\$ Again, sounds like terrible cross regulation. However, 1 W on the output regardless of 5 V or 12 V will load the primary almost identical and at the same point you should cross over into CCM. I don't know what your end goal is, but you can either fix the control loop to remain stable in DCM, change the PWM IC to one with burst mode or rewind the transformer. Say, does it squeek at around 8 kHz? \$\endgroup\$
    – winny
    Commented Feb 6, 2017 at 12:30

1 Answer 1

Reset to default
0
\$\begingroup\$

After much experimentation I find out that the cause of the problem was a degraded ESR in the capacitors on the secondary side. Solution: replace them.

How was this causing the issue I describe above? A greatly increased level of ripple on the 5V rail (the only which was regulated) was de-stabilising the regulation process, causing the supply to continuously swap between skip/continuous mode.

I do not entirely understand how but it seems that loading the 12V rail reduced the ripple on the 5V rail to the point where it made the whole supply stable.

\$\endgroup\$
1
  • \$\begingroup\$ The ESR zero was too high to be a problem before. As the ESR increased significantly, this new zero made the loop instable. Refer to page 3: exar.com/content/document.ashx?id=1244 \$\endgroup\$
    – winny
    Commented Feb 7, 2017 at 17:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.