900G power supply? Won't turn on (obvious blown caps)

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

It IS the same cap, but its ESR is increasing - which is exactly what I originally posted back in post#9. Do try to stay on track.

"Do try to stay on track."
ME stay on track?

This thread is about helping geocrasher find replacement caps.

If you want to ramble on expressing you MISGUIDED ideas and theories about how caps work go open your own thread.

.

[pedro]

Re: 900G power supply? Won't turn on (obvious blown caps)

It does.
It moves the dissipation from the cap [where you want it] to the load [where you don't want it].

AAARRGGHH!!

And more AARRGGHH!!

No..
1 - It reduces it -in the cap- installed.
2 - If you increase ESR it's not -THAT- cap anymore, you changed the cap.

1 - It's not tthat same cap.

It IS the same cap, but its ESR is increasing - which is exactly what I originally posted back in post#9. Do try to stay on track.

2 - As I said, a cap with higher ESR will have less ripple current going through it.

No it doesn't. <deep breath>

Less current through cap, same volts, less watts. -> less heat.

It ISN'T the same volts. See below or revisit post#16.

It's called Ohms Law. Please review it.

And then in post#20 you stated:

Psst - Ripple is AC.
ESR doesn't affect the DC.
DC is irrelevant to a discussion about Ripple.
ESR doesn't 'care' if the DC is 5v or 500v.

I asked (post#21) "Did I ever claim otherwise?" you responded:

Yes, you did.

Care to point out where?

I'm not sure whether you are being deliberately obtuse, argumentative for the hell of it, or just plain don't follow. I believe everyone is worth some effort to enlighten them, so here goes. I'll give it this one last go to explain why you're mistaken. Please try to keep up - I'll go through it slowly.

The ripple (AC) output of a switcher is a CURRENT SOURCE. The ripple current is directly related to the DC load current. So in a working piece of equipment (e.g. monitor) where the DC load is established, the ripple current is substantially constant.

The low-ESR filter caps are designed to shunt the overwhelming majority of that ripple current. As I explained in post#16, as the ESR rises with deterioration of the cap(s), the current through the cap(s) doesn't reduce by anything like the same proportion, and the I²R losses in the cap(s) rise. Refer again to the sample numbers in that earlier post#16 to understand why.

If you are having trouble understanding this, I suspect you are beyond comprehending something that is fairly basic.

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

You were carrying on about increasing ESR lowering the dissipation in a cap.

It does.
It moves the dissipation from the cap [where you want it] to the load [where you don't want it].

I'm not confused. Increasing ESR in a cap will increase the losses in that cap. Period.

No..
1 - It reduces it -in the cap- installed.
2 - If you increase ESR it's not -THAT- cap anymore, you changed the cap.

It all is. But I'm still unclear why YOU are consused about the increase in ESR resulting in an increase in dissipation in the cap.

1 - It's not tthat same cap.
2 - As I said, a cap with higher ESR will have less ripple current going through it.

Less current through cap, same volts, less watts. -> less heat.

It's called Ohms Law.
Please review it.

Of course. Did I ever claim otherwise?

Yes, you did.

.

[pedro]

Re: 900G power supply? Won't turn on (obvious blown caps)

I was never confused. I haven't changed a thing.

You were carrying on about increasing ESR lowering the dissipation in a cap.

I'm not sure why you're confused.
Perhaps it's your English comprehension.

I'm not confused. Increasing ESR in a cap will increase the losses in that cap. Period.

The Ripple that doesn't go through the cap goes through the IC [load].
Less through the cap -> more through the load.
Not sure why you can't comprehend that. It's simple.

It all is. But I'm still unclear why YOU are consused about the increase in ESR resulting in an increase in dissipation in the cap.

Psst - Ripple is AC.
ESR doesn't affect the DC.
DC is irrelevant to a discussion about Ripple.
ESR doesn't 'care' if the DC is 5v or 500v.
.

Of course. Did I ever claim otherwise?

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

Yup. And only a poofteenth less than before through the cap. You're slowly catching on.

I was never confused. I haven't changed a thing.

I'm not sure why you're confused.
Perhaps it's your English comprehension.

The Ripple that doesn't go through the cap goes through the IC [load].
Less through the cap -> more through the load.
Not sure why you can't comprehend that. It's simple.

Psst - Ripple is AC.
ESR doesn't affect the DC.
DC is irrelevant to a discussion about Ripple.
ESR doesn't 'care' if the DC is 5v or 500v.
.

[pedro]

Re: 900G power supply? Won't turn on (obvious blown caps)

Yup
At 99% going through the cap you have 1% going through the processor.
So you'd double the ESR.
At 98% going through the cap you now have 2% going through the processor.
Twice as much Ripple going through the Processor.

Yup. And only a poofteenth less than before through the cap. You're slowly catching on.

- Good Plan!
Let me know if it even boots with 200 mA Ripple on your CPU.

FFS the numbers were chosen to make it easy_for_you_to_follow.

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

Sorry, I ain't gonna let that pass without comment.

The switcher output will have two components - the (wanted ) DC and the unavoidable AC.



The amount that gets "re-routed" away from the cap will depend on, for want of a better term, the ... (AC-coupled) ESR of the load. The cap is what provides by far the greatest contribution to the overall shunt impedance at switcher frequency - that's its role. Quadrupling the cap's ESR will not halve the current through it - do the simple maths. Say - for the sake of simplicity - 1 A of ripple with a cap of 0.1 ohm and a load with an (AC) impedance of 10 ohms at switcher freq. Then approximately 99% of the ripple current passes through the cap, I²R dissipation ~100mW.

Double the cap's ESR to 0.2 ohms - now about 98% of the ripple current is shunted by the cap. Dissipation? I²R now a tad short of 200mW.

Yup
At 99% going through the cap you have 1% going through the processor.
So you'd double the ESR.
At 98% going through the cap you now have 2% going through the processor.
Twice as much Ripple going through the Processor.
- Good Plan!
Let me know if it even boots with 200 mA Ripple on your CPU.
.

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

This is picky as heck, but wouldn't it more correct to say that it's doing useful work in the load?

In the meantime, the total power dissipation in the system decreases with a lower ESR capacitor.

PlainBill

How is AC Ripple doing 'useful work' in a DC load?
Like for instance, a Processor.

[pedro]

Re: 900G power supply? Won't turn on (obvious blown caps)

What you originally thought was correct.

Nothing changes the creation of the ripple quantitatively but the cap is NOT the load.
-> The cap is in parallel with the load.

Sorry, I ain't gonna let that pass without comment.

The switcher output will have two components - the (wanted ) DC and the unavoidable AC.

When you raise ESR the total Ripple current created at the switchers doesn't change but what isn't going through the cap to ground is now going through the load to ground.

Raising ESR reroutes Ripple to the load.

That amount of Ripple Current that gets redirected to the load isn't heating the cap by I²R anymore,
it's heating the load by I²R instead.

The amount that gets "re-routed" away from the cap will depend on, for want of a better term, the ... (AC-coupled) ESR of the load. The cap is what provides by far the greatest contribution to the overall shunt impedance at switcher frequency - that's its role. Quadrupling the cap's ESR will not halve the current through it - do the simple maths. Say - for the sake of simplicity - 1 A of ripple with a cap of 0.1 ohm and a load with an (AC) impedance of 10 ohms at switcher freq. Then approximately 99% of the ripple current passes through the cap, I²R dissipation ~100mW.

Double the cap's ESR to 0.2 ohms - now about 98% of the ripple current is shunted by the cap. Dissipation? I²R now a tad short of 200mW.

You could also do this experimentally by sticking say a pair of 0.1 ohm resistors in series with a known good (measured) low-ESR cap. Stick a DVM across one and monitor the (obviously AC) voltage as you apply a short across the other one. You may be surprised.

Anyway, if you don't believe me that's your choice, but it won't change my view.

[PlainBill]

Re: 900G power supply? Won't turn on (obvious blown caps)

Well it does do useful work, but with unwanted consequences...

You don't want a working monitor?

The load - the logic card, the LCD panel, and the inverter draw whatever power they need to operate as long as the ESR of the filter caps is low enough to keep the ripple down to satisfactory levels. It doesn't matter if the ESR of the caps is right at the critical point, or if the caps are perfect - let's say 1/10 of the critical level; the load is going to be drawing the same current. However, caps at the critical level are going to be dissipating more power than perfect caps.

Now, I wouldn't expect to see a major savings on your electric bill, but I would expect to see the monitor running at a lower temperature.

PlainBill

[Th3_uN1Qu3]

Re: 900G power supply? Won't turn on (obvious blown caps)

This is picky as heck, but wouldn't it more correct to say that it's doing useful work in the load?

Well it does do useful work, but with unwanted consequences...

[PlainBill]

Re: 900G power supply? Won't turn on (obvious blown caps)

What you originally thought was correct.

Nothing changes the creation of the ripple quantitatively but the cap is NOT the load.
-> The cap is in parallel with the load.
When you raise ESR the total Ripple current created at the switchers doesn't change but what isn't going through the cap to ground is now going through the load to ground.

Raising ESR reroutes Ripple to the load.

That amount of Ripple Current that gets redirected to the load isn't heating the cap by I²R anymore,
it's heating the load by I²R instead.
.
.

This is picky as heck, but wouldn't it more correct to say that it's doing useful work in the load?

In the meantime, the total power dissipation in the system decreases with a lower ESR capacitor.

PlainBill

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

I originally thought that way too, but it seems to generally not be the case in switcher filters. You have an AC source (the ripple component on top of the DC) which is effectively a current source - directly related to the DC current component from the upstream switcher or whatever - so the ripple current doesn't drop off materially as the ESR rises.

What you originally thought was correct.

Nothing changes the creation of the ripple quantitatively but the cap is NOT the load.
-> The cap is in parallel with the load.
When you raise ESR the total Ripple current created at the switchers doesn't change but what isn't going through the cap to ground is now going through the load to ground.

Raising ESR reroutes Ripple to the load.

That amount of Ripple Current that gets redirected to the load isn't heating the cap by I²R anymore,
it's heating the load by I²R instead.
.
.

[pedro]

Re: 900G power supply? Won't turn on (obvious blown caps)

Probably but it depends on the caps involved.

As the ESR increases less Ripple current passes through the cap making the I in I²R smaller and since I [or I²] is usually bigger than R, I would expect the total of I²R to go down.

I originally thought that way too, but it seems to generally not be the case in switcher filters. You have an AC source (the ripple component on top of the DC) which is effectively a current source - directly related to the DC current component from the upstream switcher or whatever - so the ripple current doesn't drop off materially as the ESR rises.

But:
Typically caps with a higher ESR rating will have a lower Ripple current rating too.

The Ripple current rating is how much Ripple current the cap can pass through without over heating internally. [Basically it's based on I²R.]

Yup.

When choosing replacement caps you want:
- The same or lower ESR.
- The same or higher Ripple Current rating.
- The same or higher voltage rating.
- The same uF.

Some people play around with uF values.
It's a rare case where it's not easy to get caps with the same uF.
If you MUST change uF then it's better to go up one standard value because uF goes down as caps age.
.

More "yup".

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

And as the ESR increases, so does the heating of the cap from I2R due to the ripple current. The heating accelerates the ageing (drying out) of electros and soon you have either a failed cap or a failed component in the affected circuitry. So while too low an ESR *can* in some cases create a problem, too high an ESR is fairly sure to.

Probably but it depends on the caps involved.

As the ESR increases less Ripple current passes through the cap making the I in I²R smaller and since I [or I²] is usually bigger than R, I would expect the total of I²R to go down.

But:
Typically caps with a higher ESR rating will have a lower Ripple current rating too.

The Ripple current rating is how much Ripple current the cap can pass through without over heating internally. [Basically it's based on I²R.]

You can check.
The Ripple current rating is given in the same data sheet as the ESR.

..

When choosing replacement caps you want:
- The same or lower ESR.
- The same or higher Ripple Current rating.
- The same or higher voltage rating.
- The same uF.

Some people play around with uF values.
It's a rare case where it's not easy to get caps with the same uF.
If you MUST change uF then it's better to go up one standard value because uF goes down as caps age.
.

[pedro]

Re: 900G power supply? Won't turn on (obvious blown caps)

ESR (equivalent series resistance) is basically the 'resistance' of a capacitor TO AC, that is not attributable to capacitance (uF). [As in it is not Xc.]
-
The Low ESR cap's job is to pass the noise left over from from the transistor switching action in the SMPS to ground. [noise = ripple in this case.]
-
A small ESR [small resistance] to 100 kHz will 'short' the 100 kHz noise to ground thereby removing it from the DC power sent from the supply.

And as the ESR increases, so does the heating of the cap from I2R due to the ripple current. The heating accelerates the ageing (drying out) of electros and soon you have either a failed cap or a failed component in the affected circuitry. So while too low an ESR *can* in some cases create a problem, too high an ESR is fairly sure to.

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

Use 105C caps. [Temp rating.]

[PCBONEZ]

Re: 900G power supply? Won't turn on (obvious blown caps)

ESR (equivalent series resistance) is basically the 'resistance' of a capacitor TO AC, that is not attributable to capacitance (uF). [As in it is not Xc.]
-
The values of both Xc and ESR are frequency dependent.
-
The Frequencies of Ripple encountered due to SMPS action are right around 100 kHz.
[Probably most have Ripple between 25kHz and 200kHz. Data sheets for Low ESR caps are standardized to 100 kHz.]
-
At 100 kHz the ESR has a large affect on the caps 'resistance' to AC and Xc has almost none.

The Low ESR cap's job is to pass the noise left over from from the transistor switching action in the SMPS to ground. [noise = ripple in this case.]
-
A small ESR [small resistance] to 100 kHz will 'short' the 100 kHz noise to ground thereby removing it from the DC power sent from the supply.

That said:
It is best to stay as close to the original ESR as possible as generally there are inductors as part of the 'filter' that includes the cap.
The Inductor and Capacitor's values [uH of the inductor and ESR of the cap ] are chosen so they work together to get the best filtering.
If you go wild and jump from like .050 ohms ESR to .015 ohms ESR you may de-tune the filter and end up with more noise instead of less.
It's never good to go to a higher ESR because that will remove less ripple even without being de-tuned.

FC are a good all around cap for working with monitors [typical grade of the originals] but occasionally there will be an original crap cap with better ratings than FC.
That's why -if possible[**]- always check the specs on the old caps before choosing replacements.
If not possible then fail-safe to FM.
[**] Some crap brands simply don't have data sheets available.
.

[jetadm123]

Re: 900G power supply? Won't turn on (obvious blown caps)

I suggest Panansonic FM or FC caps. FM have better specs and are about the same price if you purchase through digikey.com. You can order the cap using the steps below:


Here's the best way to order from digikey:

1) Go to digikey.com

2) In the "parts search" window type: panasonic fm and select "GO" button

3) Under the Capacitor heading select "Aluminum"

4) The first two columns you see are capacitance and voltage. Highlight the value and voltage with your mouse and select the "Apply Filters" button

5) You'll see several selections available. Here's what to look for:


a) Look at the quantity available and minimum purchase. If "0" are available, then there's none to purchase. Some selections require a minumum purchase of a 1000 units, which you also don't want. I think you get the picture.

b) height and diameter. Measure the original caps and compare to what's listed.


6) If your value is out of stock, go back to the beginning and type in "panasonic fc" and go through the same process.

7) During the checkout process select USPS First Class shipping, since this will give the best ship rate.

[geocrasher]

Re: 900G power supply? Won't turn on (obvious blown caps)

Hi Guys, thank you! I'm in the US. Reno, NV to be specific.