• Please review our updated Terms and Rules here

A Puff of Smoke and All is Well

Grandcheapskate

Veteran Member
Joined
Oct 9, 2014
Messages
886
Location
New Jersey, USA
Hi Guys,
I have a P90 computer I do not use very often. The other day I went to boot it up and it went on for about a second and then off. I tried a couple more times with the same result.
Today I took the cover off and unplugged the power supply from the motherboard and the machine stayed on (CPU fan spun) so I figured the power supply was not shorting out. I reconnected the power supply and again the machine only stayed on for a second. Figuring it may be one of the devices I removed the power from the hard drive and booted the machine.
Here's where it gets weird. There was a puff of smoke and a sound like escaping air. But the machine stayed on and got to the point where it asked for a boot disk. I shut the machine down and reconnected the hard drive. I rebooted the machine and everything is working.
I have not checked over the motherboard but my guess is it blew a capacitor. Maybe disconnecting the hard drive allowed enough power to flow to a bad capacitor.
Are there situations where a capacitor can blow and yet the machine continues to work?

Thanks...Joe
 
Happens a lot with tantalum caps, but they usually just go "bang" like a squib. Check the Rifa film caps in the power supply AC filters--those can smoke a lot. Inspect electrolytic caps for "doming" or leakage--when they go, you'll get that "whistle".
 
The other nite I heard a loud bang and the video on my LCD went blank. I figured crap I blew my video card in my main machine, but it was still running but the KVM lights were off.

Ends up a 470uf 25V thru hole capacitor exploded in the power brick to my KVM. Installed a spare and everything works again. Never seen that before.
 
A partially shorted tantalum bead capacitor could cause a current overload with the result that the power supply shuts down before it goes bang.

Removing the drive will cause the excess current to flow in the faulty capacitor. It still may not go bang, but (in your case) it looks like it burnt it sufficiently well to cause it to go open circuit.

Hunt it down and replace it...

Dave
 
Thanks guys. I looked at the MB and I do not see any caps with mushroomed tops so if there is a blow/damaged capacitor it must be inside the power supply. I'll have to remove a couple CD drives to get at it. I kept the machine running yesterday for about an hour and there were no other signs of trouble.

Thanks...Joe
 
Regarding the "Rifa" that Chuck referred to. That is an informal name. See the 'Line Suppression Capacitor' section at [here]. The crack in those is normally along the side. In that section is a link to a photo of a failed WIMA made capacitor, and from that, it can be seen that sometimes the crack is relatively small, hidden if the capacitor is up close to some other component.
 
This machine has been idle for months but I again looked at the MB for a blown capacitor as well as opening the power supply. I see nothing indicating anything wrong. And the machine continues to work flawlessly for hours on end, even drawing the power to run two CD drives simultaneously.
 
Regarding the "Rifa" that Chuck referred to. That is an informal name.
Well, it's also the proper brand name of one manufacturer of those types of caps. So they could well be actual RIFAs. And RIFA assures us that they are quite reliable, and eliminate fire risk.

rifa-cap-advert.jpg

So...."Don't Fear the RIFA"?
 
Advertisement written by the "promise department". No one wanted to mention how long they would last for (that is the "sorry department's" job)...

Dave
 
The whole thing about generating radio interference without the capacitors is complete and total nonsense now, with what is deemed acceptable with modern switch mode power supplies. Most of these generate RFI that splats out the entire MW & SW radio bands and this whether you have the Rifa caps on the line input power feeds or not.

But the manufacturers of these switch mode supplies, be they in your computer, big screen TV kitchen appliance or Wall Wart supply don't care because the general consensus is that the population of people who would be listening to these radio bands are dying out. As long as the interference doesn't affect internet navigation, this is all that counts it seems.

If you get any appliance, with an iron cored power transformer and an analog design supply, whether is has the X2 capacitors on the primary or not, it will create only a small fraction of the RFI created by modern SMPS's in everything else. Therefore I simply remove these troublesome Rifa caps on the line supply feeds from all my vintage gear, and put them in the bin, where they belong.

PS: there is no such thing as "self Healing" it is the biggest load of marketing BS you will ever hear. Healing is a property of biological systems only. What is really is, is the metallic film in the capacitors being destroyed (vaporized) by over voltage transients and as a result the uF capacity drops with time. The uF value never "heals" and never recovers because the metallic film is lost. This causes appliances to fail where X2 caps are used as dropper capacitors.

Long before Rifa style X2 capacitors appeared on the market, the approach by most manufacturers was to simply used standard capacitors on the line, but with very high voltage ratings of greater than 1kV to 2kV, and they virtually never failed. But of course these were big sized parts, the better ones were sealed in metal containers mitigating fire risk much much better than any Rifa part. Even appliances I have that were made pre-WW2, these capacitors are normally fine over 80 years later, you can't say that for Rifas. As time went by, people wanted "small" and the size of the Rifa parts, for their voltage and uF rating, were quite small, hence the insulation thinner.

The Rifa caps also can short out as well as burn up. Unfortunately there was a tendency for these to be placed across switch contacts in series with motors in industrial machinery, and a failure can start a motor, I have seen this happen. I was woken one night from sleep when a sewing machine in the room started up because a Rifa cap across the control pedal's terminals spontaneously shorted out and it started running at full speed.

If I felt compelled to filter the line power inputs to any of my gear for RFI, I would much rather use a 2kV rated poly cap, than a Rifa X2, even if it is not officially rated for the task, because I know that would be better and safer, or better still, a 2kV rated part in a metal canister which gives the better fire mitigation. Modern gear though has no physical place for such a large sized part. So, to some extent, we are looking at one of the prices to pay for miniaturization.

I have been conducting an experiment over the last decade, using modern X2 rated capacitors as droppers. For example I have some 1930's vintage General Electric Cyclometer clocks. They use a shaded pole motor and have a lamp, the total power consumption is about 3 to 4 watts. They are designed to run from 115V 60Hz, but I modified the rotors to correct them to 50Hz use. To run them off 230V here, I used a series capacitor of around 1.5uF. Initially I used an X2 ceiling fan type, after about 3 to 4 months I noticed the lamp getting dim and the clock stopped. Testing showed that the X2 cap was down to about 50% its initial capacity, welcome to self healing, the same thing happens in people's ceiling fans. I then fitted a 2kV rated poly cap and and these clocks have been running fine for the last 5 years, testing shows no loss of capacity. In the application it is not risky because the capacitor is in a metal box and I have added fuses, though some argue not necessary because of the very fine sized wire in the motor coil.
 

Attachments

  • Clock.jpg
    Clock.jpg
    124.5 KB · Views: 7
Last edited:
Before the metal-film caps became the standard for line bypass, large ceramic disc capacitors were often used. They don't fail short . Because of self-heating effects, and voltage spikes, they could fracture. Otherwise, they're pretty safe.
 
Before the metal-film caps became the standard for line bypass, large ceramic disc capacitors were often used. They don't fail short . Because of self-heating effects, and voltage spikes, they could fracture. Otherwise, they're pretty safe.
Yes, the ceramics are better than Rifas, mostly. Generally these are used in the metal enclosed IEC panel connectors that contain a line filter.

However, excluding the X2 capacitors, Rifa did make some very good capacitors. One type of electrolytic they made was excellent, with a special seal and rated to around 130 Deg C . Used in under the bonnet car auto-electrics and other harsh environment applications, CDI units etc, I have some made in the 1970's and they are still as good as new.

Another reason I could have mentioned that many line powered instruments that have a power transformer (at least lab grade ones) have an electrostatic shield between the primary and secondary winding that is grounded, and this prevents any significant RFI getting in or out via the transformer and line cord. The transformer's core has little/no response at radio frequencies and the line filter caps are more of a hazard than a help. But even without the shield I have not seen any issues. The main problem is all the modern SMP's on the power line.

About the remark on post #10: "And RIFA assures us that they are quite reliable, and eliminate fire risk"

Rifa caps do not eliminate fire risk, they mitigate it it. They can and do burn, Whether that causes a fire depends on the adjacent components and if they might be flammable and if the instrument is in a metal or plastic housing. Plastic housings often make an excellent hydrocarbon fuel. The only way to eliminate fire risk from the actual capacitor is to have it sealed in a metal housing, but this is a lot less compact and uses up more space.

It is like calling a company a Pest Exterminator. Very few if any pests ever get exterminated or made extinct, the sensible people call it "Pest Control". So if the marketing says "eliminate fire risk" think twice. Also, it is ironic to some degree that metalized coatings are known to burn, and paper is an excellent fuel. In fact metalized plastic or paper, once alight is very hard to put out, this has happened on skyscrapers that have had aluminium coated plastic panels fitted to their exterior, I think there was a classic case of this in the UK. The bottom line if you don't want an appliance causing a fire, the first move is to put it in a metal, not a plastic housing.

While the metalized paper in the Rifa cap is very resistant to burning on its own because of the tight winding and limited oxygen supply, the thing is once it starts to burn and electrically leak the energy from the line supply, of which there is plenty available, ensures the combustion. One trick is to put a low ohms, say 10 to 33R 1/4W resistor in series with the Rifa, it will still snub off some RFI, but when the capacitor fails( as it always will with enough time), the resistor goes open. Tek did this in some of their scope supplies.
 
Last edited:
I know that Rifas are popular with the tube audio folks (I've got a tub of the things from the µF to pF range that I picked up many years ago at a "buy them by the pound" type of parking-lot sale. Generally, the metallized film caps do perform quite well; one of my chainsaws uses one of my bargain caps in the ignition circuit. A good MF cap has very low ESR (changes with frequency, of course), which is an advantage in some cases.
 
Back
Top