SCR Failure Analysis - Soft Starters 1

[GGOSS]

Hello all,

Can anyone refer me to a web site or on-line document that contains photographic images of failed SCR's with explanations on fault cause?

Look forward to hearing from you.

Regards,
GGOSS

 

[jOmega]

You can use a search engine such as

http://www.google.com,

etc. to search the web.

SCR failures typically come from three main causes:

- too much voltage

- too much current

- too much heat


or any combination of the three.


If your application is a SoftStart.... I would ask,
- are you violating the duty cycle limitations ; i.e starting too often without proper elapsed time between starts ?
- Does the unit have proper cooling ?

Also, allowing the SoftStart to source high current for too long a time will raise the heat, and cause SCR degradation and ultimate failure (elevated heat reduces the reverse blocking capability of the devices and they fail)...

Be sure you are operating the SoftStart within the limits and guidelines of the manufacturer.

If you are, the question the plant power. Is it subjected to transient spikes ? A suitable line analyzer can be used to monitor the input voltage to the SoftStart....

Finaly, a defective motor or an interfmittently defective motor...(i.e., stator winding shorts or grounds) can cause failure....


As to pictures of failed SCR's ... you would need an SEM (Scanning Electron Microscope) to be able to see the necessary detail, and an expert in semiconductor analysis to interpret the photos, to be able to "pin-the-tail-on-the-donkey".

 

[GGOSS]

Hello jOmega,

Many thanks for your help. I have exhausted all possibilities with google.

For your information I do not have a problem with failing SCR's as such. What I am looking for is images of failed devices with notations re fault cause from independant bodies. My aim is then to use this information/present to customers who may have experienced a failure for their reference.

Having said all this, I believe you have covered most causes of SCR failures in your post and as such the information will be useful to other forum members.

Here are a couple of other fault causes.

1. One cause of SCR failure (although not common) is transport damage. As you will be aware 'puck' style SCR's are used in larger soft starters. These are generally 'tensioned' between two pieces of heatink forming a stack assembly. I am aware of 3 cases now in which the SCR's have been crushed during transport. In all cases the starter had been dropped flat on its back (or its side dependant on product/design) so there were no obvious signs of physical damage, yet when checking the SCR's they measured short ie zero ohms.

2. I have also experienced heat sensitive gates that result in severe misfiring, this being at temperatures marginally higher than ambient ie say 40 - 45 celcius. Again this is not common but I have come accross a few. Needless to say, these generally result from manufacturing defects.

Thanks again and regards,
GGOSS

 

[nukeman]

GGOSS,
If you do find such a site please post it. I tried to find similar information several years ago but could not. I was evaluating multiple failures of "hockey puck" style SCRs in several inverters. I do have some SEM photos of our failures if you are interested.

 

[Donale]

Westinghouse, then Power X, use to have a post mortum of photographs in their Power Semiconductor Handbook. Its probably out of print. IR had one too. Check with the power semiconductor folks I'm sure they have references. Typically failure at the edges were caused by either voltage or lack of cooling since those have the poorest cooling. As the device heats up its dv/dt rating drops. If it was around a gate it was typically caused by over current, di/dt. The size of the damage also indicated if the fault was di/dt limited helping to determine if the recovery/protection circuitry and fuses were proper.

 

[GGOSS]

Many thanks Nukeman and Donale,

I have been conducting post mortum analysis of SCR's for 20 years now and still have difficulty with end clients who are not prepared to accept the reports issued to them, hence the need to identify an independant source to back up my claims.

I have tried to obtain the information from several SCR manufacturers and have had no joy from them. It would seem that many of them charge for the service of diagnosing and reporting on failure modes, hence their reluctance to make the information readilly available to people like you and I.

I will most certainly try the companies and documentation referred to in your posts and hopefully have a win.

In the meantime (nukeman) receiving copies, hard or electronic' of the images you have available would be greatly appreciated. As an alternative, posting them here (not sure if that's possible) would benefit all forum members.

Regards,
GGOSS

 

[nukeman]

I tried posting the photos to this forum but not sure how. If you include your email address I will send it to you and anyone else interested.

 

[jOmega]

Nukeman,

Would appreciate seeing the pics ..

Thanks,

jOmega
mailto:jomega@email.com

 

[electricpete]

If you e-mail them to me, I can post them on the web with a link back here on the forum. My e-mail address is on my profile. (I think you get junk-mail if you post your e-mail too many places)

 

[electricuwe]

To my experience it's quite difficult to analyse a failure based on the image of a failed SCR.

In the past I had access to some literature from Siemens on that topic but the problem was that there is no way to track a failure to a single reason.

Commonly in most circuits a voltage breakdown leads to an excessive current flow resulting in a severe destruction of the thyristor wafer.

 

[jOmega]

Electricuwe,

My experience over the last 43 years is that there are some very talented, skilled engineers out there that are very adept at analysising semiconductor failures and coming up with the failure mode.

Over the years, I've been fortunate to work with such specialists at Westinghouse, Reliance Electric, IR, Motorola, GE, Hi Rel Labs, and many others.

Voltage faults are percussive; current faults are a fusing.
And yes, as you state, one fault can mitigate to the other. But that is where the knowledge, talent and experience come in to play.

 

[GGOSS]

Hello Nukeman & Electricpete,

Nukeman, I think the offer made by Electricpete (email images to Electricpete and he will post via a link) presents the best possible outcome for all forum members.

Many thanks to both of you for your interest in this subject. I look forward to seeing the posted images soon.

Regards,
GGOSS

 

[jraef]

GGOSS,
I too once embarked on a similar mission, for the purpose of training field techs on how to recognize failures. My conclusions, after talking to several SCR suppliers, was that there really is not much that you can tell directly from looking at a failed SCR. There are a few generalities, but everyone was of the same basic opinion: that you can, at best, tell what DIDN'T cause it, but not necessarily what started it. Ulitmately, almost all failures will look like a current failure, because once something else goes wrong, the final event that takes it out will be a di/dt burn.

The three generalities I could garner were:
1) If the damage is only touching the outside edge of the silicon, it is a voltage burn / spike damage. Excessive dv/dt causes the silicon to start conducting from the outside, away from the gate area and does not evenly ignite the SCR.

2) If the damage is at or near the gate, it is an insufficient gate energy failure, because the silicon didn't "light" fully so all the current attempted to pass through the area closest to the gate.

3) If the damage is between the gate and the edge, but not touching the edge, it is most likely a di/dt failure, probably a short in the load that happened when the SCR was already conducting.

After hearing these rules I disected several failed hockey pucks and came to the earlier conclusion: that you can't tell much since the whole thing looked failed. I did see one with a burn on the outer edge, and the customer had suspected lightning, so it did corraborate the theory a little.

I had some poor quality photos at the time, but I can't locate them any more. This was in the days before scanners and hard drives that store everything now. If I find them, I'll scan and email them to electricpete as the "keeper of the flame" (pictures).
Quando Omni Flunkus Moritati

 

[GGOSS]

Much appreciated jraef,

I currently have in my possesion about 4 or 5 failed SCR's collected over several years for which the fault cause is quite clear ie not what I what consider to be catestrophic or unrecognisable damage.

Although I could quite comfortably write a 'fault report' for each of these devices I have nothing to collaborate my thoughts, hence my problem and need for documentation produced by others.

I'll have a chat to the boys at Semikron and Eupec tommorrow. They may have pics of failed SCR's on file and/or they may wish to comment on the cause of failure of the devices in my possession. This comnbined with the efforts of Electricpete, nukeman in posting other images and comments by other forum members including yourself, should provide all the information needed to bring this to a close.

Thanks & regards,
GGOSS

 

[Marke]

GGOSS

Email a copy of photos to me and I will add my two cents worth. It can be very difficult to tell too much from pictures alone, partially due to the destruction of the die, and partially due to the limited information that you can see on pictures. Photographing silver objects is not easy.
Unfortunately, I no longer have access to failed devices, otherwise I would provide some examples myself.

best regards,
Mark Empson

http://www.lmphotonics.com

 

[GGOSS]

Hello Marke,

Many thanks for your offer also, however I do not have the photographic equipment, patience or skill to take the 'quality' shots you require.

As I know you enjoy photography and have all the necessary the 'gadgets', I would be more than happy to send you the devices, on the proviso of-course that you return them once you have finished.

Please feel free to email me your postal or residential address and I'll get it under way.

Regards,
GGOSS

 

[Marke]

For all those interested, I have posted the Westinghouse paper here

http://www.lmphotonics.com/forum/files.php?action=category&id=4

Best regards Mark Empson

http://www.lmphotonics.com

 

[jbartos]

Suggestion: Visit

http://www.barthelectronics.com/tp_tlptesting.pdf

etc. for more info including pictures