Low MOSFET IDSS current

On Fri, 15 Dec 2017 18:24:02 GMT, Steve Wilson <no@spam.com> wrote:

>Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:
>
>> On 12/15/2017 05:34 AM, Steve Wilson wrote:
>
>>> FIT is "Failure in Time" and is a reliability term. It is defined as a 
>>> failure rate of 1 per billion hours. It obviously applies after the parts 
>>> have already been accepted and placed in service. 
> 
>> And is used based on the now completely discredited reliability
>> estimation method of MIL-HDBK-217.  The idea that the failure rate of a
>> system can be calculated as the sum of tabulated, invariant failure
>> rates of the individual parts is no better than a random number generator.
>
>John used the term, not me.
>
>It has nothing to do with AQL.
>
>Let's talk about AQL instead of FIT.
> 
>> Cheers
> 
>> Phil Hobbs
> 

OK, let's. How do you manage incoming parts quality?

If you get a reel of 3000 0603 capacitors, what do you do?



-- 

John Larkin         Highland Technology, Inc
picosecond timing   precision measurement 

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

On 12/15/2017 01:24 PM, Steve Wilson wrote:
> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:
> 
>> On 12/15/2017 05:34 AM, Steve Wilson wrote:
> 
>>> FIT is "Failure in Time" and is a reliability term. It is defined as a 
>>> failure rate of 1 per billion hours. It obviously applies after the parts 
>>> have already been accepted and placed in service. 
>  
>> And is used based on the now completely discredited reliability
>> estimation method of MIL-HDBK-217.  The idea that the failure rate of a
>> system can be calculated as the sum of tabulated, invariant failure
>> rates of the individual parts is no better than a random number generator.
> 
> John used the term, not me.
> 
> It has nothing to do with AQL.
> 
> Let's talk about AQL instead of FIT.

But FIT-bashing is way more fun. ;)

I've had one or two more-or-less bad batches of stuff, e.g. a reel of
JFETs that were 2 dB noisier than the ones another reel, but in general
I don't buy enough parts to test to 0.1% AQL even if I had the inclination.

And lots of parts are likely to be damaged during test--back in the day,
JAN TX screened transistors were notoriously less reliable than the COTS
ones due to all the extra handling.

IME parts die due to maltreatment, not quality problems.  Now if I were
making car air bag controllers, I'd worry more about that sort of stuff,
but for us, if it passes final test, it's OK.

Cheers

Phil Hobbs






-- 
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
https://hobbs-eo.com

John Larkin <jjlarkin@highland_snip_technology.com> wrote:

> Yes. Incoming inspection is not reasonable (parts on reels)

What's wrong with AQL on reels?

Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

> On 12/15/2017 01:24 PM, Steve Wilson wrote:
> And lots of parts are likely to be damaged during test--back in the day,
> JAN TX screened transistors were notoriously less reliable than the COTS
> ones due to all the extra handling.

Bad handling procedures from long ago, and less reliable designs. This can 
kill parts during assembly. Modern parts and procedures should be a lot more 
reliable.
 
> IME parts die due to maltreatment, not quality problems.  Now if I were
> making car air bag controllers, I'd worry more about that sort of stuff,
> but for us, if it passes final test, it's OK.

We are not talking about failures or design problems. We are talking about 
incoming inspection. 

If it passes final test, how do you know how close it is to failing?

Maybe your quantities are not high enough to warrant AQL, but others may need 
it.
 
> Cheers
 
> Phil Hobbs

John Larkin <jjlarkin@highland_snip_technology.com> wrote:

> OK, let's. How do you manage incoming parts quality?
 
> If you get a reel of 3000 0603 capacitors, what do you do?
 
See 6.2. Test Product for Acceptability: Lot Acceptance Sampling

This section describes how to make decisions on a lot-by-lot basis
whether to accept a lot as likely to meet requirements or reject the
lot as likely to have too many defective units.

This section consists of the following topics.

  What is Acceptance Sampling?
  What kinds of Lot Acceptance Sampling Plans (LASPs) are there?
  How do you Choose a Single Sampling Plan?
      Choosing a Sampling Plan: MIL Standard 105D
      Choosing a Sampling Plan with a given OC Curve
  What is Double Sampling? 
  What is Multiple Sampling?
  What is a Sequential Sampling Plan?
  What is Skip Lot Sampling?

http://www.itl.nist.gov/div898/handbook/pmc/section2/pmc2.htm	

Other excellent references on the web. 
 

On 12/15/2017 02:48 PM, Steve Wilson wrote:
> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:
> 
>> On 12/15/2017 01:24 PM, Steve Wilson wrote:
>> And lots of parts are likely to be damaged during test--back in the day,
>> JAN TX screened transistors were notoriously less reliable than the COTS
>> ones due to all the extra handling.
> 
> Bad handling procedures from long ago, and less reliable designs. This can 
> kill parts during assembly. Modern parts and procedures should be a lot more 
> reliable.

Maybe.  But the sheer amount of fixturing and labour involved in testing
every kind of part that we use would make the whole thing uneconomic.
Parts are cheap, but labour is expensive, and fixtures take up a lot of
space.

>  
>> IME parts die due to maltreatment, not quality problems.  Now if I were
>> making car air bag controllers, I'd worry more about that sort of stuff,
>> but for us, if it passes final test, it's OK.
> 
> We are not talking about failures or design problems. We are talking about 
> incoming inspection. 
> 
> If it passes final test, how do you know how close it is to failing?

How would AQL help me know that?  Reliability vs. quality again.

> 
> Maybe your quantities are not high enough to warrant AQL, but others may need 
> it.

I'm sure that's true.  Joh's talking about low volumes, and I'm talking
about really low volumes, so far.

Cheers

Phil Hobbs


-- 
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
https://hobbs-eo.com

On Fri, 15 Dec 2017 19:56:04 GMT, Steve Wilson <no@spam.com> wrote:

>John Larkin <jjlarkin@highland_snip_technology.com> wrote:
>
>> OK, let's. How do you manage incoming parts quality?
> 
>> If you get a reel of 3000 0603 capacitors, what do you do?
> 
>See 6.2. Test Product for Acceptability: Lot Acceptance Sampling
>
>This section describes how to make decisions on a lot-by-lot basis
>whether to accept a lot as likely to meet requirements or reject the
>lot as likely to have too many defective units.
>
>This section consists of the following topics.
>
>  What is Acceptance Sampling?
>  What kinds of Lot Acceptance Sampling Plans (LASPs) are there?
>  How do you Choose a Single Sampling Plan?
>      Choosing a Sampling Plan: MIL Standard 105D
>      Choosing a Sampling Plan with a given OC Curve
>  What is Double Sampling? 
>  What is Multiple Sampling?
>  What is a Sequential Sampling Plan?
>  What is Skip Lot Sampling?
>
>http://www.itl.nist.gov/div898/handbook/pmc/section2/pmc2.htm	
>
>Other excellent references on the web. 
> 

That's sounding like the quality-control ritual blather.

What exactly do YOU DO when you get that reel of caps?


-- 

John Larkin         Highland Technology, Inc
picosecond timing   precision measurement 

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:

> On 12/15/2017 02:48 PM, Steve Wilson wrote:
>> Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:
 
>>> On 12/15/2017 01:24 PM, Steve Wilson wrote:
>>> And lots of parts are likely to be damaged during test--back in the
>>> day, JAN TX screened transistors were notoriously less reliable than
>>> the COTS ones due to all the extra handling. 
 
>> Bad handling procedures from long ago, and less reliable designs. This
>> can kill parts during assembly. Modern parts and procedures should be a
>> lot more reliable. 
 
> Maybe.  But the sheer amount of fixturing and labour involved in testing
> every kind of part that we use would make the whole thing uneconomic.
> Parts are cheap, but labour is expensive, and fixtures take up a lot of
> space.
 
I've been thinking about that. An automated tester shouldn't be very 
complicated. Separate jigs for diferent size smd's. A little stepper to 
advance the reel to the next component. A fast precision LCR meter to 
measure the components. 0.1% should be good enough for most parts. 

Perhaps a small Teensy or Arduino to log the measurements and step the 
reel. Some of the guys on Hackaday do fantastic work on small mechanisms.

One problem is to find a way to mark bad components. If the test is fast 
enough, maybe test the whole reel. 3600 parts at 100ms per part would only 
take 6 minutes.
  
>> If it passes final test, how do you know how close it is to failing?
 
> How would AQL help me know that?  Reliability vs. quality again.
 
You set the limits for the test. Parts that are outside the limits are 
rejected and not installed.
 
>> Maybe your quantities are not high enough to warrant AQL, but others
>> may need it. 
 
> I'm sure that's true.  Joh's talking about low volumes, and I'm talking
> about really low volumes, so far.

Hopefully that will change.
 
> Cheers
 
> Phil Hobbs
 
 

John Larkin <jjlarkin@highland_snip_technology.com> wrote:

> On Fri, 15 Dec 2017 19:56:04 GMT, Steve Wilson <no@spam.com> wrote:
 
>>John Larkin <jjlarkin@highland_snip_technology.com> wrote:

>>> OK, let's. How do you manage incoming parts quality?
 
>>> If you get a reel of 3000 0603 capacitors, what do you do?
 
>>See 6.2. Test Product for Acceptability: Lot Acceptance Sampling

>>This section describes how to make decisions on a lot-by-lot basis
>>whether to accept a lot as likely to meet requirements or reject the
>>lot as likely to have too many defective units.

>>This section consists of the following topics.

>>  What is Acceptance Sampling?
>>  What kinds of Lot Acceptance Sampling Plans (LASPs) are there?
>>  How do you Choose a Single Sampling Plan?
>>      Choosing a Sampling Plan: MIL Standard 105D
>>      Choosing a Sampling Plan with a given OC Curve
>>  What is Double Sampling? 
>>  What is Multiple Sampling?
>>  What is a Sequential Sampling Plan?
>>  What is Skip Lot Sampling?

>>http://www.itl.nist.gov/div898/handbook/pmc/section2/pmc2.htm     

>>Other excellent references on the web. 

> That's sounding like the quality-control ritual blather.

The math is real. A lot of companies depend on it.
 
> What exactly do YOU DO when you get that reel of caps?
 
That's what I am setting up now. I am leaning towards automated testing and 
logging the results. A simple reel tester shouldn't cost much to make, and 
could be very fast.

I have found if you do not specify the incoming test, companies will ship 
you the dregs that failed AQL at other companies. As an example, I have  
received 16 pin HC7474 through hole parts. They were marked 74HC7474, but I 
don't know of any 16 pin versions. Obviously someone rejected them, so they 
sent them to us.

I was hoping some mfg guys would chime in and tell us their horror stories.

>>> Modern parts and procedures should be a 
>>> lot more reliable. 
  
>> Maybe.  But the sheer amount of fixturing and labour involved in testing 
>> every kind of part that we use would make the whole thing uneconomic. 
>> Parts are cheap, but labour is expensive, and fixtures take up a lot of 
>> space. 
  
>I've been thinking about that. An automated tester shouldn't be very 
>complicated. Separate jigs for diferent size smd's. A little stepper to 
>advance the reel to the next component. A fast precision LCR meter to 
>measure the components. 0.1% should be good enough for most parts. 

>Perhaps a small Teensy or Arduino to log the measurements and step the 
>reel. Some of the guys on Hackaday do fantastic work on small mechanisms.

I'm sure they do, but my shop is about 1100 square feet and 2-1/2 people, and there's only so much mojo to go round. The cost of the equipment is a nit by comparison with labour and (especially) opportunity cost.  

Plus we'd have to put those parts back into reels after testing, and all for (very probably) no benefit. 

No thanks. 

Cheers

Phil Hobbs