Measurement accuracy

[kd]

Hi all,

I was wondering if a specifications says that it must have leakage current
less than 0.1mA...and I measured on my meter 108mA (assuming that everything
is calibrated and accurate...)

Does that mean I can pass this measurement? I'm arguing that even though the
absolute value is higher than the specification, it's an interpretation of
the specification's accuracy to 1 decimal point only?

Your thoughts would be appreciated.

Thanks

 

[kd]

Sorry let me re-phrase that...

I measured 108uA

(ie 108 microamperes....)






"kd" <nospam@damn.it> wrote in message
news:41f9a11d$1@duster.adelaide.on.net...

Hi all,

I was wondering if a specifications says that it must have leakage current
less than 0.1mA...and I measured on my meter 108mA (assuming that
everything is calibrated and accurate...)

Does that mean I can pass this measurement? I'm arguing that even though
the absolute value is higher than the specification, it's an
interpretation of the specification's accuracy to 1 decimal point only?

Your thoughts would be appreciated.

Thanks

 

[Adrian Jansen]

kd wrote:

Sorry let me re-phrase that...

I measured 108uA

(ie 108 microamperes....)






"kd" <nospam@damn.it> wrote in message
news:41f9a11d$1@duster.adelaide.on.net...

Hi all,

I was wondering if a specifications says that it must have leakage current
less than 0.1mA...and I measured on my meter 108mA (assuming that
everything is calibrated and accurate...)

Does that mean I can pass this measurement? I'm arguing that even though
the absolute value is higher than the specification, it's an
interpretation of the specification's accuracy to 1 decimal point only?

Your thoughts would be appreciated.

Thanks






Depends who you are arguing with, your supplier or your customer...

--
Regards,

Adrian Jansen adrianjansen at internode dot on dot net
Design Engineer J & K Micro Systems
Microcomputer solutions for industrial control
Note reply address is invalid, convert address above to machine form.

 

[Phil Allison]

"kd"

I was wondering if a specifications says that it must have leakage current
less than 0.1mA...and I measured on my meter 108uA (assuming that
everything is calibrated and accurate...)

Does that mean I can pass this measurement? I'm arguing that even though
the absolute value is higher than the specification, it's an
interpretation of the specification's accuracy to 1 decimal point only?

** The " less them 0.1 mA " also implies less than 100 uA.

BTW

If this is electrical leakage to ground from a SMPS or similar - you need
to use a low pass filter in the AC path.






............. Phil

 

[David L. Jones]

kd wrote:

Sorry let me re-phrase that...

I measured 108uA

(ie 108 microamperes....)
"kd" <nospam@damn.it> wrote in message
news:41f9a11d$1@duster.adelaide.on.net...
Hi all,

I was wondering if a specifications says that it must have leakage
current
less than 0.1mA...and I measured on my meter 108mA (assuming that
everything is calibrated and accurate...)

Does that mean I can pass this measurement? I'm arguing that even
though
the absolute value is higher than the specification, it's an
interpretation of the specification's accuracy to 1 decimal point
only?

Your thoughts would be appreciated.

Thanks

What's your application?
Is the spec critical or a typical figure?
What is your meters claimed accuracy on that range?
Has inserting your meter into the circuit upset your measurement
somehow?
Will you get your butt kicked if it's out?
What are your quality requirements?
What do other units typically measure?
The list goes on...

Leakage currents typically fall well below an upper spec though, as you
need plenty of margin, so your reading might be an indication something
is wrong.

Dave

 

[Mike Harding]

On Fri, 28 Jan 2005 12:49:07 +1030, "kd" <nospam@damn.it> wrote:

I was wondering if a specifications says that it must have leakage current
less than 0.1mA...and I measured on my meter 108mA (assuming that everything
is calibrated and accurate...)

Does that mean I can pass this measurement? I'm arguing that even though the
absolute value is higher than the specification, it's an interpretation of
the specification's accuracy to 1 decimal point only?

Error noted - you meant 108uA.

I don't see how you can, rounded it becomes 0.11mA, and it does
say "less" than.

Is it _really_ going to be that hard to loose 10uA from the leakage?
Have you tried other samples of the same product?

Mike Harding

 

[John Crighton]

On Fri, 28 Jan 2005 12:50:31 +1030, "kd" <nospam@damn.it> wrote:

"kd" <nospam@damn.it> wrote in message
news:41f9a11d$1@duster.adelaide.on.net...
Hi all,

I was wondering if a specifications says that it must have leakage current
less than 0.1mA...and I measured on my meter 108mA (assuming that
everything is calibrated and accurate...)

Does that mean I can pass this measurement? I'm arguing that even though
the absolute value is higher than the specification, it's an
interpretation of the specification's accuracy to 1 decimal point only?

Your thoughts would be appreciated.

Thanks



Sorry let me re-phrase that...

I measured 108uA

(ie 108 microamperes....)

I can see what you are on about, but to me "less than"
means exactly what it says.

100 uA is a fail

108 uA is greater than 100uA so it is definitely a fail.

99.9 uA is a border line pass. but is so close to failing
I would not be happy to my name on the test sheet.

What are you checking anyway?

Regards,
John Crighton
Hornsby

 

[kd]

"John Crighton" <john_c@tpg.com.au> wrote in message
news:41f9a4be.24656684@News.individual.net...

What are you checking anyway?

We have a product (Li-Ion battery charger) that went through an electrical
approval type testing (IEC 60601.1) and the test house (NATA approved) said
the ENCLOSURE LEAKAGE CURRENT was measured at 108uA.

The standards (AS3200.1.0 which is more or less equivalent to 60601.1) has a
table that refers to the maximum leakage current to be "0.1mA"

The test house gave us a report and a certificate saying that is PASSES that
particular test (clause 19.4.g)

Our client did a audit/review of the report and noted that 108uA does not
comply with the <0.1mA requirements.

Before I approach the test house, I want to know why they passed it? Is it
because they interpreted that "0.1mA" is not the same as "0.100mA" and thus
108uA is a PASS since the requirements only asks for accuracy in "mA" to 1
decimal point??

However, having said that, I do note that it is VERY BORDER LINE and perhaps
a small redesign is needed to be well below that limit.

 

[Mike Harding]

On Fri, 28 Jan 2005 16:14:42 +1030, "kd" <nospam@damn.it> wrote:

Before I approach the test house, I want to know why they passed it? Is it
because they interpreted that "0.1mA" is not the same as "0.100mA" and thus
108uA is a PASS since the requirements only asks for accuracy in "mA" to 1
decimal point??

Just ask them.

Mike Harding

 

[David, not to be confused]

"Mike Harding" <mike_harding@nixspam.fastmail.fm> wrote in message
news:v4mjv01t3vnasjtegetosd0ih8i09ufgcu@4ax.com...

On Fri, 28 Jan 2005 16:14:42 +1030, "kd" <nospam@damn.it> wrote:
Before I approach the test house, I want to know why they passed it? Is it
because they interpreted that "0.1mA" is not the same as "0.100mA" and thus
108uA is a PASS since the requirements only asks for accuracy in "mA" to 1
decimal point??

Just ask them.

Mike Harding

A specification that states a limit does not have a tolerance applied. So 0.1
mA means no more than 0.1 mA to the n'th decimal place. If the spec said "the
leakage shall be 0.1 mA" then you could measure it and apply a tolerance using
an accepted method of practice. But if it says "not exceding 0.1 mA" then
thats definitive - not one poofteenth over 0.1 is allowed.

Before asking for more info you need to think about a few relevant points:

1. Is the sample you measured the exact same sample they measured?
2. Is it the same sample the electro-medical lab measured?
3. The sensitivity of the meters used.
4. The 'measurement uncertainties' of the test lab, the electro-med lab and
you.
5. Did you all use the same test method?
6. The leakage source was loaded with 1K Ohms as per Clause 19.4 (e) 1) of the
Standard?
7. The 'measuring device' of Figure 15 or an equivalent method was employed?

Points 5, 6 & 7 are critical.

If you did not load the leakage source with 1K Ohm or use the Fig 15 device the
current may (likely will) be higher than from not loading the source. You
can't just put a meter in series with the device and the reference point, there
is a specific test method to be followed. For reference, this is a very small
part of clause 19.4:

19.4
e) Measuring device (MD)
1) The measuring device shall load the source of LEAKAGE CURRENT or PATIENT
AUXILIARY
CURRENT with a resistive impedance of approximately 1 000 ? for d.c. and a.c.
and for
composite waveforms with frequencies up to and including 1 MHz.

Ask the test lab and the electro-med lab for their measurement uncertainty, it
is relevant to know these figures for this kind of dispute. If the test lab has
an uncertainty of, say, 100 nA then their test result means it passed by 100 nA
or more (else they should say it had a result lying within the range of
measurement uncertainty or it failed).

Have fun

 

[kd]

"David, not to be confused with the other Davids." <dave@mail--swap.bpa.nu>
wrote in message news:n6nKd.138140$K7.56728@news-server.bigpond.net.au...


Thanks for your reply so far guys... been very helpful.

1. Is the sample you measured the exact same sample they measured?

NO... random sample from the batch

2. Is it the same sample the electro-medical lab measured?

NO... I also realised the standards requires the EUT to be put under
specific environmental conditions such as humidity, temperature, etc... so
again, different lab conditions would probably yield different results...

3. The sensitivity of the meters used.

+/- 3%

4. The 'measurement uncertainties' of the test lab, the electro-med lab
and
you.

Not given...

5. Did you all use the same test method?
6. The leakage source was loaded with 1K Ohms as per Clause 19.4 (e) 1) of
the
Standard?
7. The 'measuring device' of Figure 15 or an equivalent method was
employed?

I would assume that the test house being NATA certified would have followed
the correct procedures and protocols...

Have fun

Definitely will try to have some fun...

 

[David, not to be confused]

I would assume that the test house being NATA certified would have followed
the correct procedures and protocols...

A reasonably safe assumption, but not always so.

Did you get the actual measurement value of the leakage current in the test
report?

To my mind, if you wish to peform some kind of QA then it is essential you
establish a baseline for the data, and that would include finding out the
measurement values obtained by the test house. Although without the same
sample(s) that measurement data is pretty much useless as you can't work out
the error between measurements done by the different parties. Nevertheless it
is still handy to know.

It sounds like the device passed by a very small margin - possibly too small a
margin to consider suitable for production. Ask the test house for their
measurement uncertainty for this test (which will be different to their
measurement uncertainty for plain old voltage or current as there are several
'components' making up the total uncertainty). Then establish what variance
your production devices have. If you are aiming for 100% of production to pass
100% of the time then you need your highest leakage values to be less than the
0.1 mA specified taking into account measurement uncertainty (Uc). Perhaps
this explains it better the above:

"100% QA limit = 0.1 mA - Uc" So if Uc is 0.01 mA then you have to make sure
all devices have leakage of less than 0.09 mA.

Now this would be OK if the test house did all your measurements - but they're
not. So you have to replace the test house Uc with your own value of Uc. Your
value of Uc will probably be larger than the test house, so by reducing the QA
limit by your value of Uc will increase the margin of compliance. As long as
your devices all lie below your QA limit then you will always comply with the
specification.

Calculating Uc is beyond a news group post, however NIST (USA) have an
excellent guide here:

http://physics.nist.gov/Pubs/guidelines/contents.html

Now you can really get into the fun stuff... not !!