Shunt ? (was Re: Ideas for inexpensive True-RMS Metering?) [Design]

Prev: NASA to Provide Missile Technology to Arabs...
Next: Buy Designer Handbags

From: Fred Bartoli on 7 Jul 2010 03:20

John Larkin a �crit :
> On Tue, 6 Jul 2010 19:06:37 -0500, "Tim Williams"
> <tmoranwms(a)charter.net> wrote:
>
>> "whit3rd" <whit3rd(a)gmail.com> wrote in message
news:c6d94b54-aa3a-4eb1-8810-4417e6f366cd(a)32g2000vbi.googlegroups.com...
>>> The saturation of the core makes the coupling of primary and
>>> secondary go away; only the primary side gets the fault current,
>>> the secondary current is no longer proportional (unless you choose
>>> a really odd winding scheme that stays highly flux-coupled when
>>> the core is removed).
>> Yes, which is why the energy delivered is proportional to flux as
well. It only delivers power until the transformer stops transformering.
>>
>> Say you get 1A fault current from a CT (i.e., secondary referred),
which throws your circuit into overload, so the voltage on the windings
jumps to 5V (clamped by a perfect 5V supply, assuming ideal clamp
diodes). The delivered power is evidently 1A * 5V = 5W, going into your
supply. If the winding has a saturation flux of 1mWb, this fault
current will flow for 1mWb / 5V = 0.2ms. The energy is 5W * 0.2ms =
1mJ, or 1A * 1mWb.
>>
>> Increase the fault to 10A. The winding is clamped at 5V, so 50W is
delivered, and the fault again lasts for 0.2ms, because the flux is
1mWb. The energy is 10mJ.
>>
>> Increase the fault to 1kA. Now you get 5kW and 1J, beyond the
capacity of a 1.5KE6.
>>
>> Strike it with a bolt of lightning. 100kA gives 500kW peak and
100J, assuming the transformer doesn't fail first; if it has an internal
resistance of 0.01 ohm, it will drop 1kV, probably breaking down the
meager insulation in a CT.
>>
>> Tim
>
> The usual practise in electronic metering is to have the CT secondary
> drive a low-resistance wirewound or manganin strip shunt. The shunt
> resistance is considerably less than the winding resistance.
> Outrageous CT overloads don't damage the shunt... most of the power
> dissipation is in the winding. The signal conditioning opamps or
> whatever are protected by high value resistances between them and the
> shunt.
>

I'm designing an energy metering ammeter and am looking after a
0.1/0.2R, preferably SMT, shunt.

On the average all of them will sum up to half a billion euro energy, so
it has to be accurate :-)

It'll work in some harsh environment and must :
* work @ 85�C Tamb, (100�C PCB temp)
* be low tempco (preferably lower than 20ppm/K)
* real low aging for less than yearly calibration
* preferably high initial accuracy to hopefully bypass one calibration step

None of the usual suspects fit the bill.

Any manufacturer / part series to suggest?

--
Thanks,
Fred.

--
Thanks,
Fred.

From: Grant on 7 Jul 2010 04:54

On Wed, 07 Jul 2010 09:20:18 +0200, Fred Bartoli <myname_with_a_dot_inbetween(a)free.fr> wrote:

....
>I'm designing an energy metering ammeter and am looking after a
>0.1/0.2R, preferably SMT, shunt.
>
>On the average all of them will sum up to half a billion euro energy, so
>it has to be accurate :-)
>
>It'll work in some harsh environment and must :
>* work @ 85°C Tamb, (100°C PCB temp)
>* be low tempco (preferably lower than 20ppm/K)
>* real low aging for less than yearly calibration
>* preferably high initial accuracy to hopefully bypass one calibration step

1%, 0.1% better?
>
>
>None of the usual suspects fit the bill.
>
>Any manufacturer / part series to suggest?

Only thing I've found for accurate shunts better than 1% is lots of 0.1%
resistors in parallel.

But I use axial lead or SMD big enough to hand solder. .1, .2R is high
value for a shunt anyway?

0.1% 250mW 15ppm start at 10R, 5 or 10 in parallel too silly?

Grant.

From: John Larkin on 7 Jul 2010 10:04

On Wed, 07 Jul 2010 18:54:39 +1000, Grant <omg(a)grrr.id.au> wrote:

>On Wed, 07 Jul 2010 09:20:18 +0200, Fred Bartoli <myname_with_a_dot_inbetween(a)free.fr> wrote:
>
>...
>>I'm designing an energy metering ammeter and am looking after a
>>0.1/0.2R, preferably SMT, shunt.
>>
>>On the average all of them will sum up to half a billion euro energy, so
>>it has to be accurate :-)
>>
>>It'll work in some harsh environment and must :
>>* work @ 85�C Tamb, (100�C PCB temp)
>>* be low tempco (preferably lower than 20ppm/K)
>>* real low aging for less than yearly calibration
>>* preferably high initial accuracy to hopefully bypass one calibration step
>
>1%, 0.1% better?
>>
>>
>>None of the usual suspects fit the bill.
>>
>>Any manufacturer / part series to suggest?
>
>Only thing I've found for accurate shunts better than 1% is lots of 0.1%
>resistors in parallel.
>
>But I use axial lead or SMD big enough to hand solder. .1, .2R is high
>value for a shunt anyway?
>
>0.1% 250mW 15ppm start at 10R, 5 or 10 in parallel too silly?
>
>Grant.

I did some meters using this:

MFR2 DIGIKEY KRL32C.005CT-ND
MFR3 MOUSER 66-ULRB1R005LFSLT

which is a Susumu 0.005 ohms, 1%, 2512 package. There are lots of
similar low-ohm resistors around these days. There are also 4-lead
resistors and manganin shunts, lower TC, but maybe hard to find as
high as 0.1 ohm.

A serious problem with electric metering is magnetic field pickup,
causing crosstalk from adjacent channels or nearby transformers, fans,
whatever. Layout matters a lot, as does pcb copper-foil heat sinking
of small parts like this. We make our own shunts out of sheet manganin
with exotic geometries to minimize hum pickup and eddy-current
effects. But we get a few per cent initial accuracy and software
calibrate around that.

Vishay makes 4-lead laser-trimmed metal-foil TO-3 can power resistors
that meet all your specs, but they aren't cheap, and mag field pickup
could be a problem. They have horrible transient response due to eddy
effects in the can, but that wouldn't matter at 60 Hz.

Energy meters are fun, especially the processing algorithms.

John

From: Capt. Cave Man on 7 Jul 2010 10:39

From: Grant on 8 Jul 2010 08:41

On Wed, 07 Jul 2010 07:39:37 -0700, Capt. Cave Man <ItIsSoEasyACaveManCanDoIt(a)upyers.org> wrote:

>On Wed, 07 Jul 2010 18:54:39 +1000, Grant <omg(a)grrr.id.au> wrote:
>
>>On Wed, 07 Jul 2010 09:20:18 +0200, Fred Bartoli <myname_with_a_dot_inbetween(a)free.fr> wrote:
>>
>>...
>>>I'm designing an energy metering ammeter and am looking after a
>>>0.1/0.2R, preferably SMT, shunt.
>>>
>>>On the average all of them will sum up to half a billion euro energy, so
>>>it has to be accurate :-)
>>>
>>>It'll work in some harsh environment and must :
>>>* work @ 85°C Tamb, (100°C PCB temp)
>>>* be low tempco (preferably lower than 20ppm/K)
>>>* real low aging for less than yearly calibration
>>>* preferably high initial accuracy to hopefully bypass one calibration step
>>
>>1%, 0.1% better?
>>>
>>>
>>>None of the usual suspects fit the bill.
>>>
>>>Any manufacturer / part series to suggest?
>>
>>Only thing I've found for accurate shunts better than 1% is lots of 0.1%
>>resistors in parallel.
>>
>>But I use axial lead or SMD big enough to hand solder. .1, .2R is high
>>value for a shunt anyway?
>>
>>0.1% 250mW 15ppm start at 10R, 5 or 10 in parallel too silly?
>>
>>Grant.
>
> Precision current shunts are readily available at 0.1% ratings.
> They have to be trimmed to that level of precision.
>
> It is also very easy to perform said trimming for accuracy, so why
>would they not make them that accurate?
>
> Answer: They don't.
>
> They, in fact do make precision shunts and using paralleled banks is
>pretty damned silly.

Not on my budget ;) But I'm not designing for production, prototyping.

Grant.

| Next | Last
Pages: 1 2
Prev: NASA to Provide Missile Technology to Arabs...
Next: Buy Designer Handbags