[volt-nuts] Matched resistors

[Andreas Jahn]

Hello Randy,

I think the only difference is in oscillator section (and thus power 
consumption)
and of cause the TSSOP-package.
The LTC1043 is easily available from stock e.g. from digikey.
The LTC6943 is more difficult to get.
Within the Keithley 2002 LTC1043 is used.
http://www.eevblog.com/forum/testgear/keithley-2002-8-5-digit-dmm-review-and-teardown/
http://dev.xdevs.com/projects/kei2002/repository/entry/photos/K1/small/K2002_1-2251.jpg

I have added a drift chart with longterm drift data.
Note: the drift is for the whole measurement arrangement.
It consists of 3 7V references (2 LTZ1000A and 1 LM399)
measured via a LTC1043 divider with a 24 bit LTC2400 ADC with
temperature compensated voltage reference AD586LQ.
So most of the drift is related to the AD586LQ reference.
(X-axis is in days, Y-axis in ppm)

I get around 2 ppm drift for the LTZ1000A  over 1 year
which I guess is mostly humidity related
from the ADC printed cirquit board + AD586 reference drift
and usually below 0.25 ppm standard deviation over 1000 hours.  (42 days).
All at unstabilized room temperature.
I guess with resistors you will need ovenized temperature stabilisation 
to achieve this.

with best regards

Andreas

Am 19.07.2014 05:57, schrieb Randy Evans:
> Andreas,
>
> That is good information, I appreciate it.  I have contacted LT application
> support but they have yet to get back to me on my questions except they did
> recommend to use the LTC6943 instead of the LTC1043.  Later generation I
> guess.
>
> I think i am going to try both the LTC6943 and the LT5400 resistor array
> and characterize them.  The LT5400 matching ratio looks pretty good over
> temp (0.2ppm/C) but the absolute resistor change over temp is -10 to +25
> ppm/C, a little larger than I would like for the circuit I am using.
>
> Randy
>
>
> On Thu, Jul 17, 2014 at 10:37 PM, Andreas Jahn <Andreas_-_Jahn at t-online.de>
> wrote:
>
>> Hello Randy,
>>
>> I am using the LTC1043 in 1/2 VIN or 1/3 VIN configuration.
>> A matching of the caps is not necessary.
>> In the 1/2 VIN or 1/3 VIN configuration a matching would give
>> the advantage that the settling time of the cirquit is reduced.
>> But in 2* VIN or inverting configuration a matching gives no advantage.
>> So perhaps it is better to put a 1/2 VIN divider into a feedback loop.
>>
>> The most important point: you will need a low leakage buffer amplifier at
>> the output.
>> The caps should be low leakage foil capacitors. (polypropylene would be
>> best).
>> The ESR is negligible against the switch resistance of around 1000 Ohms
>> And dielectric absorption would also affect only settling time.
>>
>> In 1/2 VIN configuration I am using cheap small mylar capacitors (WIMA
>> MKS02)
>> (isolation time constant is given only with  >1250 sec (3000 sec typ)).
>> Buffer amplifier is a LTC1050.
>> The circuit is very stable over temperature (10 - 40 deg C).
>> The absolute amplification error is usually some ppm lower than exact 2:1
>> value.
>> (depends somewhat on the pinning which is used so I am not shure wether
>> the pins are mixed up regarding the charge compensation)
>>
>> So I dont know wether the ±1 ppm is more a stability figure than a
>> absolute value.
>> Even polypropylene capacitors do not change the amplification error.
>>
>> With best regards
>>
>> Andreas
>>
>> Am 17.07.2014 17:26, schrieb Randy Evans:
>>
>>> Frank,
>>>
>>> The high cost is my concern, although high performance demands high price
>>> typically.  I am trying to double the voltage reference from either an
>>> LM399 or LTZ1000, hence the need for precision matched resistors for a x2
>>> non-inverting amplifier (using a LT1151 precision op amp).  An alternative
>>> I am investigating is using the LTC1043 in a voltage doubling circuit as
>>> shown in Linear Technology app note AN 42, page 6, Figure 16.  It states
>>> that Vout = 2xVin ± 5 ppm.  I am less concerned about the absolute
>>>
>>> accuracy than I am about the long term stability.  I assume that a high
>>> quality capacitor is required (low leakage, low ESR, low dielectric
>>> absorbtion, etc.) but the circuit does not appear to be dependent on the
>>> absolute value of the capacitors.  I'm not sure if the two 1uF caps  need
>>> to be matched.  If they do then that would be a show stopper.
>>>
>>> Does anyone have any experience using the LTC1043 in such a circuit?
>>>
>>> Thanks,
>>>
>>> Randy
>>>
>>>   _______________________________________________
>> volt-nuts mailing list -- volt-nuts at febo.com
>> To unsubscribe, go to https://www.febo.com/cgi-bin/
>> mailman/listinfo/volt-nuts
>> and follow the instructions there.
>>
> _______________________________________________
> volt-nuts mailing list -- volt-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
> and follow the instructions there.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 20140713_ADC13_AGE.PNG
Type: image/png
Size: 73836 bytes
Desc: not available
URL: <http://www.febo.com/pipermail/volt-nuts/attachments/20140719/5f274de5/attachment-0001.png>