LTZ1000A stack-up

Hi Mehdiaghayi,

I suspect you meant to post this in the "Voltage References" forum. "Reference Circuits" generally refers to Circuits from the Lab and other orderable hardware. Apologies for any confusion - which is definitely understandable given that "reference" can refer to lots of different things.

I have worked with the LTZ1000 in the past, as well as the LT399 style devices that include a built-in heater controller. Stacking this many LTZ1000s will be a challenge - you will need to provide a floating power supply to each circuit. Also note that the initial tolerance on the LTZ1000 is quite large - 7.0V to 7.5V. This means that a stack of 6 may have a voltage between 42 and 45V. Is this okay?

You may be better off using precision, low-drift resistors to multiply the LTZ1000 output by 6 - the engineers in the Voltage References forum should be able to provide better advice.

Also - another device that may be useful, at least for ideas, is the LTC1043. This part can act as a precision multiply by 2 (and higher, when stacked.). Take a look at the datasheet and Linear Technology Application Note 3. This part only operates up to 18V, but you MIGHT be able to use a similar technique using higher voltage devices. In the distant past, mechanical choppers were used for this sort of thing - but of course they are not silent and subject to wear over time. Depends what your exact requirements are.

-Mark

Hello Mark,

First of all, I'm sorry for this inconvenience and thank you for finding a better forum for this question.
also thank you so much for your detail answer. really I didn't find any recommendations on shunt mode operation for LTZ1000A so I asked this question first here.
I'm not standing over a certain voltage like 42V. Actually when I can find a low noise reference combination as high as 21V it's also great for my design. The voltage that I'm talking about will be choped to +21v and -21v by a simple chopper. It must provide only the low noise supply voltage for a current source circuit. Therefore, the voltages as 21, 20, 22, 19 volts are considered similar to the current source.

as I described earlier because this voltage that I called it V_ref will come directly to the circuit, a low noise profile would be considered preferrable.  

Is it possible to multiple the output voltage of LTZ1000 without any change to Its noise qualification? 

regarding the LTC6655LN That's also considered good for my design. However, That's also not designed for shunt mode operation.

With the best regards,
mehdiaghayi

Hi Mehdiaghayi, I'll have to defer to the experts on this forum for the detailed answers, but a couple of observations and hints:

The LTZ1000 is fundamentally a shunt-mode device, but you MUST use one of the recommended application circuits from the datasheet -  with no modifications. In contrast, you could simply string multiple ADR5045-style devices together to generate a higher voltage. (I'm not suggesting the ADR5045 for your application, I'm just using it as an example.)

Note that "low" noise will need to be quantified - you'll need to provide exact requirements. How many nanovolts per root Hz, or how many microvolts peak-to-peak, and over what frequency band?

ANY circuit that multiplies or chops / inverts a voltage will necessarily add noise - you just need to make sure that it's acceptable in your application.

One technique for reducing noise is to parallel multiple devices, as shown on page 26 of the LTC6655 datasheet.

-Mark

Hi Mehdiaghayi, I'll have to defer to the experts on this forum for the detailed answers, but a couple of observations and hints:

The LTZ1000 is fundamentally a shunt-mode device, but you MUST use one of the recommended application circuits from the datasheet -  with no modifications. In contrast, you could simply string multiple ADR5045-style devices together to generate a higher voltage. (I'm not suggesting the ADR5045 for your application, I'm just using it as an example.)

Note that "low" noise will need to be quantified - you'll need to provide exact requirements. How many nanovolts per root Hz, or how many microvolts peak-to-peak, and over what frequency band?

ANY circuit that multiplies or chops / inverts a voltage will necessarily add noise - you just need to make sure that it's acceptable in your application.

One technique for reducing noise is to parallel multiple devices, as shown on page 26 of the LTC6655 datasheet.

-Mark

Hi Mark, thanks for moving this to the voltage references forum.

Hi mehdiaghayi  Mark is right, we must use the recommended applications from the datasheet. Some suitable parts for this application are ADR5045 that Mark mentioned and ADR550. These parts are proven to work for stacking voltage. Note that the noise/ accuracy characteristics of the stacked parts will sum up.

Hello GioBalgemino,

I solved the problem with stacking multiple LT6657 reference voltages.

However, I believe that I've destroyed the noise and thermal drift specifications for LT6657 by adding some opamps and specially the ADG5236 switch. 

thank you for your time and considerations.

Switch +-20V.asc

Regards