# Reference Voltage of 1.71875V design challenge

#### artmaster547

Joined Jan 6, 2016
409
Hi all
Can anyone help me design a circuit that will give out a constant voltage of specifically 0.78125V, if anyone can help with this would be greatly appreciated, I have used TL431's before for creating a reference but always get stuck on how to select resistors any help would be greatly appreciated.

Kind Regards

Art

Last edited:

#### Alec_t

Joined Sep 17, 2013
12,001
Do you really need a reference voltage that is accurate to ~ 0.001% ? What is it for? How much are you prepared to pay for resistors of that tolerance?

#### artmaster547

Joined Jan 6, 2016
409
Do you really need a reference voltage that is accurate to ~ 0.001% ? What is it for? How much are you prepared to pay for resistors of that tolerance?
on second thoughts tolerance isn't so much of an issue now if you can get as close as you can with 0.1% that would be fine even 1%

#### kubeek

Joined Sep 20, 2005
5,751
You can for example use TL431 to get 2.5V and then divide that with resistors or a trim pot.

#### dl324

Joined Mar 30, 2015
12,792
That's still only 7.8mV. What is the application that requires such accuracy?

#### artmaster547

Joined Jan 6, 2016
409
That's still only 7.8mV. What is the application that requires such accuracy?
hey just realised I made a big typo I meant 1.71875 and I also need to make one for 3.28125, the application space I had in mind was to create an over current signal. These voltages are basically used as reference voltages that will be used with a comparator the application was to monitor current.

#### Alec_t

Joined Sep 17, 2013
12,001
If you use a TL431 , note that there is a 2% tolerance to its internal reference voltage, and that voltage is also subject to a few mV drift with temperature over the rated range. So even with 0.1% resistors you will need to do some tweaking to get your target voltages. That will require your meter to be equally accurate.

#### WBahn

Joined Mar 31, 2012
26,398
Why does an overcurrent limit need to be accurate to less than one part in part in a million? The fact that your values end in '875' and '125' indicate that these are coming from fractions. In fact 1.71875 is 55/32. That implies that you only really need a limit that is good to about one part in one hundred.

Consider what you would be asking for if you needed a limit of 4/3. Would you really be asking for a reference that is good to an infinite level of accuracy since, after all, 1.3333333333333333333333333 still isn't exactly 4/3?

#### EM Fields

Joined Jun 8, 2016
583
hey just realised I made a big typo I meant 1.71875 and I also need to make one for 3.28125, the application space I had in mind was to create an over current signal. These voltages are basically used as reference voltages that will be used with a comparator the application was to monitor current.
Rather than using all of our time and effort to work out solutions for unrealistic problems, it would be much better if you'd just describe your application and let us work out a solution and then feed you back the details.

Last edited:

#### artmaster547

Joined Jan 6, 2016
409
Rather than using all of our time and effort to work out solutions for unrealistic problems, it would be much better if you'd just describe your application and let us work out a solution and then feed you back the details.
Hi yes sorry basically I am using the following current sensor:

http://uk.farnell.com/lem/hais-400-...meResp=All&searchView=table&iscrfnonsku=false

Datasheet:
http://www.farnell.com/datasheets/51649.pdf?_ga=1.141388658.194281532.1473847905

I will be measuring currents in the range of 500A to -500A, and I want the output of this sensor to go to a microcontroller (ADC). The error I made before is that I wanted to override the Reference Voltage of 2.5V. Because I was concerned the voltage would go over 3v3, over the uC limit.

In addition to this I am using two compartors, the first compartor notifies me if the current is at 500A and the second notifies me if the current is at -500A, now I also need reference voltages for these to compare the output voltage of the current sensor any guidance would be greatly appreciated.

Thanks

Art

#### MrChips

Joined Oct 2, 2009
23,233
Get a Weston cell. It delivers a standard voltage of 1.018638V.

#### artmaster547

Joined Jan 6, 2016
409
Get a Weston cell. It delivers a standard voltage of 1.018638V.
errm was hoping that I could generate reference voltages via electronics components rather than another cell, but do appreciate the suggestion.

#### Alec_t

Joined Sep 17, 2013
12,001
Because I was concerned the voltage would go over 3v3, over the uC limit.
A Schottky diode could be used to clamp the signal from the sensor to the 3.3V rail to prevent that.
Are you now intending to use the 2.5V internal reference of the sensor, without over-ride?
Sensor reference and output tolerances are ~1%, so presumably any other references you intend using could be derived from the internal reference and should have a similar tolerance.

Last edited:

#### Sensacell

Joined Jun 19, 2012
2,782
A reference with an adjustable voltage output is not hard to design, you can tweak it to anything you want.
The voltage accuracy then is only as good as the meter reading it.

But... it wont stay there, it will have drift, and it will also change depending on load and input voltage.

#### Kermit2

Joined Feb 5, 2010
4,162

#### Dodgydave

Joined Jun 22, 2012
9,938
hey just realised I made a big typo I meant 1.71875 and I also need to make one for 3.28125, the application space I had in mind was to create an over current signal. These voltages are basically used as reference voltages that will be used with a comparator the application was to monitor current.
Use 1K and 3.2K as the resistors for theTL431.

just use 1% or better resistors, and to drop it down to 1.71V use resistor dividers.

Last edited:

#### crutschow

Joined Mar 14, 2008
26,997
Sorry, this is not enough of a design challenge for me.

#### wayneh

Joined Sep 9, 2010
17,152
Is this related to the current sensor project you asked about in another thread? All you need for that is a window comparator circuit.

I don't believe you need the accuracy implied by your posts in this thread. I'm skeptical that even 1% tolerance is needed, but there's no way you need ppm accuracy on a 500A circuit.

Why don't you back up and tell us the overall goals first. Use the experience of the people here to help you formulate and execute a strategy.

#### KeepItSimpleStupid

Joined Mar 4, 2014
4,643
An OP-amp by Linear Technology, particularly their Over The Top series should be able to handle overloads nicely. You do have to know th absolute maximum Voltage to generally select the current limiting resistors. Clamping can also be employed. So, this could buffer your signal.