LM4040 vs TL431 for vRef over wide voltage range

[ACharnley]

Hi,

I have a 2.5v vRef requirement and at first I looked to improve over a 2.5v zener using an LM4040-2.5. I then realised it has the same issue as a low voltage zener whereby a minimum current is required - and when the voltage rises this can become significant.

Which had me wondering if a LM4040 is nothing more than a trimmed zener?

Which brings me to the TL431. It requires 10% of the LM4040 current and is cheaper to purchase. The ATL431 uses 1% of the LM4040.

Am I missing a trick? Other than the TL431 only goes down to 0.5% accuracy vs 0.1% for the LM4040 I can't see any benefit?

Cheers!

Andrew

 

[dknguyen]

Which had me wondering if a LM4040 is nothing more than a trimmed zener?

If it is a zener, but it then it is more than just a trimmed zener. The construction would be different to get better noise and stability. So-called "buried zener". Sometimes also other stuff inside like temperature compensation.

The TI datasheet says it is a bandgap reference with compensation From general literature, it seems to be a more accurate than the typical bandgap reference but that could be due to the compensation.

Read and be enlightened:

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&ved=2ahUKEwi9womf8YPdAhVD7VQKHfZwBhYQFjABegQIBxAC&url=https://www.ti.com/lit/an/slyt183/slyt183.pdf&usg=AOvVaw1qIwyEHPCj2qNoYs0EP5ev

 

[ACharnley]

So LM4040 and similar for very high accuracy, TL431 / ATL431 for higher voltages? Anything better?

 

[dknguyen]

So LM4040 and similar for very high accuracy, TL431 / ATL431 for higher voltages? Anything better?

Accuracy isn't the only thing to look for in voltage references. The noise and stability are really important too.

The TL431 is also adjustable, in case you did not notice.

 

[ACharnley]

Ah I've been looking at the wrong figure, the TL431 has a min operating current of 1mA and the LM4040 around 70uA, so in fact it'll work just fine.

 

[tomizett]

It sounds like you're just driving from the supply via a resistor.
For better performance across supply voltage, you could consider driving your reference from a current source.

 

[Nigel Goodwin]

It sounds like you're just driving from the supply via a resistor.
For better performance across supply voltage, you could consider driving your reference from a current source.

Does it make any difference on the 'super zener' IC's that are under discussion?.

 

[tomizett]

If I understand correctly, they still have some dependence on current - but I was mainly thinking of reducing dissipation, which would depend on V, rather than V^2.

 

[ACharnley]

So are you proposing to limit the current with an LDO first?

It's all good anyhow as I read the LM4040 datasheet wrong - it has a very low cathode current drain compared to an equivalent zener.

 

[tomizett]

Well you certainly could use a voltage regulator plus resistor to set the current, but hat I had in mind was something like this:

(just a picture from the net... obviously you'd put the zener/shunt reference where the LEDs are). I believe it's quite a common technique for getting the best performance from shunt regulators, and particularly from zeners.

On the other hand, I may have misinterpreted your problem. I'd assumed you where saying that with a resistor sized to meet the regulator's minimum current requirement at the minimum supply voltage, you ended up with excessive current at the maximum supply voltage. "Excessive" could mean several things; exceeding the regulator's maximum current, drawing too much from your supply (maybe it's a battery, etc) or dissipating an uncomfortable amount of power in the resistor. In that case, this kind of circuit helps as the current is more-or-less independent of the supply voltage.

Sorry I couldn't elaborate more yesterday - I was a bit pressed for time.

 

[ACharnley]

That was my problem but I'd read the datasheet incorrectly. Indeed the LM4040-2.5 needs minimal cathode current to operate so 4-30V with a 40k resistor is possible. So all sorted!