Output capacitor on a shunt reference? (ZR431)

[Futterama]

Hello forum,

Should I use some output capacitor (C1) for this circuit, using a ZR431 (IC1) shunt regulator as voltage reference (2.50V) for a PIC AD converter input?

Thanks.

Regards,
Futterama

 

[Nigel Goodwin]

Hello forum,

Should I use some output capacitor (C1) for this circuit, using a ZR431 (IC1) shunt regulator as voltage reference (2.50V) for a PIC AD converter input?

I suggest you read the datasheet - but I didn't use one in my analogue PIC tutorial!.

 

[Futterama]

I suggest you read the datasheet - but I didn't use one in my analogue PIC tutorial!.

Surely I have read the datasheet (the Zetex version) before posting, and it doesn't mention output capacitors, but in some of the test circuits, they use a capacitor, but I'm not testing, I'm using the thing...

The only thing they mention is load capacitance, but I don't know what that is :roll:

 

[Nigel Goodwin]

I shouldn't bother with one! - I didn't, and your application is VERY non-critical.

Have you ever heard of a 'neon relaxation oscillator'? - it looks very similar to fitting a capacitor acoss the TL341!.

 

[Futterama]

...and your application is VERY non-critical.

I wouldn't call an application that keeps my $1000 RC car from running loose with 50mph "VERY non-critical" - the car is expensive, and someone could get hurt if 4kg metal hits them with 50mph :?

Have you ever heard of a 'neon relaxation oscillator'? - it looks very similar to fitting a capacitor acoss the TL341!.

No, never heard of it...

I'm still worried about not using a capacitor. All 3 datasheets I have, shows a capacitor when using the XX431 as shunt regulator, but the capacitor type or value is never mentioned:

 

[Futterama]

Perhaps they don't mention the capacitor value due to the statement on the Zetex website:

**broken link removed**

One of the features of, at least the ZR431, is: "No stabilizing capacitor required".

So I'll just use it without capacitor, and perhaps make room on my PCB for a capacitor if it should be necessary.

 

[audioguru]

In National's datasheet, they show the value of capacitor and output current that cause the LM431 to oscillate.

 

[Futterama]

In National's datasheet, they show the value of capacitor and output current that cause the LM431 to oscillate.

And how could one get around this? Use a 10uF cap then?

 

[Nigel Goodwin]

...and your application is VERY non-critical.

I wouldn't call an application that keeps my $1000 RC car from running loose with 50mph "VERY non-critical" - the car is expensive, and someone could get hurt if 4kg metal hits them with 50mph :?

Isn't it just for checking how well charged the battery is?, I would call that "VERY non-critical".

And how could one get around this? Use a 10uF cap then?

As I've already said, and the graph audioguru posted shows, you don't need a capacitor at all!.

You're probably too young to remember neon oscillators?, but adding a capacitor across the TL341 looks exactly like one - and from audioguru's graph, it looks like they oscillate in exactly the same way?.

 

[Optikon]

In National's datasheet, they show the value of capacitor and output current that cause the LM431 to oscillate.

And how could one get around this? Use a 10uF cap then?

First, Nationals version of LM431 shown is only a region of concern for nationals part. Other companies make this part and they may be more or less immune to osc for a particular value of output C. You'll have to check for specifically the part you are using.

And while having a capacitor doesnt seem to be critical, sometimes it is a good idea to improve the transient response, filter noise etc... as long as you stay away from the region of oscillation, it will only help maintain a stable quality reference voltage.

You probably have much bigger problems to worry about keeping your design safe. Oh yeah, if you dont want to risk it, get a different reference. Plenty of manufacturers make oscillation free references with or without caps.

 

[Russlk]

I have never used a capacitor on a TL431 and never had a problem with noise but if you are worried about noise, use an R-C filter (100 ohms, 100uF).

 

[Futterama]

Isn't it just for checking how well charged the battery is?, I would call that "VERY non-critical".

No Nigel, it's used to monitor the battery voltage during use, and when the voltage drop because of low battey level, the unit needs to activate and brake the car. If the voltage gets too low, or rather the batteries get too depleted, the throttle/brake and steering servo will stop working, and if this happens at WOT (Wide Open Throttle) the car will be completely out of control and dangerous.

 

[Nigel Goodwin]

Isn't it just for checking how well charged the battery is?, I would call that "VERY non-critical".

No Nigel, it's used to monitor the battery voltage during use, and when the voltage drop because of low battey level, the unit needs to activate and brake the car. If the voltage gets too low, or rather the batteries get too depleted, the throttle/brake and steering servo will stop working, and if this happens at WOT (Wide Open Throttle) the car will be completely out of control and dangerous.

Yes, I'm aware of the reasoning behind it, but the level of accuracy required is only very low, 10% accuracy would probably be plenty? - the 0.1% from a 10 bit A2D isn't required. You really don't want to be pushing the battery too low, that's where you're going to hit problems - and rechargable batteries drop down quickly once they start!.

As I've suggested previously, there's no need for a capacitor, and it could well be VERY troublesome if fitted (from the graph posted earlier!).

 

[Futterama]

Perhaps I should go for another IC then, I have a couple of LM4050 from Maxim, and the datasheet says this:

The LM4050/LM4051 do not require external capacitors for frequency stability and are stable for any output capacitance.

 

[Nigel Goodwin]

Perhaps I should go for another IC then, I have a couple of LM4050 from Maxim, and the datasheet says this:

The LM4050/LM4051 do not require external capacitors for frequency stability and are stable for any output capacitance.

You could if you wanted?, but I see no reason to do so? - the 341 works perfectly without a capacitor, and there's no reason to want to include one.

My tutorial board gives 0.1% resolution, I don't say 'accuracy', because you are limited by the tolerance of the components - using the LM4050 won't make any difference whatsoever!.

Personally I think you're making a big fuss other absolutely nothing! - I presume you've got the 431?, then test it without a capacitor!.

 

[Futterama]

Personally I think you're making a big fuss other absolutely nothing! - I presume you've got the 431?, then test it without a capacitor!.

Perhaps you are right, but Nigel, I started this thread because I was uncertain whether to use a capacitor or not - I will try the 431 without a capacitor then (yes, I have a couple of those in SOT23).

And since you obviously know a bit about the 431, could you then tell me a bit about the resistor used in series? My version of the 431 (ZR431) should have a minimum of 50μA cathode current to keep regulation, so wouldn't it be alright to calculate the resistor value from this? The input leakage current of the Vref pin on the PIC used, is only 1μA, so if I use a 33k resistor with the ZR431 and a supply voltage of 5V, the cathode current would be 2.5/33000=75μA.

I think this should be alright, or am I wrong?

 

[Nigel Goodwin]

I've used a 2.2K on mine (I've used the 431 on a number of occasions), and going from memory I used that value as a comprimise between preformance and supply current. However, exceptionally low current wasn't an issue for me!.

If you really want to minimise the current, how about feeding the 431 and resistor from a PIC output pin, and only power it when you are taking a battery reading?. There's no need for a continuous monitor of the battery voltages - every 5 or 10 secinds would be more than enough!.