Prev: Death & Destruction of a Fluke Multimeter
Next: Variable Square Wave to Sine Wave
From: John Walliker on 8 Mar 2010 09:47
The TI data sheet for the TL431 shows two different stability graphs,
one for "All TL431 and TL431A devices except for SOT23-3, SC-70 and Q-
temp devices" while the other is for "All TL431B, TL432, SOT23-3, SC70
and Q-temp devices". Both graphs are for 25 deg C only.

Are these really all the same device, or are there two die variants?
The first group need less capacitance for stability at low output
voltages.

Similarly, ST make a TL431 which appears to need less capacitance than
the TI versions. Is this device really "better", or have they found a
more favourable way of measuring it?

None of the data sheets mentions the esr of the capacitors used for
stability measurement, nor do they discuss temperature effects on
stability. I have tried adding between 1 and 2.2 Ohms series
resistance to the load capacitor and this allows much lower value load
capacitors than without any series resistance. (I am using ceramic
capacitors and I can cope with a slight reduction in regulation
accuracy. Operating voltages are 2.7 and 3.3V and I am using the SO8
package.)

I'm trying to minimize cost but I don't want a stability problem to
crop up after 10k units have been built.

Any thoughts or painful past experiences?

John

From: legg on 8 Mar 2010 16:50
On Mon, 8 Mar 2010 06:47:47 -0800 (PST), John Walliker
<jrwalliker(a)gmail.com> wrote:

>The TI data sheet for the TL431 shows two different stability graphs,
>one for "All TL431 and TL431A devices except for SOT23-3, SC-70 and Q-
>temp devices" while the other is for "All TL431B, TL432, SOT23-3, SC70
>and Q-temp devices". Both graphs are for 25 deg C only.
>
>Are these really all the same device, or are there two die variants?
>The first group need less capacitance for stability at low output
>voltages.
>
>Similarly, ST make a TL431 which appears to need less capacitance than
>the TI versions. Is this device really "better", or have they found a
>more favourable way of measuring it?
>
>None of the data sheets mentions the esr of the capacitors used for
>stability measurement, nor do they discuss temperature effects on
>stability. I have tried adding between 1 and 2.2 Ohms series
>resistance to the load capacitor and this allows much lower value load
>capacitors than without any series resistance. (I am using ceramic
>capacitors and I can cope with a slight reduction in regulation
>accuracy. Operating voltages are 2.7 and 3.3V and I am using the SO8
>package.)
>
>I'm trying to minimize cost but I don't want a stability problem to
>crop up after 10k units have been built.
>
>Any thoughts or painful past experiences?
>
>John

Work in the low side of the capacitance range. Less than 4n7 gives no
trouble, is cheaper and has fewer temperature considerations.

If you want filtering or load immunity, filter at the point of load -
this works better, anyways, while allowing optimum low cost R and C.

RL
From: Robert Baer on 9 Mar 2010 04:12
John Walliker wrote:
> The TI data sheet for the TL431 shows two different stability graphs,
> one for "All TL431 and TL431A devices except for SOT23-3, SC-70 and Q-
> temp devices" while the other is for "All TL431B, TL432, SOT23-3, SC70
> and Q-temp devices". Both graphs are for 25 deg C only.
>
> Are these really all the same device, or are there two die variants?
> The first group need less capacitance for stability at low output
> voltages.
>
> Similarly, ST make a TL431 which appears to need less capacitance than
> the TI versions. Is this device really "better", or have they found a
> more favourable way of measuring it?
>
> None of the data sheets mentions the esr of the capacitors used for
> stability measurement, nor do they discuss temperature effects on
> stability. I have tried adding between 1 and 2.2 Ohms series
> resistance to the load capacitor and this allows much lower value load
> capacitors than without any series resistance. (I am using ceramic
> capacitors and I can cope with a slight reduction in regulation
> accuracy. Operating voltages are 2.7 and 3.3V and I am using the SO8
> package.)
>
> I'm trying to minimize cost but I don't want a stability problem to
> crop up after 10k units have been built.
>
> Any thoughts or painful past experiences?
>
> John
>
Be advised that every manufacturer has their own design; there are at
least 3 different circuits being used.
The TI design is completely different than the ST design.
From: John Walliker on 9 Mar 2010 05:56
On 8 Mar, 21:50, legg <l...(a)nospam.magma.ca> wrote:
>
> Work in the low side of the capacitance range. Less than 4n7 gives no
> trouble, is cheaper and has fewer temperature considerations.>
> If you want filtering or load immunity, filter at the point of load -
> this works better, anyways, while allowing optimum low cost R and C.>
> RL

That does sound a good idea when it is possible, but unfortunately I
need more capacitance to absorb esd events and to cope with external
devices being connected, each of which has 100nF of its own
decoupling.

Thanks

John
From: John Walliker on 9 Mar 2010 06:02
On 9 Mar, 09:12, Robert Baer <robertb...(a)localnet.com> wrote:
>    Be advised that every manufacturer has their own design; there are at
> least 3 different circuits being used.
>    The TI design is completely different than the ST design.

Thanks. That is very useful to know. I will bench test with all
makes that are likely to get used in production. Do you know whether
TI are shipping two different designs or just specifying them in
different ways?

John
 |  Next  |  Last
Pages: 1 2
Prev: Death & Destruction of a Fluke Multimeter
Next: Variable Square Wave to Sine Wave