TL431 not working properly as tunable negative reference by the datasheet

[john23]

Hello, I am using TL431 in the configurations below to tune the negative voltage for the Vg1 port of PMA5-83-2W.
I know that TL431 is generaly used as positive tunable reference voltage.
I have been suggested the following configuration shown in the LTSICE where the input is -12V.
but as you can see in the simulation now cathode is connected to GND and the resistors in the formula are switched in the order.
So my Vref=-2.5V R_load3=2k R_load2=1k V_ka=-2.5(1+R_load3/R_load2)=-7.5V as shown below.
The resistor order is exactly the opposite than the forumla in the data sheet.
Is it a valid configuration to use in real life because in the simulation it works?
Thanks.

LTspice file is attached.
https://www.minicircuits.com/pdfs/PMA5-83-2W+.pdf

[magic]

I see nothing unusual here. You have a resistor divider between K and A, with R connected in the middle. The result is always a "zener diode" with breakdown voltage equal to 2.5V·(Rkr/Rar+1). It doesn't matter how the resistors are numbered and it doesn't matter which side is connected to ground, because it's as simple two-terminal circuit. Just make sure to always run at least (1mA + 2.5V/Rar) through it and you are good to go.

[john23]

Hello magic,Yes i need to see  what are Iref and I_Anode.
I have a problematic(so i was told) case where R_A=1K R_K=2K
Iref=2uA Ianode=-4mA as shown below.
I found in the data sheet  the  line attached.
How can i see if my Iref and I anode current are good for normal functioning?
LTspice file is attached.

[Konkedout]

Just based on your original schematic; please confirm that the cathode is positive WRT the anode?  Confirm that the reference is also positive WRT the anode?

In using the TL431 I think of it as similar to an NPN BJT, with:

1) The cathode is the collector, the anode is the emitter, and the reference is the base

2) Vbe is a precise 2.5 (or maybe it is 2.45 or ?? on the datasheet) instead of ~ 600 - 700 mV.

[magic]

Iref is simply the current flowing into the reference pin. Ideally it would be zero, but internal transistors need base current to work so it's 2~4μA in real chips. You don't need to think about it much, it's just there and it's too small to cause significant problems with the divider.

What you need to care about is the "minimum current for regulation" - there has to be at least 0.4~1mA flowing internally from K to A for the reference voltage to be accurate.

[iMo]

Works nice..

[TimFox]

Be careful when looking at the data sheet's pin labels for voltage regulator devices, especially for TL431 type circuits.
Some manufacturers label the negative-most pin as "GND", assuming the whole world uses positive DC voltages.
Others label the device as if it were a Zener diode, so the positive-most terminal is labeled "CATHODE", which would be technically correct for a Zener diode, but confusing since Zener diodes are operated in reverse-bias.
I sent a valid schematic to a PCB layout guy, including a two-terminal reference IC shown with its Zener equivalent symbol, who downloaded the pinout from the manufacturer, and connected "GND" to ground, which didn't work for my intended negative reference circuit.
[Why does no IC manufacturer sell a proper three-terminal reference voltage source for negative outputs, such as a negative version of the REF05 series, although they all make 7905-style non-reference-level negative linear regulators?]

[Phil1977]

What advantage do you see for a negative TL431 version? I suppose you`d like to make IMO's circuit symmetrically so that R1n and R2n can be swapped. But for what reason?

[TimFox]

What advantage do you see for a negative TL431 version? I suppose you`d like to make IMO's circuit symmetrically so that R1n and R2n can be swapped. But for what reason?

As I wrote, I wanted a negative version of the REF05, which is a three-terminal series regulator specified for precision use.
Lacking such a negative REF05, I used a higher-specified equivalent of TL431 (from LTC)  for negative voltages in my design. 
Otherwise, the parts count would need a precision op amp (e.g., OP-07) and two expensive resistors to obtain the stable negative bias voltage I needed.
In general, three-terminal series regulators are often superior to two-terminal shunt regulators.

Since the TL431 with the appropriate voltage divider (or wire) to the REF pin forms a circuit equivalent to a Zener diode, that combination can be used for either positive or negative voltages.
Note that the TI datasheet identifies the three pins as CATHODE, ANODE, and REF.  That is the possible confusion I was warning about in my post.
In other projects, I have used the very flexible TL431 for positive and negative voltage circuits. 
When done carefully, it can be used at higher voltages than specified with a cascode NPN transistor (or even vacuum triode) or series high-voltage Zener and additional diodes to protect the device during turn-on.

[magic]

What advantage do you see for a negative TL431 version? I suppose you`d like to make IMO's circuit symmetrically so that R1n and R2n can be swapped. But for what reason?

Look at the "System Examples" section of TI datasheet, plenty of them can't be easily converted to negative.

LM385-ADJ is something approximately similar to a negative TL431, but I don't have this chip and have never tried abusing it this way. It's also low power (read: low speed) and very limited in terms of Vka.

[TimFox]

Right, but the basic Zener-like configuration of the TL431 (and possibly two resistors) can be connected to replace a Zener diode for either a positive or negative shunt regulator, as I wrote.
Amplifier or comparator configurations are different.

At the time where I designed that circuit for work (mid 1990s), I looked hard for a negative three-terminal series reference regulator, including asking the relevant sales engineers from AD, TI, and LT, and none were available so I chose the best shunt reference regulator I could find plus a reasonable-price series resistor to get my negative reference voltage.

[john23]

Hello, I have made a configuration for  -2.5V.
As you can see in the photos and attached LTSPICE files.
My currents are 1uA in the Iref and 9 mA to the I_anode=is it a good resistor configuration to create -2.5V?
Thanks.