How "Rail-to-Rail" are we really talking here???.......

[Smokey]

Lets take the TL900x series:
https://www.ti.com/lit/ds/symlink/tlv9002.pdf

"TLV900x Low-Power, RRIO, 1-MHz Operational Amplifier for Cost-Sensitive Systems"

"The TLV900x family includes single (TLV9001), dual (TLV9002), and quad-channel (TLV9004) low-voltage (1.8 V to 5.5 V) operational amplifiers (op amps) with rail-to-rail input and output swing capabilities."

7.10 Electrical Characteristics:
Output:
Vo -  Voltage Output Swing from Supply Rails:
         Vs = 5.5V, RL=10K  ---> Typ 10mV, Max 20mV
         Vs = 5.5V, RL=  2K  ---> Typ 35mV, Max 55mV

So this Rail-to-Rail-Input-Output opamp can only actually get to best case output of 10mV of the rails, and worst case 55mv?

I understand this is significantly better than a nonRRIO part, but that could totally screw a design if those 10mV around zero were important. 

Note: 10mV is good, but it's not "Rail-to-Rail". 
Note2: I propose this:
For linear regulators we have "Low Drop Out".  They don't call them "Zero drop out*" ....."*still need 0.4V".
How about "RRI-LHO" = Rail-Rail-Input, Low-Headroom-Output

[Shay]

Seems quite decent. I don't know any RR op-amp that can swing better than that.

[Marco]

Unfortunately there's no zero ohm FETs.

[Smokey]

I'm less complaining about the 10mV here, which as I said I agree is pretty good.

It's more about the advertising as RRIO as a banner spec.  There isn't even a * footnote.  Yes I know you should be reading the whole datasheet. 

I feel it's like advertising "Free Pizza Today!".  Then when you get there, it's actually $1.  That's a great price for a pizza but it's not "Free".

[SiliconWizard]

I understand this is significantly better than a nonRRIO part, but that could totally screw a design if those 10mV around zero were important.

Yes, this is significantly better than common, non-RRIO opamps with swings limited to hundreds of mV below rails.

The zero can't be achieved, at least not for outputs. What is zero anyway in physics? If it was 10µV rather than 10mV, would you still feel ripped off? If not, why, it's still not hitting the rails?
For inputs, some opamps actually allow going all the way down to the negative rail or even slighly lower.

The way to handle this is well known, just use power rails with a slight headroom. The benefit of using RRIO opamps even in that case is that the required headroom can be pretty small.

[TimFox]

IIRC, there are/were complicated op amps that had built-in charge pumps or similar to derive a negative voltage below the negative rail applied to the device, for those who actually need zero volts to the rail.
Note that many op amps have inputs that are capable of going slightly negative with respect to the negative rail.
I would follow SiliconWizard's suggestion to provide a slightly negative rail to your device if you need < 10 mV.
In general, a positive voltage cannot go to zero unless it is capable of going negative.

[thm_w]

I'm less complaining about the 10mV here, which as I said I agree is pretty good.

It's more about the advertising as RRIO as a banner spec.  There isn't even a * footnote.  Yes I know you should be reading the whole datasheet. 

I feel it's like advertising "Free Pizza Today!".  Then when you get there, it's actually $1.  That's a great price for a pizza but it's not "Free".

Its convention. As Shay says, its about as good as you get.

You can probably find a hundred examples of this: ideal diodes, zero drift opamps, etc. There was a thread with a bunch of examples but can't recall offhand.
https://www.eevblog.com/forum/beginners/why-are-manufactorers-allowed-to-lie-(misrepresent)-datasheet-values/

[langwadt]

I'm less complaining about the 10mV here, which as I said I agree is pretty good.

It's more about the advertising as RRIO as a banner spec.  There isn't even a * footnote.  Yes I know you should be reading the whole datasheet. 

I feel it's like advertising "Free Pizza Today!".  Then when you get there, it's actually $1.  That's a great price for a pizza but it's not "Free".

more like "Pizza $1" but when you get there it's actually $1.002 with tax ....

[SiliconWizard]

Yeah it's a naming convention, but how would you call it otherwise? If you have a better idea?
ARRIO? "Almost rail-to rail"

Now how "almost" are we really taking here...?

[Smokey]

...
For inputs, some opamps actually allow going all the way down to the negative rail or even slightly lower.
...

That's true here (TLV900x). 
Input Voltage Range:
    Min = (V-) - 0.1V
    Max = (V+) + 0.1V

[Someone]

... wait until you read about zero-drift.

[Smokey]

Yeah it's a naming convention, but how would you call it otherwise? If you have a better idea?
ARRIO? "Almost rail-to rail"

Now how "almost" are we really taking here...?

Ok.  How about this?
For linear regulators we have "Low Drop Out".  They don't call them "Zero drop out*" ....."*still need 0.4V".

How about "RRI-LHO" = Rail-Rail-Input, Low-Headroom-Output

[Smokey]

I'm less complaining about the 10mV here, which as I said I agree is pretty good.

It's more about the advertising as RRIO as a banner spec.  There isn't even a * footnote.  Yes I know you should be reading the whole datasheet. 

I feel it's like advertising "Free Pizza Today!".  Then when you get there, it's actually $1.  That's a great price for a pizza but it's not "Free".

more like "Pizza $1" but when you get there it's actually $1.002 with tax ....

It's too completely different things to advertise "Cheap Pizza" and then have extra fees, than it is to advertise "Free Pizza" then charge anything.

[Smokey]

I'm less complaining about the 10mV here, which as I said I agree is pretty good.

It's more about the advertising as RRIO as a banner spec.  There isn't even a * footnote.  Yes I know you should be reading the whole datasheet. 

I feel it's like advertising "Free Pizza Today!".  Then when you get there, it's actually $1.  That's a great price for a pizza but it's not "Free".

Its convention. As Shay says, its about as good as you get.

You can probably find a hundred examples of this: ideal diodes, zero drift opamps, etc. There was a thread with a bunch of examples but can't recall offhand.
https://www.eevblog.com/forum/beginners/why-are-manufactorers-allowed-to-lie-(misrepresent)-datasheet-values/

Nice link. I feel like this is different than the thermal limits case that guys is talking about.  Under "ideal conditions" that part can output 1A, but under real world conditions, it's often much less. 
Ok, but in this case, there are no possible conditions that the output of this opamp can hit the rails. 

[peter-h]

It is possible to have rail to rail input (because you can use "depletion mode" devices or LM358-style PNPs) but it is physically impossible to swing an output rail to rail.

With very low output load you could get it within microvolts however.

This is why TI make that funny chip which generates -200mV. It opens up all kinds of interesting stuff. You can swing down to 0V, but cannot swing too far down to forward bias the protection diodes in whatever you are driving. Personally I achieve -200mV with a schottky diode Vf and some other stuff, but it has a big TC.

[MasterT]

Options.

[SiliconWizard]

Yeah it's a naming convention, but how would you call it otherwise? If you have a better idea?
ARRIO? "Almost rail-to rail"

Now how "almost" are we really taking here...?

Ok.  How about this?
For linear regulators we have "Low Drop Out".  They don't call them "Zero drop out*" ....."*still need 0.4V".

How about "RRI-LHO" = Rail-Rail-Input, Low-Headroom-Output

Why not. May sound unnecessarily convoluted, but yeah why not! I doubt you will be able to change an "industry convention" that easily though.

In any case, that's yet another case for always reading the datasheets, in particular the parameters that are critical for a given application.
No naming will save you from having to do that.

[phil from seattle]

Ever bought lumber in the USA?  At my local Home Depot (local crap lumber purveyor) I helped a couple of Japanese guys with a measuring tape who appeared flummoxed by 2x4s. When they understood, they had dissatisfied frowns. 

The fact is, marketing is almost always either an exaggeration or outright lie. I just think of RtoR as a search term to help me find a better op amp.

[thm_w]

Ok.  How about this?
For linear regulators we have "Low Drop Out".  They don't call them "Zero drop out*" ....."*still need 0.4V".

How about "RRI-LHO" = Rail-Rail-Input, Low-Headroom-Output

Yeah that is a valid counter example.
But don't worry, we already have "ultra low dropout regulators": https://www.st.com/en/power-management/ultra-low-dropout-ldo-regulators.html
and Maxim has used it to describe a buck boost: "The active rectifier also acts as a zero-dropout regulator if the input exceeds the regulated output".

LDO -> ULDO -> ZDO

[Terry Bites]

I'm moving to your super low 0.2% tax region. Go Denmark!

[Siwastaja]

You have to take basic laws of physics into account.

(Well, theoretically an opamp could have an integrated switch mode power supply generating supply voltages higher than external supply pins, but that would be a large and expensive part!)

It is possible to have input which goes exactly from rail to rail, or even beyond the rails, yes.

But output? It has to drive significant current into the load (resistive or capacitive), and it only has the rails to work with, using transistors in its output stage. Cheap, room-temperature superconducting zero-ohm transistors are not a thing, so there will be a voltage drop.

[NiHaoMike]

(Well, theoretically an opamp could have an integrated switch mode power supply generating supply voltages higher than external supply pins, but that would be a large and expensive part!)

That already exists for audio applications, eliminating the need for a negative supply rail.
https://www.analog.com/en/products/max97200.html

[Smokey]

(Well, theoretically an opamp could have an integrated switch mode power supply generating supply voltages higher than external supply pins, but that would be a large and expensive part!)

That already exists for audio applications, eliminating the need for a negative supply rail.
https://www.analog.com/en/products/max97200.html

https://www.digikey.com/en/products/detail/analog-devices-inc-maxim-integrated/MAX97200AEWC-T/2239115
That part is $1.68 in 1s qty.  Reasonable. 

[Smokey]

You have to take basic laws of physics into account.

(Well, theoretically an opamp could have an integrated switch mode power supply generating supply voltages higher than external supply pins, but that would be a large and expensive part!)

It is possible to have input which goes exactly from rail to rail, or even beyond the rails, yes.

But output? It has to drive significant current into the load (resistive or capacitive), and it only has the rails to work with, using transistors in its output stage. Cheap, room-temperature superconducting zero-ohm transistors are not a thing, so there will be a voltage drop.

I'm not trying to break the laws of physics.  But for a physics analogy, what would happen if someone came out with a cryocooler that they advertised as "True Absolute-Zero Cryocooler"   "Drops temperature to Absolute-Zero!*"   
"*Min temp = 1.75 mK"

[langwadt]

You have to take basic laws of physics into account.

(Well, theoretically an opamp could have an integrated switch mode power supply generating supply voltages higher than external supply pins, but that would be a large and expensive part!)

It is possible to have input which goes exactly from rail to rail, or even beyond the rails, yes.

But output? It has to drive significant current into the load (resistive or capacitive), and it only has the rails to work with, using transistors in its output stage. Cheap, room-temperature superconducting zero-ohm transistors are not a thing, so there will be a voltage drop.

I'm not trying to break the laws of physics.  But for a physics analogy, what would happen if someone came out with a cryocooler that they advertised as "True Absolute-Zero Cryocooler"   "Drops temperature to Absolute-Zero!*"   
"*Min temp = 1.75 mK"

about 1/1500 the temperature of space, so pretty damn close to zero ...