Min or Typ?

[zesla]

Hi guys,
I have a TL072CN OP-Amp chip, As you know most of datsheets indicate 2 values (Min & Typ). the gain of the siad op-amp with the configuration I have used is 101 (1+1M/10k).
Now I am not sure which parameters should be used (Min or Typ) to calculate the GBP?

Thanks for any help

 

[MikeMl]

Depends if you are real engineer or a Hobbyist

 

[Nigel Goodwin]

You should always design using min values.

 

[zesla]

Thanks but I need more info please.

 

[zesla]

You should always design using min values.

Can you explain it more please? I do not know why there are Min, Typ and Max values there? the real reason is vague to me.

 

[DerStrom8]

Can you explain it more please? I do not know why there are Min, Typ and Max values there? the real reason is vague to me.

The "min" is what the chip needs in order to function. For example, if a chip has a 5 volt min, it means that the chip cannot work with less than 5 volts. The typical ("typ") is what the chip will usually use while providing the best performance. So, if the chip mentioned above has a typical voltage input of 7.5 volts, it means that this is the most common supply voltage and the chip will function best at this value. The "max" value is the maximum value the chip can withstand. So, if the chip has a maximum supply voltage of 10 volts, it means that if you put more than 10 volts into it, it will likely burn out or break.
I hope this helps (I realize this may seem obvious, so let me know if this isn't what you were asking).
-Der Strom

 

[Nigel Goodwin]

Can you explain it more please? I do not know why there are Min, Typ and Max values there? the real reason is vague to me.

If you go and buy a pint of milk, it is guaranteed to have 1 pint inside (that's Minimum), it may have 1 pint plus a little bit (Typical), or it may have 1 pint plus a bigger bit (Maximum)

That's basically what it means, although spreads in transistors etc. are FAR, FAR more than that example.

So to design a circuit you should use the MINIMUM figure, then it will work with any example - if you're using Typical figures then you would need to manually test each part and select the ones which meet that requirement. It's been common in various circuits over the decades for transistors to be selected for a particular circuit, or for component values to be adjusted during manufacture to take account of value spreads. Both techniques are very rare now, and to be frowned upon, as they add expensive to the manufacturing process.

 

[dknguyen]

I use worst case values (whether worst case is the maximum or minimum of a particular value). Because typical is typical and isn't guaranteed.

Min, max and typical values are there because there is variation between components. They'll probably be somewhere around the typical value, you can't be sure unless you test the devices individually...and seriously, who's going to have the equipment and time to do that? So realy, I guess the reason typical values are there is to make the device sound better than it really is for marketing purposes? Otherwise you'd only really need to list worst case specs.

I guess if for some reason your project did allow individually testing components, you could use min, typ, and max to narrow down what components might have individuals that would meet your specs so you could speed up your individual testing. Though no one ever does that because it's bad practice.

If you were building a bridge you wouldn't want pick a steel by how strong it "typically" is. Heavens no! You want steel that is always at least a certain strength or stronger! . You don't want a bridge that will work most of the time- you want a bridge that will work ALL of the time! Same deal in circuits

 

[lilimike]

Isn't TYPICAL the safe value to use since it is between MIN and MAX?
If components in general have a % tolerance, shouldn't I want to aim in the middle to be safe?

Mike

 

[MRCecil]

Hi guys,
I have a TL072CN OP-Amp chip, As you know most of datsheets indicate 2 values (Min & Typ). the gain of the siad op-amp with the configuration I have used is 101 (1+1M/10k).
Now I am not sure which parameters should be used (Min or Typ) to calculate the GBP?

Thanks for any help

The actual reason for these distinctions is the variability in the manufacturing process. It is a matter of the statistical variation of die across a wafer in a lot. The manufacturer bases the minimum, typical and maximum on specific target criteria adjusting the process to meet the criteria for maximum die yield. A number of manufactures will place the three sigma curves on their datasheets for more complex IC's.

 

[dknguyen]

Isn't TYPICAL the safe value to use since it is between MIN and MAX?
If components in general have a % tolerance, shouldn't I want to aim in the middle to be safe?

Mike

I think you are thinking more along the lines of "more likely to getyour circuit to be close to the calculations" but ending up with multiples of the same circuit that are closer to the calculations on average does not guarantee that all circuits will be close enough to the average to work properly. Designing circuits in that way also makes them vulnerable to specifications can drift with operating conditions (temperature, input voltage, supply voltage, etc) as the circuit simply might stop working if it gets too hot or the voltage supply becomes too far from nominal even if it might not actually cause any damage. That specification drift along with component variation can be a real problem.

It is likely cheaper to use typical specs since a typical spec part costs less than a worst case spec part of the same value. However, it is definately not safer than basing your design so it works off the worst case spec (whether it be the min or max for the specific specification). Again, think of the bridge example. You don't design a bridge so that it can support between X and Y kilograms of load. You design it so it can support more than Z kilograms.

There are some circuits where you aim for the middle because you need a specific number. The most obvious that comes to mind is the RLC values when used to tune frequency cutoffs filters for filters or set closed loop gain for amplifiers. But in general, you don't do that with semiconductor components because they have many many different specs that can drift and vary with operating conditions. Why design for an operating window when you can design for an operating ceiling or floor?

 

[EngIntoHW]

Hey,

I have a question regarding Min,Typ,Max values.

I designed a simple LED driver circuit:

VF = [3.3V(Typ), 4.1V(Max)] @ 20mA.

VCE(SAT) characteristics are as follows (I'm interested in IC=20mA):

According to what VF and VCE(SAT) would you calculate the current limiting resistor?

Thank you

 

[Nigel Goodwin]

According to what VF and VCE(SAT) would you calculate the current limiting resistor?

I would ignore the transistor entirely, just calculate the resistor based on the voltage drop across the LED - the tiny drop across a saturated transistor is far too small to bother about.

 

[dknguyen]

Yup, just ignore it.

But you could figure it out though by just using Kirchoff's voltage law and very basic circuit analysis fundamentals. Using V=IR around the circuit loop...

R = (Voltage across the resistor)/Iled = (Vcc-Vss-Vled-Vce_sat)/Iled

But Vce_sat << Vcc-Vss-Vce_sat so you can just ignore Vce_sat and treat it as zero. In english, it the transistor will reduce the voltage drop across the resistor a little bit. So your resistor can be a little bit smaller.

You should learn how you come up with the current limiting resistor is sized because its really fundamental and really simple. It's not something you should memorize because it's something you should know well enough that you can just look at the circuit and write it down without memorizing or thinking about it too much.

 

[EngIntoHW]

I would ignore the transistor entirely, just calculate the resistor based on the voltage drop across the LED - the tiny drop across a saturated transistor is far too small to bother about.

Yup, just ignore it.

But you could figure it out though by just using Kirchoff's voltage law and very basic circuit analysis fundamentals. Using V=IR around the circuit loop...

R = (Voltage across the resistor)/Iled = (Vcc-Vss-Vled-Vce_sat)/Iled

But Vce_sat << Vcc-Vss-Vce_sat so you can just ignore Vce_sat and treat it as zero. In english, it the transistor will reduce the voltage drop across the resistor a little bit. So your resistor can be a little bit smaller.

You should learn how you come up with the current limiting resistor is sized because its really fundamental and really simple. It's not something you should memorize because it's something you should know well enough that you can just look at the circuit and write it down without memorizing or thinking about it too much.

Thanks guys.

What value of VF would you take when calculating current limiting resistor's resistance?

 

[Nigel Goodwin]

Thanks guys.

What value of VF would you take when calculating current limiting resistor's resistance?

If you really want to be pedantic you could consult the graph from the datasheet and see what value it shows for the current you are going to use. But really it's a complete waste of time

It's VERY, VERY, VERY unimportant - a huge range of values will work - personally (if I was going to calculate a value at all) I'd just assume 2V drop, so 3V left and whatever current you were thinking of. I certainly wouldn't get out pencil and paper, or a calculator - 1K would give you 3mA, so 330 would give you 10mA - it's as easy as that.

 

[EngIntoHW]

If you really want to be pedantic you could consult the graph from the datasheet and see what value it shows for the current you are going to use. But really it's a complete waste of time

It's VERY, VERY, VERY unimportant - a huge range of values will work - personally (if I was going to calculate a value at all) I'd just assume 2V drop, so 3V left and whatever current you were thinking of. I certainly wouldn't get out pencil and paper, or a calculator - 1K would give you 3mA, so 330 would give you 10mA - it's as easy as that.

How come it's not important.

The graph you mentioned of IF VS VF shows that for 20mA IF you get 3.3V VF.

However, the table states that
VF = 3.3V TYP @ IF = 20mA ---> R(LED) = (5V-3.3V)/20mA = 85Ω
and
VF = 4.1V MAX @ IF = 20mA ---> R(LED) = (5V-4.1V)/20mA = 45Ω

It makes a large difference as you can see.

 

[Nigel Goodwin]

How come it's not important.

The graph you mentioned of IF VS VF shows that for 20mA IF you get 3.3V VF.

However, the table states that
VF = 3.3V TYP @ IF = 20mA ---> R(LED) = (5V-3.3V)/20mA = 85Ω
and
VF = 4.1V MAX @ IF = 20mA ---> R(LED) = (5V-4.1V)/20mA = 45Ω

It makes a large difference as you can see.

Fit either of those resistors, and the LED will light up - what's the problem?

For roughly 20mA I've have stuck a 150 ohm in.

 

[EngIntoHW]

Fit either of those resistors, and the LED will light up - what's the problem?

For roughly 20mA I've have stuck a 150 ohm in.

The problem is that this LED is going to be placed inside a product.
I need to order many units of R(LED) and send them for assembly done by a factory.

So I want to know what value of R(LED) would give good brightness and will not damage the LED.

 

[Nigel Goodwin]

The problem is that this LED is going to be placed inside a product.
I need to order many units of R(LED) and send them for assembly done by a factory.

So I want to know what value of R(LED) would give good brightness and will not damage the LED.

All the values I've suggested will - as I said it's EXTREMELY non-critical.

Frist thing you need to think about is how bright you want it - this is obviously related to the current, but you've no way of knowing how bright a particular current will make it. So chuck a resistor in and see how bright it is, if you then want it brighter then reduce the value, if it's too bright, then increase the value. Calculating doesn't really come in to it, as even if you knew how bright a specific current would be, I suspect the brightness value would be totally meaningless to you (it certainly would to me).