Rise Time

[JustAnotherGuy999]

I know what rise time is, but why is it so important? it pops up frequently in electronics

[ataradov]

Depends on context. It limits your maximum frequency, for example. You can't start switching back to 0, if your signal has not reached its maximum value yet. So your minimum period ends up being P = (Trise + Tfall)  and maximum frequency F = 1 / (Trise + Tfall).

At the same time fast rise time generates transient currents and can lead to EMI issues. So you always need to balance your speed and EMI requirements.

[uncle_bob]

I know what rise time is, but why is it so important? it pops up frequently in electronics

Hi

The main reason you see it popping up is that it is a standard way to specify a square wave signal. Just about every logic signal you see coming out of anything will be specified with a rise / fall time.

Bob

[Alex Trofimov]

I think it worth adding that fast rise and fall times may be not only the issue from an EMI point of view. The shorter you rise/fall times — the more high-frequency your signal is. The most HF-component of the signal will be like rise and imediately fall and then rise and so on. So it's frequency will be 1 / (rise_time + fall_time). It is called -3dB cutoff frequency. The more it is — the shorter is the the wire that you must consider as a transmission line. In transmission lines signal does not propagate immediately, there may be reflections from the ends and all sorts of wierd and wonderful stuff) 

As an example with an ordinary RS-485 transmitter you must consider anything longer than 30 cm as a transmittion line. With slew-rate-limited (e.g. much longer rise/fall times) transmitter this length is about 50m. Also with the second one you will have less EMI problems as well, as was mentioned before.

Sorry for all the mathematical stuff, I hope you caught the idea.

[Brumby]

While the term 'rise time' is a generic one, using the special case of a square wave makes the basic understanding a bit easier to grasp.  Square waves are also very common, so using them for an explanation is valid..

There are two different applications of the term 'rise time'.  One is for a pieces of equipment that create a square wave and the other for pieces of equipment that accept a square wave.

Square waves are routinely created in digital devices such as an output from a microprocessor or a USB transmitter.  In this situation, the rise time is a measure of how quickly the output circuitry can change the logic level.  As mentioned above, the quicker it can get from Low to High, the quicker it can change back again.  If you try and drop back from High to Low before the output actually went high enough to do its job, then the circuitry receiving this signal may not register the fact that the signal DID go High.  Therefore, your timing must allow for the circuitry creating the square wave to reach a guaranteed High value before switching again.  (The same argument applies for High to Low transitions.)

There's also the issue of when a logic level change occurs, you will want it to change NOW! not ... Let's change the level ... start now ... yep - voltage is going up ... a little faster would be good ... nearly there ... we made it!  Aside from not wasting time, the crisp and snappy change is very important when you have 2 or more square waves running around that have some relationship where consistency in timing is important.  Parallel data transfer is a classic example - and sending data from your computer's RAM to the CPU is perhaps the most common example where this is critical.

The other is for a piece of equipment that accepts a square wave as an input, such as an oscilloscope, a USB receiver or any number of circuits in a computer.  In this case, the rise time is a measure of how well the receiving circuit can keep up with the signal coming in.  If the rise time of the receiving circuit is too slow, the transmitting circuit may have waited at the 'High' level long enough and already made the transition to Low before the receiving circuit registered the High.


You can see how rise times are very important in all areas of digital electronics - and become more and more critical the faster things go.

[Brumby]

Continuing on from the square wave example, the rise time is directly correlated to the frequency range the circuitry can handle.

If you know anything about Fourier analysis, the reason is clear.  If not, here's a quick explanation for the special case of a square wave:

Mathematically, a square wave (of frequency f) can be made up by adding an infinite number of sine waves, using a formula that goes something like this: ¹/₁ sin(1*f) + ¹/₃ sin(3*f) + ¹/₅ sin(5*f) + ¹/₇ sin(7*f) + ...etc.  Yes, they are all the odd harmonics - at reduced amplitudes, the higher the frequency you go.

In this formula, the factors 1, 3, 5, 7, etc... can go on forever, but the following diagram shows how how the mathematical shape gets closer to a real square wave using more terms.  In this, the number 'K' is the highest numbered factor used.

So, for K=1, the formula is simply ¹/₁ sin(1*f).
For K=5 the formula is ¹/₁ sin(1*f) + ¹/₃ sin(3*f) + ¹/₅ sin(5*f)
For K=11 the formula is ¹/₁ sin(1*f) + ¹/₃ sin(3*f) + ¹/₅ sin(5*f) + ¹/₇ sin(7*f) + ¹/₉ sin(9*f) + ¹/₁₁ sin(11*f)
... and so on.



As you can see, the higher the frequencies that are pumped into this formula, the closer the resultant waveform is to an 'ideal' square wave.  You will notice the rise time of the square wave is faster as the included frequencies range higher.

This is why - if you have a 100Mhz square wave and you want to display it on an oscilloscope, you will probably want a scope capable of 1GHz ... or maybe more.

[JustAnotherGuy999]

Square waves are routinely created in digital devices such as an output from a microprocessor or a USB transmitter.  In this situation, the rise time is a measure of how quickly the output circuitry can change the logic level.  As mentioned above, the quicker it can get from Low to High, the quicker it can change back again.  If you try and drop back from High to Low before the output actually went high enough to do its job, then the circuitry receiving this signal may not register the fact that the signal DID go High.  Therefore, your timing must allow for the circuitry creating the square wave to reach a guaranteed High value before switching again.  (The same argument applies for High to Low transitions.)

I think it doesn't matter how small the rise time is, if the voltage stay at high level enough time, for example even if rise time is 0 as far as the voltage stay enough time at high level, say 10 seconds, a slow circuitry of 0.1 Hz will detect the change of voltage. Am'I missing something?

[Zbig]

I think it doesn't matter how small the rise time is, if the voltage stay at high level enough time, for example even if rise time is 0 as far as the voltage stay enough time at high level, say 10 seconds, a slow circuitry of 0.1 Hz will detect the change of voltage. Am'I missing something?

Have fun streaming a movie over 0.1 Hz link.

[ataradov]

Am'I missing something?

You are not. But we live in a world, where 100 MBps links are nothing special and pretty mundane. So for real world applications you need to think about it.

Also, some parts (mostly MCUs and MPUs) do care about power supply rise time and may lock if you don't meet the spec.

[Lukas]

By the the way: As per IEEE 181-2011, we shouldn't use the term rise time any more, because it's

[...] deprecated because they [risetime, falltime, and transition time] are ambiguous and confusing. First, the use of the word time in this standard refers exclusively to an instant and not an interval.

The term 'falling edge' is frowned upon as well:

NOTE 2— The term falling edge is deprecated because 1) the word edge describes the property of a
geometric figure, which is not contained by, or representative of, the physical signal that corresponds to the waveform,
and 2) the word falling refers to motion or position of physical objects.

Helpfully, this standard provides some new terms:

Risetime (Rise time)  : transition duration
Falltime (Fall time)  : transition duration
Falling edge : negative-going transition
Duty cycle : duty factor
Preshoot : overshoot or undershoot in the pre-transition aberration region

Smells like academic BS, or have you ever heard anyone say 'positive-going transition duration' instead of 'rise time'?

[ataradov]

By the the way: As per IEEE 181-2011, we shouldn't use the term rise time any more, because it's

[...] deprecated because they [risetime, falltime, and transition time] are ambiguous and confusing. First, the use of the word time in this standard refers exclusively to an instant and not an interval.

Well, is not the point of the specification to make things unambiguous and non-confusing?

It is like saying "the word resistance is ambiguous and confusing", so lets call it [some super long and descriptive definition]. Just define what it means in this context and it will be fine.

[Brumby]

Square waves are routinely created in digital devices such as an output from a microprocessor or a USB transmitter.  In this situation, the rise time is a measure of how quickly the output circuitry can change the logic level.  As mentioned above, the quicker it can get from Low to High, the quicker it can change back again.  If you try and drop back from High to Low before the output actually went high enough to do its job, then the circuitry receiving this signal may not register the fact that the signal DID go High.  Therefore, your timing must allow for the circuitry creating the square wave to reach a guaranteed High value before switching again.  (The same argument applies for High to Low transitions.)

I think it doesn't matter how small the rise time is, if the voltage stay at high level enough time, for example even if rise time is 0 as far as the voltage stay enough time at high level, say 10 seconds, a slow circuitry of 0.1 Hz will detect the change of voltage. Am'I missing something?

How long did it take to get to the high level?  How long between one change of state and the next?  These are two parameters that really need to be specified before a proper assessment can be done - but I'll throw in a couple to illustrate the point...

Taking your rate of 0.1Hz, if the square wave has a rise time (sorry) transition duration of 12 seconds, you might not expect reliable operation.

If you want to make it a situation that is more 'real world' then throw a factor of 10⁹ on the above numbers...  You'd then have a 100Mhz square wave with a 12nS rise time (stuff them) will be a problem.