You can't go wrong referencing the data sheet

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
I absolutely agree you get a better design following guide lines in the data sheet. I just would not limit the design to the suggestions in the data sheet and text book.

My LED flashlight, and probably yours also, works just fine with battery and LEDs. No resistors needed.

Good theory suggests not charging batteries in series, but every rechargeable battery pack does so.

The data sheet for TTL chips may say 4.75 to 5.25 V for VCC. Of course we can run them at voltages outside these suggestions but performance may vary outside the specs on the data sheet. Most work just fine at 4.5 or 6 V. You may even get some to operate at 9 V, but I would not market such a product.

Filter caps on battery powered circuits? Well, I have omitted them on occasion, but again ... I would not put such a product on the market without them.

Suggested current for an LED may say 20 mA. That doesn't mean you MUST drive them at 20 mA. If you don't need full brightness they may work just fine at 5 or even 1 mA. You can drive an LED from a CMOS logic chip/ I just would not try to drive other logic gates from that same output.

Drive an LED, cathode to ground, from a TTL gate or op amp? Sure. use the current limiting high side driver as the ballast resistor.

Drive an LED from TTL, anode to VCC through a resistor, and still use it to drive other logic gates? Sure, just keep the current down to a few mA and use a parallel resistor to give a good high out.
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
Caution for the uninformed.
There are a lot of bad advice in the above post.
Follow at your own risk.
I agree completely. You can't go wrong following good advice. Most of the above suggestions may work fine on the breadboard but I would not put them into a product.
That being said ... what's in your flash light or battery pack?
 

kubeek

Joined Sep 20, 2005
5,794
Suggested current for an LED may say 20 mA. That doesn't mean you MUST drive them at 20 mA. If you don't need full brightness they may work just fine at 5 or even 1 mA. You can drive an LED from a CMOS logic chip/ I just would not try to drive other logic gates from that same output.
The only thing I can agree with, the rest of it is all just bad practice.
 

wayneh

Joined Sep 9, 2010
17,496
For any design goal, there is a continuum of solutions ranging from state-of-the-art to "barely works". It's the skill of the designer to choose which solution is appropriate. An awfully large proportion of advice dispensed here relates to choosing an appropriate solution.

For instance lighting an LED without a resistor can be the absolute best choice in one application and a terrible choice in most others. Need a square wave? That might be a one-op-amp circuit or a thousand dollar function generator.

It's why we have to pull teeth for specifications. Context matters.
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
The only thing I can agree with, the rest of it is all just bad practice.
So what is in you flashlight? Do you not have anything with a multi-cell battery pack? In general, yes, follow good advice. But what I do on the breadboard is not what I would put into a product.
 

kubeek

Joined Sep 20, 2005
5,794
In my flashlight I have three primary AA cells and a PWM controller.
As for the charging, I am not entirely sure which battery packs are you talking about, but any lithium based battery charger will have balancing which basically substitutes parallel charging. I haven´t seen any NiCd or NiMH packs for a long time so not sure how it was done with those.
 

crutschow

Joined Mar 14, 2008
34,285
I think the only time you can power an LED directly from a battery without a current limit circuit of some sort, is when using a small button or coin cell that has a high internal resistance.
Try doing that with low resistance batteries such as AA or AAA alkalines or NiMHs and you will likely fry the LED.
 

dannyf

Joined Sep 13, 2015
2,197
I think the only time you can power an LED directly from a battery without a current limit circuit of some sort, is when using a small button or coin cell that has a high internal resistance.
Look up led drivers - a lot more leds are lit without any resistors at all.
 

wayneh

Joined Sep 9, 2010
17,496
The example of an LED on a button cell battery doesn't count, because it does not technically deviate from the datasheet of either component. In fact it relies on the artful combination of properties.
 

ErnieM

Joined Apr 24, 2011
8,377
I'd like an example of that.
In my 30 year career I have but one. Design was based on a PIC controller selected because it was small and could run with a clock up to 40MHz, and I needed 24MHz. About a year into development (not a full time effort) I found a note in the data sheet that crystal clocks were only spec'd to run up to 20MHz.

As a good engineer I searched for a replacement. There was none.

As a pragmatic engineer I look at all my test data that showed no problem running at that frequency. Further more, were there to ever be a problem in production it would be caught 100% in acceptance testing. Rather than cancel the project over a "paper problem" I kept my doubts to myself.

The product has been their top seller for 15 years now with excellent reports back from the field.
 

dl324

Joined Mar 30, 2015
16,846
A dumb engineer follows the datasheet;
Disagree.
A good engineer knows when to deviate from the datasheet and what risks s/he is taking by doing that.
A smart engineer would contact the manufacturer to determine whether an out-of-spec feature can be depended on. A smart manufacturer would say no.

If you're going to operate a part at out-of-spec conditions, you need to do 100% screening for the characteristic that isn't guaranteed.
 

dannyf

Joined Sep 13, 2015
2,197
No one said the limit had to be from a resistor.
That's the point about needing to understand why things are done certain ways.

A dumb engineer would conclude that you have to use a resistor to limit current to an led;
A smart engineer would conclude that the essence here is to limit current to an led, however you achieve that is application specific. ie in some cases you may have to use a resistor and in other cases it is entirely OK to NOT use a resistor.

It is the understanding of why the resistor is there that matters. The fact that the resistor is there doesn't.
 
Top