[nux]

Quote:

Originally Posted by cerbie /img/forum/go_quote.gif The thing is, 1ma could be 2V, 10ma could be 2.5V, and 50ma could be 3.5V--not a nice neat voltage scale (it's logarithmic, as is your perception of light). So, with a plain RC circuit, you would be ramping voltage, the LED would appear to turn on late but quickly.

This is all wrong. An LED is a diode, hence its forward voltage is almost constant across a wide range of current. 0.1mA could be 2V, 20mA could be 2.001V. Look at the datasheet for an LED, it should have a V/I graph. Hence, you cannot control an LED's brightness well with a voltage source, you need to control the current going to it, hence a CCS. As I said, an LED's brightness is roughly logarithmic, the relative brightness change decreases as the current through it increases linearly. A PWM can also be used because it turns on and off the LED very fast, the brightness is roughly linear with the duty cycle.

Light-emitting diode - Wikipedia, the free encyclopedia

Have a look at the I-V diagram, and you can see at Vd (the forward turn on voltage), the voltage changes very little for a big change in current.

Quote:

Originally Posted by cerbie /img/forum/go_quote.gif If you ramp the voltage logarithmically, it will appear as if it is lighting up close to linearly. If you ramp the current linearly, it will appear to light up slightly logarithmically (however, it will probably be subtle enough that it will "look" exponential or linear).

You would want an exponential current to cancel the logarithmic brightness of an LED to make the brightness appear to linearly increase over time. Exponential is the inverse of logarithmic functions, so if it lights up logarithmically it can't 'look' exponential =)

 

[cerbie]

Quote:

Originally Posted by nux /img/forum/go_quote.gif This is all wrong. An LED is a diode, hence its forward voltage is almost constant across a wide range of current. 0.1mA could be 2V, 20mA could be 2.001V. Look at the datasheet for an LED, it should have a V/I graph. Hence, you cannot control an LED's brightness well with a voltage source, you need to control the current going to it, hence a CCS.

http://www.nichia.com/specification/...SPW545CS-E.pdf
Page 8. I've been itching to mod my X5 (CS->DS, fit in a ~150mA buck), so I have had Nichia's datasheet page bookmarked. Also, I went to few of Tangent's projects, and the indicator LEDs had datasheets with no more info than Vf at rated current(s).

You can drive an LED quite well with a voltage source (LM317+pot+spare DIP socket--try it). However, 20mA is more likely to be the same brightness between samples than 3.0V will be.

Quote:

Light-emitting diode - Wikipedia, the free encyclopedia Have a look at the I-V diagram, and you can see at Vd (the forward turn on voltage), the voltage changes very little for a big change in current.

I like real ones better than text book ones. Things do start flattening out, but usually, it seems to be in the 100+ mA range where it really starts. Even then, it's not ruler-flat through the range they'll work at.
Quote:

You would want an exponential current to cancel the logarithmic brightness of an LED to make the brightness appear to linearly increase over time.

Yes. A linear voltage rise, however you'd do it, would provide this (appear is the key word).
Quote:

Exponential is the inverse of logarithmic functions, so if it lights up logarithmically it can't 'look' exponential =)

Linear V: approximately exponential current, but very little until near the rated Vf (for the linked white LED, you'd need to be in a dark room to see that anything was happening at 2.5V), then it would quickly brighten as it neared whatever stopping point it had. Starting not too much lower than the Vf would mitigate the apparent delay.
Exponential V: kind of like linear, but more exaggerated.
Logarithmic V: about linear brightness increase. You'd reach the lower end of Vf quickly, so making it start near there wouldn't be a big deal.

A linear brightness increase works about like a linear sound increase. Just the same, in a well-lit room (think noisy room), a dim LED may not be noticeable until it has gotten to some reasonable output level. That's what I meant. I did kind of jumble that all up.

That said, an actual linear increase in emitted light (very close to a linear increase in current) would probably look the most pleasing.

Finally, all of this exemplifies why it's so easy to choose to build them with high frequency PWM and figure the rest out in software

.

 

[nux]

Quote:

Originally Posted by cerbie /img/forum/go_quote.gif You can drive an LED quite well with a voltage source (LM317+pot+spare DIP socket--try it). However, 20mA is more likely to be the same brightness between samples than 3.0V will be. I like real ones better than text book ones. Things do start flattening out, but usually, it seems to be in the 100+ mA range where it really starts. Even then, it's not ruler-flat through the range they'll work at.

I presume you are doing this with an LED in series with a resistor. Increasing the voltage applied does not mean that the voltage across the LED has increased. The extra voltage is being dropped across the resistor, causing a higher current (V=IR for linear devices like resistors, NOT for diodes/LEDs). The higher current causes the LED to be brighter, the voltage drop across the LED will not change very much, as the V-I curve for diodes show.

If you are simply trying to drive an LED directly from an adjustable voltage source, as soon as you set the voltage much above the forward voltage, the LED current would exponentially run away and blow the LED. With a resistor, the resistor and voltage source is acting as a current source to the LED. This is why in any LED circuit, a resistor is needed to set the current through the LED..

I suggest you measure the voltage drop across only the LED in an LED+resistor+adjustable voltage source circuit. You will see that the voltage drop across the LED is almost constant as the applied voltage goes up, as the extra voltage is simply dissipated in the resistor. There is an initial rise in forward voltage of a few hundred millivolts, but then almost constant. Same as a normal silicon/germanium diode.

And the 'text book' V-I characteristics are there because they are what real LED/diodes exhibit. I probably have some excel graphs from labs I've done on diodes many years ago..

 

[cerbie]

*scurries to find that bag of more small LEDs than he'll ever need from Futurlec*
Edit: I've got things half packed up, and can't find them. I've exposed the contact points on my beat-up L2D-CE, instead, and gotten the Vfs for those currents that will fry most indicator LEDs (~30mA, ~80mA, ~330mA, ~700mA). Once I find a spare bare-leads DC supply, I'll bread-board it up.

However, tomb is pretty much right. LEDs complicate things and can be a PITA to work with. I've fried more than a few

--if I am indeed wrong above (maybe if something in previous circuits had the same effect as a low-Ohm resistor in series?), I can add one more (upgrade/mod time!). With the SS, I wonder if a low-voltage incan could be tied in somewhere to do the same job?

 

[tomb]

Aside from this argument over the LED brightness, I would not recommend this for the Starving Student. First of all, the combination of the power supply and the fact that the tube heaters are providing the current source for the MOSFETs means that the tubes take much longer to come up than other tube amps. Secondly, even with the SSMH PCB's improved parts selection, there is still a hiccup that occurs upon power up. The LEDs will first light, then dim, then come back up. Meanwhile, the heaters in the tubes themselves take well over 10 seconds to come up.

Another wrinkle in this that's been alluded to is that LED's don't have a linear relationship between current and brightness. They may still retain 90% at 1/2 to 1/4 the current. All these factors make it very difficult to accomplish what you're asking - much more than what's appropriate for the Starving Student, IMHO. Simple is the theme ...

P.S. Some of these currents I see mentioned would fry every LED I've used. We size resistors for tube LEDs on the Starving Student and Millett MAXes/MiniMAXes at about 10ma. Just a tiny bit over 20ma will burn the LEDs out very quickly in my experience. At 10ma, however, you might as well say they'll last forever.

 

[nikongod]

Whats the voltage drop across the cathode resistor? Is it high enough to simply swap for an LED? Its not exactly a slow controlled ramped up lightshow, but if it works the LED wont light until the amp is ready to plug headphones into.

 

[glennb]

Going back to the original idea of just putting a big cap across the LED.... Assuming there is a dropping resistor from DC power supply somewhere in the amp (even if its the heater or high tension supply !), a largish electrolytic capacitor should do the trick. It only needs to be a low voltage rating, say 6.3, 10 or 16 volts. For best effect, increase the value of the dropping resistor to as high as possible without dimming the LED too much. Then connect a cap and observe the result. I suggest starting with 470uF, then 1000uF, 2200, 4700 and see what happens. Don't forget to use the correct polarity, the - wire of the cap goes to the cathode of the LED.

 

[Voodoochile]

Would it be too easy to use an incandecent grain of wheat lamp? If you get one rated higher than your max voltage it will last a very, very long time.

 

[nullstring]

What about this?
Soft Start PSU

I am kind of busy today, I am gonna try to figure out how to (or if I can) modify that for my use later.

 

[peranders]

Have you seen the white LED on an iMac which breathes when the Mac is asleep. The LED dims also when the light is low and the dimming process goes smoothly.

 

[nullstring]

So, I decided that the soft start schematic with the LM317 should work well.
(If I am wrong, tell me so)

things I need to figure out:

1. I need to figure out how to calculate the warm up time based on the capacitors and resistors used.
2. the LM317 can't take in a voltage as high as 48V. I can put resistance before the regulator in order to reduce input voltage, right?