[sdiy] LEDs and noise

Neil Johnson neil.johnson71 at gmail.com
Tue May 12 00:24:21 CEST 2015

Hi John,

>> no diodes? (gnd to base)
> I was puzzled about that too.  I think what Neil meant is a constant
> current driver using a PNP transistor as described in the transistor
> chapter of the Art of Electronics.  Here's a drawing of what me meant (I
> think):
> https://dl.dropboxusercontent.com/u/23053666/CCLed.pdf
> R1/R2 are selected to put the transistor into the active mode (not
> saturated).  Rset sets the LED current assuming the current gain is high
> enough (usually true).  The shorting transistor diverts current into the
> load (LED or shorting transistor).
> R1/R2 is a simple voltage divider in which the base current can be
> considered negligible.  The base voltage but must be constant to deliver
> constant current.  A zener diode or two or three forward biased series
> diodes are another way to guarantee a constant base voltage.

Exactly this.  The particular application does not require a specific
current, just a constant (or very near constant) current somewhere in
the region of whatever the designer intended (e.g., 10mA).  So whether
it turns out to be 9mA or 11mA is not important, and in a mixing desk
(I think I saw this in an API desk schematic...) parts cost is king so
common component values are important to keep manufacturing costs

If you want a more precise current source then you can use a zener
diode or a string of silicon diodes (you need at least two) to give
you a fixed base bias voltage... although Zeners are typically +/- 10%
tolerance, 5% if you pay more.

Or with more complex arrangements you can end up with something like this:
Here, R4 sets the current flowing in Tr4, and Tr1,2,3,5,6,7 then
mirror that current out to all the LED chains, so that whatever
comparators have flipped their outputs the supply rails see an almost
constant current.  With R4 = 2k7 sets the current to about 10mA
(assuming +/- 15V rails).


More information about the Synth-diy mailing list