Searching the "best" CCS

[forr]

I mean left from right.

[keantoken]

Hellooo. I'm sure everyone is still sorta interested in a good CCS.

Below is a comparison. On the left is a CCS from the thread below. Sure it's old, but WHO CARES. IT WORKS WELL. It does better than MikeB's favorite (no. 7).

Constant Current Source article (Old)

On the right is a design of my own. Yes, it has the same temperature defects as usual, but it has excellent characteristics. READ: YOU MUST MUST MUST MUST MUST TUNE R1 TO THE RIGHT VALUE OR IT WILL NOT WORK RIGHT. IF YOU POST THAT IT DOESN'T WORK I WILL POST SAYING THAT IT DOES WORK AND THAT YOU TUNED IT WRONG.

Please read the uppercase sentences in the above paragraph before critiquing my design.

Now, my circuit has the drawback that it doesn't have as good frequency characteristics as most others because the tuning goes off at higher frequencies. This might could be fixed by adding a filter somewhere.

The transistors used here have low beta and low Cjo. I believe this is fair. In my experience, some have shunned designs not knowing that if you used different transistors, it actually works quite well. This is the difference between selecting different transistors for different tasks, and selecting the first one that pops up in the sim menu!

At any rate, circuit and simulation below.

- keantoken

[AndrewT]

Hi K,
which node does the load connect to?

Why does R1 appear to pass ~5.4mA, but R2 passes ~3.3mA? Surely the missing 2mA cannot go through the bases.

[lineup]

AndrewT
Are you blind
Q1 passes like 2 mA

[AndrewT]

Originally posted by lineup
Are you blind

I'm getting tooo ooold for this.
Now, where did I put my other pair of specs?

[keantoken]

In my design the modifications from a normal CCS should be apparent. I only just realized this but you can do the same modifications with the other CCS and probably have a better circuit. Hmmm.

- keantoken