Constant Current Source (CCS) For Audio Applications

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disclaimer: I'm just a lowly technician, not an engineer. Many here are well above me in areas related to the knowledge of electronics and more creative as well. However, given the prior statement, please do qualify any statements / assertions made. Don't state something as a fact without some verifiable reference, preferably a link to a professional paper / resource, or at least the name of the resource and author. To simply say 'because it sounds better' will be considered inadequate for the purposes of this thread.

----------

I'd like to open a discussion on the humble 'Constant Current Source', or CCS, as this fundamental building block relates to its uses in audio amplification.

We often see the CCS used as a source or sink for current where we want a relatively stable and defined current to flow in the face of variable conditions in adjacent areas of the circuit. Some people feel a specific arrangement offers better performance over other circuits in an amplifier, while different folks feel that even basic CCS is more than adequate for almost all needs. The idea here is not to settle any sort of dispute but rather to discuss and perhaps demonstrate, at least in simulation, what might be preferable with the where and when qualified.

Perhaps the application dictates how 'good' the CCS needs to be? Is there a point of 'good enough'? Is a better CCS always better in audio?

----------

If possible it would be nice to lay down a few guidelines. For starters:

1) Avoid exotic components. The best design is useless if the parts are made of Unobtanium. Parts should be easily available from main stream distributors. Components found in DIY-friendly SMD package sizes are fine.
2) Limit complexity. A CCS that takes more PCB real estate than the amplifier is not practical, especially if the gain in performance of the circuit as a whole is minimal. Let's target not more than two active devices plus required supporting passives unless the increase in performance is dramatic enough to warrant an extra active device.
3) Let's have some fun. Please no poo-flinging, flaming, trolling. Keep things friendly, courteous and civil.

----------

As a starter for resources:

http://waltjung.org/PDFs/Sources_101_P1.pdf
http://waltjung.org/PDFs/Sources_101_P2.pdf

----------

And attached are a few basic CCS circuits and a little of their performances. Please do point out any fundamental errors or ways to improve the simulations. Comments, ideas and suggestions?
 

Attachments

  • CCS Comparison Sch.png
    CCS Comparison Sch.png
    71.3 KB · Views: 3,938
  • CCS Comparison Current.png
    CCS Comparison Current.png
    78.7 KB · Views: 3,781
  • Current Variation v Freq.png
    Current Variation v Freq.png
    77.8 KB · Views: 3,686
  • CCS Comparison.asc
    2.7 KB · Views: 361
  • Cordell-Models.txt
    19.6 KB · Views: 266
  • Fairchild Models.txt
    933 bytes · Views: 219
Just found a paper:
Low noise constant current source for bias dependent noise measurements
D. Talukdar et al.

A low noise constant current source used for measuring the 1/f noise in disordered systems in ohmic as well as non-ohmic regime is described. The source can supply low noise constant current starting from as low as 1 μA to a few tens of mA with a high voltage compliance limit of around 20 Volts. The constant current source has several stages which can work in a standalone manner or together to supply the desired value of load current. The noise contributed by the current source is very low in the entire current range. The fabrication of a low noise voltage preamplifier modified for bias dependent noise measurements and based on the existing design available in the MAT04 data sheet is also described.
 

Attachments

  • CCS2.gif
    CCS2.gif
    46.6 KB · Views: 4,930
  • CCS.gif
    CCS.gif
    48.6 KB · Views: 5,033
Founder of XSA-Labs
Joined 2012
Paid Member
Just found a paper:
Low noise constant current source for bias dependent noise measurements
D. Talukdar et al.

A low noise constant current source used for measuring the 1/f noise in disordered systems in ohmic as well as non-ohmic regime is described. The source can supply low noise constant current starting from as low as 1 μA to a few tens of mA with a high voltage compliance limit of around 20 Volts. The constant current source has several stages which can work in a standalone manner or together to supply the desired value of load current. The noise contributed by the current source is very low in the entire current range. The fabrication of a low noise voltage preamplifier modified for bias dependent noise measurements and based on the existing design available in the MAT04 data sheet is also described.

Nice find! I love 2sk170's and was hoping there would be a way to use them for a high performance CCS.
 
Just found a paper:
Low noise constant current source for bias dependent noise measurements
D. Talukdar et al.

A low noise constant current source used for measuring the 1/f noise in disordered systems in ohmic as well as non-ohmic regime is described. The source can supply low noise constant current starting from as low as 1 μA to a few tens of mA with a high voltage compliance limit of around 20 Volts. The constant current source has several stages which can work in a standalone manner or together to supply the desired value of load current. The noise contributed by the current source is very low in the entire current range. The fabrication of a low noise voltage preamplifier modified for bias dependent noise measurements and based on the existing design available in the MAT04 data sheet is also described.


Here's my results with the schematic on the paper.
Y-axis: decibel
 

Attachments

  • CCS3.gif
    CCS3.gif
    11.2 KB · Views: 1,493
Last edited:
Nice thread!

We can add the shunt voltage reference CCS. So either a Zener diode or a programmable reference like the TL431. I hear that the TL431 is supposedly quieter than a Zener as it doesn't have the Zener switch noise. The shunted voltage the controls then base of the CCS transistor.

Yes the TL431 is definitely an option. I haven't yet procured a model for the device that works in LTspice yet. Admittedly I haven't looked too hard either.
 
The op's sim is flawed since R4/5 are fed ac as well. They should be connected to a dc source with no ac signal.

Noted. I will set up a sim with a stable bias source and see how they look.

Edit: Changes the results quite a bit. Things actually look closer to how I would expect. Thanks for pointing out my error.
 

Attachments

  • CCS Compared Alt Bias.png
    CCS Compared Alt Bias.png
    120.8 KB · Views: 1,629
Last edited:
Member
Joined 2011
Paid Member
An under-appreciated attribute of current source (Q2/Q3/nodeB) from post #1 is: extremely high output impedance thanks to negative feedback. Much higher than the 1T circuit Bob Cordell recommends (6.3V zener drops 5.6V across emitter resistor) to people who seek high output impedance. Careful / clever choice of R5, or R5 replacement, gives excellent PSRR too.
 
An under-appreciated attribute of current source (Q2/Q3/nodeB) from post #1 is: extremely high output impedance thanks to negative feedback. Much higher than the 1T circuit Bob Cordell recommends (6.3V zener drops 5.6V across emitter resistor) to people who seek high output impedance. Careful / clever choice of R5, or R5 replacement, gives excellent PSRR too.

Can you offer any insight into optimizing the 2 BJT biasing arrangement?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.