I have seen FET's only used once as a constant current source in an amp design (A JLH circa 1974) but note that mostly 2 BJT's are used for this purpose. With the higher resistance of an FET surely they would be preferable (and reduce component count)?
Would the current supplied by an FET CCS be more or less temperature sensitive?
What are the other pros and cons of each method?
Any advice comments would be appreciated.
Regards,
Would the current supplied by an FET CCS be more or less temperature sensitive?
What are the other pros and cons of each method?
Any advice comments would be appreciated.
Regards,
I think the curren trend is JFET current sources. They are simple to do but as for performance i have no idea.
My personal favorite is a cascaded single bipolar ccs with LED's as voltage references, but i am an inept audio designer who beleives everything spice tells me so do not take my word for it.😀
My personal favorite is a cascaded single bipolar ccs with LED's as voltage references, but i am an inept audio designer who beleives everything spice tells me so do not take my word for it.😀
Hi !
At least in sims the jfet-ccs is far superior to the 2xbjt.
But: For a known current you need to measure/match the jfets
before using them. Without this you get currents with +/- ~30%.
That's because jfets differ much in production. And bjts are cheap !
For a finished highend amp i would prefer the jfet-ccs, but for only
testing some topology...
Additionally jfets don't withstand high voltages, except some special
devices like the sk330 (max-Vgds = 50volt)
Mike
At least in sims the jfet-ccs is far superior to the 2xbjt.
But: For a known current you need to measure/match the jfets
before using them. Without this you get currents with +/- ~30%.
That's because jfets differ much in production. And bjts are cheap !
For a finished highend amp i would prefer the jfet-ccs, but for only
testing some topology...
Additionally jfets don't withstand high voltages, except some special
devices like the sk330 (max-Vgds = 50volt)
Mike
Hi, polsol.
Fet's CCS are the easiest and often the best way to implement current sources in the range of 1 to 10 mA and with some types of fets even 15 mA.
Of course you can paralleling them for more current but not to much because of the decreasing impedance.
Fet's CCS are noisy below 100-200 Hz, this can be a problem, but depends where you use them, and the gain of your design.
You can do good CCS with bipolars (prefer low noise types), the emitter resistor must be high, at least 5 kOhms, and the more important the reference voltage at the base must be, very very well decoupled for noise an hum.
Fet's CCS are the easiest and often the best way to implement current sources in the range of 1 to 10 mA and with some types of fets even 15 mA.
Of course you can paralleling them for more current but not to much because of the decreasing impedance.
Fet's CCS are noisy below 100-200 Hz, this can be a problem, but depends where you use them, and the gain of your design.
You can do good CCS with bipolars (prefer low noise types), the emitter resistor must be high, at least 5 kOhms, and the more important the reference voltage at the base must be, very very well decoupled for noise an hum.
I do the same. It's simple and it works well. Low noise, good stability, and not too critical on matching transistors like a fet setup is.hjelm said:
Thanks all for the information.
Must say that the noise aspect woried me somewhat. Their use would be on the input stage of a power amp.
Moving to BJT's what about using a matched pair on a single die such as the LM394 (from recollection) to improve thermal stability and also matched characteristics rather than resort to measuring individual BJT's. But are they available in NPN/PNP pairs with low noise characteristics?
Regards
Must say that the noise aspect woried me somewhat. Their use would be on the input stage of a power amp.
Moving to BJT's what about using a matched pair on a single die such as the LM394 (from recollection) to improve thermal stability and also matched characteristics rather than resort to measuring individual BJT's. But are they available in NPN/PNP pairs with low noise characteristics?
Regards
Hi !
Don't be too worried about the noise from jfets. Unless you are
using very high openloopgains, this is no big problem. I've tried
several amps with jfets, and noise was no real issue. The only
way to recognize audible noise, was to keep the ear directly to
the tweeter.
Mike
Don't be too worried about the noise from jfets. Unless you are
using very high openloopgains, this is no big problem. I've tried
several amps with jfets, and noise was no real issue. The only
way to recognize audible noise, was to keep the ear directly to
the tweeter.
Mike
- Status
- Not open for further replies.