Is high, or low, Hfe better for a CCS?
 

Using a PNP transistor in a CCS, is it better to select transistors with a low Hfe ... or a high Hfe? :confused:

I am currently measuring Hfe for 40 x 2AS1142s I bought recently and have observed readings ranging from 123 to 527.

Sure, a pair of transistors used for R&L channels should be matched ... but does it make any difference whether they measure 123 or 527? I can see that a different Hfe will probably change the value of the CCS resistor but is there any other effect?


Thanks,

Andy

Are we supposed to guess which CCS circuit you are using?

If you're using a bipolar ccs or cascode (e.g., the diyAudio circuits or the ones in Valve Amplifiers), higher hfe is better. Note that hfe will have almost no effect on the choice of current setting resistor.

Ah, sorry ... I had no idea there were variants. ;)

I have attached a jpg schematic, showing the way my CCSes fit into the circuit:

Thanks,

Andy

Thank you very much, SY.

Can you elucidate as to:
* why high Hfe is better, and
* how come it makes no difference to the value for the current-setting resistor (although I'm delighted to hear it! :) ).

Regards,

Andy

Higher hfe means higher output resistance (what you want in a CCS). The current is determined by the reference voltage minus Vbe (0.6-0.7V) compared to the voltage dropped across the emitter resistor. Assuming that hfe is reasonably high, one can neglect base current and assume that collector and emitter current are roughly equal, and hfe doesn't enter into your calculations.

Have you thought your circuit through?

Humour me for a moment - I have never studied CCS in any detail. I can come up with an argument that lower hfe means higher output impedance.

The emitter resistor voltage does not depend on hfe, as the resistor scales inversely with emitter current. Whether you have an xK resistor with hfe of infinity, or a 0.5xK with hfe of unity you get the same voltage and same voltage change at the emitter when the collector current changes. hfe comes in when you consider the effect of the source resistance Rb feeding the base. The CCS output resistance gets a term proportional to Rb/hfe, which argues for low hfe. Of course, in most designs people try to get a low Rb but I am not sure why (noise?).

I could call this 'back of envelope'. except that there is no envelope. There is probably a flaw in my reasoning. I assume that the Early voltage is independent of hfe.

Basically, the output resistance will be 1/hoe + Re*hfe, but the latter term usually dominates. This assumes a low source resistance at the base, which is the case for most good references.

So how does the mu (voltage gain) term for a valve turn into hfe (current gain) for a BJT? Provided Rb/hfe is sufficiently small how can base current affect the result? It is merely a tiny correction, not a scaling factor.

If hfe is reasonably large, it doesn't- Ic and Ie can be considered equal. If the source resistance to the base is high, that introduces an error into Ve ~ Vref - 0.6.

Or am I not understanding what you're saying?