C4S for differential pair input stage cathode bias

[Argopo]

Hey Paul B, Paul J, and Doc B,

Based on this previous thread (https://forum.bottlehead.com/index.php?topic=1373.0), I bought a single C4S last month with the intention of using it as a CCS for an input differential pair for a two-stage PP headphone amp.

In my use case, I plan to "feed" the CCS with a negative voltage supply, similar to the examples used in Morgan Jones' book, 4th Ed. The simplified circuit diagram is attached. I have also attached the LTspice sim of the two-stage circuit.

Using a 20V voltage source for the negative voltage, and assuming 0.855V drop across R1 with ~30mA, the value for R1 should be 28ohms. And the value for R2 should be around 10k. But, based on the sim in order to get it work correctly, the value for R2 exceeds what I calculated above based on the examples in the previous thread. So, I must not understand how the C4S would work in my application, or I setup the sim incorrectly.

If I decrease the value of R2 to ~2K and increase the value of R1 to ~75ohms, the circuit works more reasonably.

If possible, could you take a look at the attached LTspice schematic, and tell me if I configure the C4S correctly? Or provide any advice on using the C4S in my application. (Which is different than those examples provided in the previous thread.)

Thank you!

[Paul Birkeland]

I would recommend sending at least 3mA of current through the LEDs.  The linearity of voltage vs. current for the HLMP6000 may not be well reflected in the SPICE simulation, which would account for seeing a lot of variation depending on how much current you send through the LEDs.  Still, in the simulation you posted, you are sending a little more than 500uA through the LED string, which will cause some deviation from the characteristic 1.57V we assume is present across the HLMP6000s themselves. 

[Paul Joppa]

The other reason you need plenty of R2 current is that the transistors consume some base current - for example, the current gain of the MJE340 is between 30 and 240 times. So it can consume as much as 1mA all by itself. For the LED to maintain a reasonably constant voltage, the should carry something like 10 times the base current, which gives 10mA. I have on occasion provided a smaller margin without problems, but I can't defend that as a regular thing.

[Argopo]

Ok. I tweaked the value of R2 for 10mA of current through the LEDs, and adjusted R1 accordingly.

Thank you both!