Have you solved your probleme yet?
There are two things about your circuit that came to my mind:
First is: the circuit is a current source feeding a shunt regulator. It obviously cannot source more current than the current source does. Your current source is set to 0.125A. Be sure sure the circuit you are feeding is not trying to pull more than that. If you are testing the regulator with a resistor-load, it shouldhave more than , say 100 ohms. Not to forget the current that runs through the divider that gives FB to the 431.
Second is: the 431 in your regulator is sensing the base voltage of your darlington transistors, not the output voltage. This gives you some regulation but Vbe of the two transistors depends on their base current. So the voltage on the output will increase a bit when the load draws more current. You have seen that the negative voltage was more negative with load than without.
You could try to connect the 431 like this:
There are two things about your circuit that came to my mind:
First is: the circuit is a current source feeding a shunt regulator. It obviously cannot source more current than the current source does. Your current source is set to 0.125A. Be sure sure the circuit you are feeding is not trying to pull more than that. If you are testing the regulator with a resistor-load, it shouldhave more than , say 100 ohms. Not to forget the current that runs through the divider that gives FB to the 431.
Second is: the 431 in your regulator is sensing the base voltage of your darlington transistors, not the output voltage. This gives you some regulation but Vbe of the two transistors depends on their base current. So the voltage on the output will increase a bit when the load draws more current. You have seen that the negative voltage was more negative with load than without.
You could try to connect the 431 like this:
Really thanks, Grommeteer.
I will try to do it on tomorrow and let you know status.
naniking,
C95 & C96 should connect to the TL431 Cathode, not the Ref pin as shown.
Connecting them to Ref puts a large delay between the output and the TL431 feedback input. With the pot set to ~ 4k7, the time constant is (4k7//18k)*C95 = ?? (assuming 1uF for C95, C96 thats a delay of about 4ms). I'd expect this to make the TL431 oscillate, but slowly - around a few hundred Hz (set by the time delay).
Simply removing C95/C96 will prevent this closed-loop oscillation. But TL431s also can and do oscillate all by themselves - there is a range of Cathode capacitive loading that causes high-frequency oscillation. The TI datasheet has graphs of Cathode Current vs Load Capacitance, which shows the unstable regions for varying Cathode voltage (fig. 16, p.24). >= 10uF will stabilise the TL431 for any Cathode voltage/current combination. As V_cathode increases, the amount of load capacitance required to guarantee stability reduces.
But you need to make sure that this capacitor has reasonably low ESR. and by reasonably low I mean << 10 Ohms. I found this out the hard way with an smt circuit in 1998 - it was our first ever SMT design, and I was making an SMT version of an existing design. I used a 10uF smt electrolytic in place of a leaded 10uF OSCON, and the new circuit oscillated terribly. Turned out the crappy electrolytic I used had 27 Ohms (yep, 27R) of ESR. dreadful! the OSCON was << 1R. I tested the circuit with an OSCON and an external resistor, and IIRC found that 10R was enough to make it unstable.
Grommeters point is a very good one too -
C95 & C96 should connect to the TL431 Cathode, not the Ref pin as shown.
Connecting them to Ref puts a large delay between the output and the TL431 feedback input. With the pot set to ~ 4k7, the time constant is (4k7//18k)*C95 = ?? (assuming 1uF for C95, C96 thats a delay of about 4ms). I'd expect this to make the TL431 oscillate, but slowly - around a few hundred Hz (set by the time delay).
Simply removing C95/C96 will prevent this closed-loop oscillation. But TL431s also can and do oscillate all by themselves - there is a range of Cathode capacitive loading that causes high-frequency oscillation. The TI datasheet has graphs of Cathode Current vs Load Capacitance, which shows the unstable regions for varying Cathode voltage (fig. 16, p.24). >= 10uF will stabilise the TL431 for any Cathode voltage/current combination. As V_cathode increases, the amount of load capacitance required to guarantee stability reduces.
But you need to make sure that this capacitor has reasonably low ESR. and by reasonably low I mean << 10 Ohms. I found this out the hard way with an smt circuit in 1998 - it was our first ever SMT design, and I was making an SMT version of an existing design. I used a 10uF smt electrolytic in place of a leaded 10uF OSCON, and the new circuit oscillated terribly. Turned out the crappy electrolytic I used had 27 Ohms (yep, 27R) of ESR. dreadful! the OSCON was << 1R. I tested the circuit with an OSCON and an external resistor, and IIRC found that 10R was enough to make it unstable.
Grommeters point is a very good one too -
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.