DIYHiFi.org

Mmmm... I'm here wondering if HB's schematic wouldn't benefit from using a dual op-amp. The output of the first half goes to the input of the second half (passing through another RC filter), cascaded.
Dual regulation, but still tight layout.

Hi HB,

Well, as long as GND+2.5V is well within the common mode range of the op-amp when it's run from a single supply, the TL431 could have been configured as a simple 2.5V reference and the FB resistor R5 made 12K.

Another thing, R7 could then be increased to 24K with a 28dB improvement in regulation, as TL431 will hold it's output Z down to < 0.5mA.

Cheers,
Greg

@Carlos:
Would adding a LPF plus a second OP as a buffer for the filter do anything else then slowing down the load regulation?

@Greg:
2,5V would be in the common mode range of the OP177, which ist about 2V above the negative rail. But this is close, and I just feel better getting further away from the rail. Adding two resistors an a cap at a few cents doesn't bother me. Any technical issues against doin so?
I agree on the value on the shunt resistor. I'll increase its value.

Hi HB,

OK - Keeping the TL431 at 5V you can increase the R's in the ref bias chain from 1K -> 10K as they draw current from R7 and, from memory the ref only needs < 20uA. R7 would need to be < 15k.

Cheers,
greg

Isn't there any advantage in terms of noise when the voltage divider Rs have lower values?

Hi HB,

I don't believe so. Besides the source impedance of the divider is bypassed with big C's.

Cheers,
Greg

I mean a small resistor and cap.
Probably not an issue, and you could use a faster opamp, but Elso doesn't like that.
I understand the reasons why you've chosen the OP177, CMRR is impressive at low frequencies, but the OPA627 is not bad at all either, and it's better at high frequencies. PSRR is better on the OPA627.
I use the OPA627 on Jung regs. It's not a dual opamp, though.

I don't believe that R1 R2 are there for current limiting they are to drop the emitter (and therefor also for base) voltage low enough that the op amp can drive the pass transistor. Look on the op amp data sheet for maximum othput swing with respect to positive rail voltage. It is usually around a couple of volts for low output currents from the op amp I think.

What transient current? The pass transistor is limited to sinking the current supplied by the current source in one suddenly removed the load. If one shorts across the shunt pass transistor. The base to emmitter voltage decreases and the pass transistor decreases amount of current through it. The are really no requirements for series resistors. If there were some need for current limiting for some scenerio... put a resistor between the collector and ground where it doesn't increases the output impedance.


A: Sorry, the five minutes is up.
M: That was never five minutes!

There's another reason I have chosen the OP177 and not OPA627; 1,50 Euros vs. 20 Euros...
I have some samples of TI's OPA277 here, another modern precision OP. Might be an alternative.

Exactly.

The story so far. Any further suggestions?

Yeah, why no resistor on the base of Q1? Right now you are prone to HF oscillation due to the gyration of Q1 near ft. If the opamp responds in that frequency range it could spell instability trouble. Hmmm, OPA177 is really slow like an OP07, right? Then you might be ok. Still it can't hurt to put a little protection in there.

Does TL431 really need a cap on it? Been a long time since I've used one of these. I think C3 will do all of the filtering work you need. C4 may not be needed. It too, might be a stability issue.

Potentiometer should have some series resistance, to prevent wiper from going to zero ohms (forcing open loop condition on regulator). Also be careful of the input section on opamp. Is it diode protected? What is common mode range? Better yet, what is differential range allowed? Just want to be sure it can handle any voltage transients via C1. Also has to survive power up and down conditions without latching opamp.

If you are going to the trouble to make such a fine discrete regulator, why not do the same for the other half of the circuit - the CCS?

Does this regulator start up? And how does it respond to a short circuit? I see the reference is powered by the load, hence it starts at zero volts. As the output voltage ramps up to about 3V, the opamp should start coming alive, but not the reference. And maybe then the opamp responds by forcing the output voltage as low as it can, trying to regulate to the 0V reference. Have you built this? Without digging too deep or putting much thought into it, seems like maybe it will clamp itself low and never turn on.

I would suggest finding a way to power both the reference and the opamp from something other than the output. Or at least give it a way to start up.

jh

Agreed.
It does. It is indeed a stability issue.Maybe, but I just feel better with seperating decoupling from the LPF.Agreed.It is.From the datasheet: "For supply voltages less than ±22 V, the absolute maximum input voltage is equal to the supply voltage."I basically agree, and I admit I didn't try other CCS topologies. I have done a lot of reading on this issue, and found several measurements proving the LM317 isn't hat bad in this application. At least, it seems to be better then the standard one or to transistor CCS. Going the OP regulated route as with the shunt reg would be an alternative, but then I would most probably lose the pure "two terminal" nature of the circuit, which makes wiring the hole thing as a positive or negative reg very easy. Trying the depletion mode MosFets Fred proposed might be worth a try, haven't looked deeply into it yet.It does start up and doesn't have any problems with shorts, a short is nothing but switching the shunt reg off (supply voltage goes zero).I did, see picture on last page. At this time it is powering a phono preamplifier, set to +/-20V, powered from a massive CLCLCLC-filtered +/-30V transformer.Might be an alternative, but this would also cost it's two terminal nature.Some safety concerning this is a good idea. Any suggestions?

Thanky you Jim for you input!

The filter cap for the reference is being charged by the bias and feedback resistors of the voltage reference as the feedback resistor bypass cap cap for the op omp is being charged during startup. As long as positive input is more positive than the negative input during start up the supply should come up. One can adjust the RC time constants to make sure this is occuring.

A couple of 100 resistor in series with the inputs to the op amp would be a good idea to limit current from caps discharging. Check the data sheet but most op amps have clamp diodes rcross the inputs internal to the part. IF not add them and the resistor to limit current through them when clampimg.

Stabilty concerns for the output transistor can be met with a 5 to 10 ohm resistor on its base. Powering the op amp and reference from regulated output is a performance advantage that should not be changed unless absolutely neccesary. Good design reveiw and things that should be considered for reliable design.




Depletion mode mosfet... DN2540 I feel like that duck on the commercial yelling AFLAC!

150 mA Idss is a little low for this application, the LM317 CCS is set to 460 mA. DN2470 has 500mA, still not too much. Cascoding with something more powerful might be an idea.
Updated schematic (still with LM317 CCS):