Replacing noisy Zener in audio PS with TL431

[slashdev]

I'm repairing a Pioneer RT-707 tape recorder, and have tracked down a recording problem to a noisy 7V Zener (7.2V actually) in the REC/ERASE power supply.  I considered buying a replacement, but because it would almost certainly have a different voltage, I would have to futz around to get the same Vout (30.94V) or adjust everything downstream (i.e., a factory recalibration on that section).  So...

I was thinking about using a TL431 "adjustable zener" as a drop-in replacement.  I don't see many projects in the audio forums about this, so I thought I'd get some EEVBlog opinions.  Here's the existing circuit:

  (subsection of )

At first, I was focused on just replacing the Zener.  Then somebody happened to post a thermal image of this circuit here, and both R332 and R334 are running at about 50C.  I wondered if the TL431 should replace the Q308 transistor and the toasty resistors?  After some simulations, I have this as a possible replacement:

• TL431 provides a ~31.6V reference for Q307, instead of a 7.2V reference to Q308
• R2 is a pot for dialing in the desired 30.94V output.
• R334 deleted
• R336 increased by 50% (cooler!)
• C323 and C324 deleted because the TL431 doesn't like (or doesn't need) a capacitive load.  R335 keeps TL431 stable WRT C325
• C325 reduced by 70%.  This decreases startup time from 3s to <1s, and gives me a little more room.  Things are a little cramped between the big caps and the heat sink for the transistors.

The ripple is reduced quite a bit, and I think the noise is 4x lower.  This seems close to the "Figure 34. Efficient 5-V Precision Regulator" in the TL431 applications section, but what do I know?  Umm... enough to find the problem, but not enough to solve it? 

Any suggestions or comments?

Thanks
/dev

[Edit - removed R1]

[Seekonk]

I have a fondness for the 431.   Without doing any calculations I suspect the voltage is way too high to trust to a low quality import and it would be running fairly hot.  Suspect those resistors were chosen to place a fairly high load on the supply. I'd keep it as designed with the 431 replacing the zener.

[Zero999]

Yes, if you're using the TL431 directly, it's possible it's too near the maximum voltage rating or power dissipation.

How about configuring it as attached which will preserve the low voltage drift of the TL431.

[slashdev]

I suspect the voltage is way too high to trust to a low quality import

Rats.

... and it would be running fairly hot.

Well, the spec says (...if I had a nickel...) at least 1mA should pass through for good regulation, so I set R336+R337 to give 1.6mA.  At 31.6V, it's dissipating ~50mW.  With R_?JA = 140 °C/W, it should run 7°C above ambient. (Did I do that right?)

Suspect those resistors were chosen to place a fairly high load on the supply

Oh?   The whole board uses a total of "only" 100mA, measured at R337.   I thought (oops!) R334 was to get enough current to be sure the Zener was at a good knee (7mA, spec'ed @ 10mA), and R332+R333 was to have a good divider for Q308 (base current << 7mA divider current?).  But... 

Thanks!

[slashdev]

How about configuring it as attached...

Sorry, is that with the TL431 set for the original 7.2V?

[Pjotr]

This is the classical "Two transistor" regulator and performs very well for audio. TL431 is certainly noisier than a good zener. Simply replace the bad zener for a new good 6.8V one and trim R333 for the correct output voltage by using a large parallel resistor (something between 22K and 100K I'll guess) when needed. B.t.w. 50 degC is not particularly hot for a resistor.

[IconicPCB]

As Pjotr says this is a regulator and the function of Q308 is to act as ERROR amplifier.

Q308 base voltage is derived from the output voltage and compaerd to the Zener voltage within the base emiter junction of Q308.

Keep the current density in the divider network so that the base impedance remains the same, low.

Looking at the base drive it is supplied by an equivalent voltage source in series with 825 ohm.

You might consider increasing the capacitor in parallel with the zener although it might cause the error amp to start oscillating.

DO NOT DECOUPLE THE BASE DRIVE it will cause poorer regulation and instabilities.

What is Q308?

 

[T3sl4co1l]

TL431 is noisier than a proper zener diode.  Replace it with a 6.8V device.

Tim

[slashdev]

TL431 is certainly noisier than a good zener. Simply replace the bad zener for a new good 6.8V one and trim R333

Easy-peasy!  Is there a second?

TL431 is noisier than a proper zener diode.  Replace it with a 6.8V device.

All in favor, so moved.   

What is Q308?

Q308 was a 2SC945, which I replaced with a KSC1845.  One other 2SC945 had gone bad, so I tried replacing this one.
Q307 is an original 2SD526.

DO NOT DECOUPLE THE BASE DRIVE it will cause poorer regulation and instabilities.

Are you referring to C324 in the mfr's original circuit?  Are you suggesting I remove it?  Isn't it a secondary smoothing of the rectified 44V?

Thanks everyone!  Your descriptions help immensely, and they point me to related topics.

[Seekonk]

Hard to resist just tearing out everything.  This regulator is at an awkward voltage to give nice simple options.  I happen to have a roll of 5,000 18V zeners that seem to work their way into everything.  If you placed about this voltage zener in series with the 431 (transistor replacement circuit) issues of voltage and dissipation would be eliminated.  Zener noise is relative.  A little zener noise isn't an issue when compared to a bad zener where it can be measured with a DVM.

[IconicPCB]

Apropos the decoupling I refer to the BASE drive tis means base to ground not power supply filtering.

[T3sl4co1l]

I see what you did there