Depends.
I imagine that regulation with this combo of current source + shunt regulator may actually be quite good.
Noise, probably nothing too special. The TL431 uses a bandgap reference, much like LM317, 78xx etc. Actually, both have been compared in parts 2 and 3 here. Best-case noise is somewhat lower for the shunt regulator... 6-12 dB tops, which is something, I guess.
Efficiency is going to be terrible since the current source literally has to be sized for (more than) peak load current, with the rest being shunted away to ground. In this case, that current is about 120 mA. That's a total circuit dissipation of close to 5 W, all the time... and you may be getting close to current limiting (a 4556A can churn out about 80 mA peak per channel, though on average you'd need substantially less).
Shunt regulators are useful for circuits with very little current draw (think a few mA tops... maybe a phono preamp or oscillator), as they allow transistors to be used at sensible amounts of current that give good bandwidth and low noise. This has a big effect on regulator performance.
I think if you get the LM3x7 feedback network to draw ~12 mA by itself (i.e. 220R in top) and the opamp draws another 6-9 mA, that should get the regs into a pretty decent zone. Ca. 800 mW of idle power dissipation sounds a little more acceptable. Once you get into circuits including a discrete Class A buffer this pretty much becomes a non-issue anyway - and I'd much rather have one of those than a fancy power supply with a barefoot opamp.
Realistically speaking, you really don't need more than a 78M/79M pair for an opamp-based headphone amplifier - the O2 got by like that, too. The risk of a noisy supply isn't so much running out of circuit PSRR but rather spoiling effective PSRR by injecting noise into the ground via decoupling caps if signal and power ground share a nonzero return resistance somewhere, which is a function of layout.
In hi-fi equipment, you will often see RC power filtering on daughterboards with shared ground connections (or alternatively, just minimal amounts of capacitance to limit the effect to the ultrasonic range). Since the R forms a voltage divider with common ground return impedance and tends to be much larger than "barefoot" regulator output impedance, noise injection is reduced correspondingly. A few dozen ohms may not seem like much but compared to <1 ohm it certainly is.
In some classic phono preamp designs, you can also see how power comes in at the output, with big capacitors near the DC connection where return resistance is small, and then gets progressively RC filtered while making its way alongside the ground towards the more critical low-level stages.
It's all much less of a problem if you can keep your ground returns nice and low in impedance and/or separate for signal and power.
I imagine that regulation with this combo of current source + shunt regulator may actually be quite good.
Noise, probably nothing too special. The TL431 uses a bandgap reference, much like LM317, 78xx etc. Actually, both have been compared in parts 2 and 3 here. Best-case noise is somewhat lower for the shunt regulator... 6-12 dB tops, which is something, I guess.
Efficiency is going to be terrible since the current source literally has to be sized for (more than) peak load current, with the rest being shunted away to ground. In this case, that current is about 120 mA. That's a total circuit dissipation of close to 5 W, all the time... and you may be getting close to current limiting (a 4556A can churn out about 80 mA peak per channel, though on average you'd need substantially less).
Shunt regulators are useful for circuits with very little current draw (think a few mA tops... maybe a phono preamp or oscillator), as they allow transistors to be used at sensible amounts of current that give good bandwidth and low noise. This has a big effect on regulator performance.
I think if you get the LM3x7 feedback network to draw ~12 mA by itself (i.e. 220R in top) and the opamp draws another 6-9 mA, that should get the regs into a pretty decent zone. Ca. 800 mW of idle power dissipation sounds a little more acceptable. Once you get into circuits including a discrete Class A buffer this pretty much becomes a non-issue anyway - and I'd much rather have one of those than a fancy power supply with a barefoot opamp.
Realistically speaking, you really don't need more than a 78M/79M pair for an opamp-based headphone amplifier - the O2 got by like that, too. The risk of a noisy supply isn't so much running out of circuit PSRR but rather spoiling effective PSRR by injecting noise into the ground via decoupling caps if signal and power ground share a nonzero return resistance somewhere, which is a function of layout.
In hi-fi equipment, you will often see RC power filtering on daughterboards with shared ground connections (or alternatively, just minimal amounts of capacitance to limit the effect to the ultrasonic range). Since the R forms a voltage divider with common ground return impedance and tends to be much larger than "barefoot" regulator output impedance, noise injection is reduced correspondingly. A few dozen ohms may not seem like much but compared to <1 ohm it certainly is.
In some classic phono preamp designs, you can also see how power comes in at the output, with big capacitors near the DC connection where return resistance is small, and then gets progressively RC filtered while making its way alongside the ground towards the more critical low-level stages.
It's all much less of a problem if you can keep your ground returns nice and low in impedance and/or separate for signal and power.
Last edited:
Thank you for your input I am learning much more from answers like this. My plan was to use lm 317 regulator but others think its outdated and ancient. I have ordered grado ra-1 clone board that I can use with 2 x 9v batteries and have good number of objective2 boards. I first want to try grado board and after that i will build o2.
I started to learn about serial vs shunt regulators last few days and I understand basics but still cant say what good shunt regulator have to look like. Also I have found from other people that tl431 can easily make problems like overheat or oscillate and I wouldnt know how to fix that.
Also as far i can tell there is no active groupbuy for psu that suits me so my best bet is ebay.
I started to learn about serial vs shunt regulators last few days and I understand basics but still cant say what good shunt regulator have to look like. Also I have found from other people that tl431 can easily make problems like overheat or oscillate and I wouldnt know how to fix that.
Also as far i can tell there is no active groupbuy for psu that suits me so my best bet is ebay.
Last edited:
Hi,
If you are not afraid of soldering SMD parts, I offer PCBs for a PSU based on TPS7A4901/TPS7A3001 low noise LDO regs.
Regards,
Oleg
If you are not afraid of soldering SMD parts, I offer PCBs for a PSU based on TPS7A4901/TPS7A3001 low noise LDO regs.
Regards,
Oleg
Last edited:
Have you seen Muffsy Hifi Dual Power Supply v4? It’s actually a dual + aux. Have not tried it myself but also looking for a PSU, and it looks good.
I thought you were asking for dual LM317/LM337 regulator boards. Muffsy PSU provides ±15VDC @ 100-200mA regulated and 15V not regulated. If you dont like the voltages you can easily change them as well (see how to change the regulated output here or the aux output here)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.