The path to the ultimate LM317 is well trodden with the VRDN, the D-Noizator, etc. Inspired by Pa's F5m I'm going
to ponder and post a minimum LM317 that has maximum performance with a minimum number of parts. Perhaps ten?
Quick notes on using 3-Pin Voltage regulators
Feel free to offer your thoughts.
Regards,
Dan
to ponder and post a minimum LM317 that has maximum performance with a minimum number of parts. Perhaps ten?
Quick notes on using 3-Pin Voltage regulators
Feel free to offer your thoughts.
Regards,
Dan
The low-parts-count implementations work great.
CRCRC the raw supply and the regulator will have barely any noise to deal with. The best improvements I've gotten have been with better passive filtration.
CRCRC the raw supply and the regulator will have barely any noise to deal with. The best improvements I've gotten have been with better passive filtration.
Diodes D1 and D2 are only needed for high input and output voltages to prevent reverse breakdown if the input or output is suddenly shorted.
12V out is "high"
12V out is "high"
Say for a preamp, or phone stage, or similar element with a low current draw you can use R=10 ohms, and your reservoir caps have an ESR around 0.1 ohms. The maximum attenuation you will get from the RC element is 40dB - a really useful amount ahead of a regulator, or for many other purposes.
D1 and D2 are recommended to protect the regulator from the effects of stored charge in C1 and C2 at switch-off,
Not only recommended. They're likely required.
You don't have to use C2. Its purpose in life is to reduce noise, which is nice, but not always required. If you skip C2 you can eliminate D2 as well.
A uA7212 will make 12 V as well. It won't be as low noise as the LM317 circuit shown above but low noise is not always required. You can make a 7812 run with a grand total of three (3) parts.
Tom
All other things being equal, popular opinion suggests skipping the 7812 for the 317 for lower noise for the same price.
Pondering a simple CRC feed to the LM317 as a good low parts compromise.
Pondering a simple CRC feed to the LM317 as a good low parts compromise.
LM 317 circuit with external protection diodes
When there is capacitance present in a circuit driven by a voltage regulator such as an LM317, either directly on the output of the regulator or within the circuit being driven, it is often wise to add protection diodes to prevent the capacitors from discharging through low current points into the regulator.
Protection diodes are recommended for output voltages in excess of 25 V or when high values of capacitance appear on the output of the regulator. It is often recommended that for capacitance values in excess 25 µF and when the values for CAdj exceed 10 µF).
In the circuit the diode D1 prevents the overall capacitance on the output from discharging through the IC during an input short circuit. The output capacitance consists of COut plus any capacitance in the circuit being powered. Although an input short circuit may appear unlikely, the path through any input circuitry may appear as a short circuit under some conditions.
Also the diode D2 protects against the capacitor CAdjj discharging through the IC during an output short circuit.
Finally, the combination of both diodes prevent CAdj from discharging through the IC during an input short circuit.
The diodes, D1 and D2 need not be any special form of diode - diodes such as the 1N4002 have sufficient current capability and reverse breakdown capability to provide excellent protection. 1N4002s are also very cheap and widely available.
Without C2, with a big capacitor at the input, with nothing but the LM317 loading that capacitor, with nothing forcing current into the output and without sudden short circuits at the input, you can do without the diodes. The input voltage then comes up faster at start-up and drops slower at power-down than the output voltage, so the voltage difference always has the right sign.
Anyway, make sure to use a sufficiently lossy decoupling capacitor at the output. A run-of-the-mill electrolytic capacitor works fine, a ceramic capacitor can cause oscillations in some brands of LM317.
Anyway, make sure to use a sufficiently lossy decoupling capacitor at the output. A run-of-the-mill electrolytic capacitor works fine, a ceramic capacitor can cause oscillations in some brands of LM317.
I once had a circuit that sometimes worked fine and sometimes produced an annoying high-pitched beep. It turned out to be due to me using only X5R ceramic decoupling capacitors at the output of an LM317; the LM317 was at the edge between oscillating and not oscillating, and when it oscillated, the X5R capacitors worked as piezotweeters. I connected an ordinary aluminium electrolytic capacitor across the output and since then, it's always been silent.
One thing that's a bit peculiar about the LM317 datasheet is that recommended application circuits typically have R1 = 240 ohm while the minimum recommended load current is 10 mA. With R1 = 120 ohm, the minimum load current requirement is met even with open output, with 240 ohm it is not.
One thing that's a bit peculiar about the LM317 datasheet is that recommended application circuits typically have R1 = 240 ohm while the minimum recommended load current is 10 mA. With R1 = 120 ohm, the minimum load current requirement is met even with open output, with 240 ohm it is not.
Yeah. But all things aren't equal. OP asked for a low parts count and the LM317 uses more parts than the 7812.
I do realize that the LM317 offers lower noise than the 7812 but that may not matter in the application. I see no need for an LM317 to regulate the supply for a string of relays for example.
If one truly wants lower noise, I suggest looking at the more modern regulators such as the TPS7Axxxx. from TI.
Tom
Even a 7812 can fail without D1 if there is a large output capacitor and the input can be shorted. How this could happen is for your risk analysis.
Any value of CAdj in excess of a few uF will interfere with the regulator’s transient response.
COut needs some ESR for stability. Tants not recommended
True. Which is why I just add the diode anyway. Diodes are cheap. Failures are expensive. But OP asked for a low parts count...
Tom
And it's a problem with LM337 too. I ran into that problem and switching to a cheap high ESR electrolytic solved the problem.
Could you say, why "minimum number of parts" is an important and righteous goal? You're going to spend 96 hours building the gear, followed by several years listening to it. In what way is the total experience (entertainment) improved by reducing the parts count?
Replacing the entire PSU by two lead acid batteries certainly minimizes parts count -- is that the microphone-drop, thread ending, obvious best?
Replacing the entire PSU by two lead acid batteries certainly minimizes parts count -- is that the microphone-drop, thread ending, obvious best?
Everest I think. Same as speed running a game. All pointless persuits to me, but if it floats your boat...
It beats me why anyone would want to use a LM317/337 regulator when newer, better spec'd, linear, through hole mountong devices exist:
Eg LT 1963a (1.5a +ve) and LT 3015 1.5a -ve. low dopout, v.low noise, low quiescent current.
Both are available in either fixed voltage or adjustable voltage and no protection diodes needed as they built in, so ideal for 'minimum component count designs'
Eg LT 1963a (1.5a +ve) and LT 3015 1.5a -ve. low dopout, v.low noise, low quiescent current.
Both are available in either fixed voltage or adjustable voltage and no protection diodes needed as they built in, so ideal for 'minimum component count designs'