FAQ
TL;DR: Field-windings on small generators rarely exceed 2 A, so a 15 V shunt regulator only dumps about 30 W max; “the winding does not give more than 2 A” [Elektroda, żałosna udręka, post #20273420] A minimal TL431-optocoupler-MOSFET design improves survivability versus sealed OEM modules.
Why it matters: Knowing why each part is there lets you build a cooler, tougher regulator that survives a lost battery.
Quick Facts
• TL431 reference voltage: 2.495 V ±1% [Texas Instruments, Datasheet].
• Optocoupler CTR span: 50 – 150 % at 5 mA [Vishay, 2023].
• IRF640 RDS(on): 0.18 Ω at 10 V GS; SOA 3 A at 25 V, 1 s [Infineon, Datasheet].
• Typical generator auxiliary winding: 18–22 VAC no-load, 2 A max [Elektroda, żałosna udręka, post #20270112]
• Zener surge current (1N4742A 12 V): 200 mA for 1 µs [ON Semi, Datasheet].
What job does the Zener diode actually perform in this regulator?
It forms a coarse 11.8 V reference so only a few milliamps flow through the divider when the engine idles; the TL431 then adds 2.5 V for a final 14.3 V set-point
[Elektroda, żałosna udręka, post #20270112] Removing it saves parts but increases leakage and makes the TL431 handle all surge energy.
Why combine a Zener and a TL431—couldn’t I use one or the other?
Using both lets you keep divider resistors high (lower standby loss) yet maintain a tight ±1 % regulation band from the TL431. A lone Zener drifts ±5 % over temperature, while a TL431 alone would need lower-value resistors and waste ≈15 mA continuously [Texas Instruments, Datasheet].
Two transistors drive the MOSFET gate—necessary or overkill?
Why was a 200 V IRF640 chosen for a 15 V system?
Does the big 33 µF capacitor hurt stability?
Is heat really a problem in a parallel (shunt) regulator?
Yes—at 2 A and 15 V the MOSFET can dissipate 30 W continuous. Without adequate heatsinking junction temperature will exceed 150 °C in less than 45 s, cutting lifetime by 50 % for every 10 °C rise [Infineon, Datasheet].
What happens if the battery disconnects while running?
OEM modules often fail because the stator no-load voltage can exceed 100 Vpk; the proposed design clamps the gate with a 33 V Transil and the MOSFET absorbs the energy within its SOA, preventing catastrophic over-voltage
[Elektroda, satanistik, post #20270819]
Can moisture or dust shift the divider voltage?
How do I add automatic shut-off when AC is absent?
Insert a second optocoupler that disconnects the TL431 divider until the auxiliary winding exceeds about 6 VAC. This drops standby current to zero and prevents battery drain
[Elektroda, Mlody_Zdolny, post #20270207]
Could a linear series regulator beat this shunt design?
Series regulation wastes less heat but needs a low-impedance, higher-current pass device and a feedback supply; complexity and parts count rise. For <2 A field coils, a shunt MOSFET remains cheaper and simpler
[Elektroda, CosteC, post #20270518]
Edge case: what if the MOSFET shorts?
A failed-short MOSFET ties the winding directly to ground, stalling charge at ≈0 V. The generator keeps running, but the battery slowly discharges—a failure you’ll notice before damage spreads. Adding a 15 A fuse on the winding minimises risk.
How can I retrofit this regulator in three quick steps?
- Unsolder and remove the potted OEM module.
- Bolt the MOSFET to the existing heatsink with thermal paste.
- Wire the TL431 board to the winding, battery sense, and ground; verify 14.3 V at 3,000 rpm.
Done—average install takes 20 minutes.
Generated by the language model.
IMG_20221105_101758.jpg
(2.37 MB)
You must be logged in to download this attachment.
IMG_20221105_175550.jpg
(2.33 MB)
You must be logged in to download this attachment.
Comments
What exactly is this zener diode for? It just ruins the stability of the 431. To limit the current, you could have added larger resistors or a second optocoupler to disconnect the voltage divider when... [Read more]
What exactly is this transformer for? I would also remove the Zener diode from the TL divider (I would protect the MOS gate with it, but with the value of C15) replacing it with a selected resistor, and... [Read more]
I'll throw in: Why two transistors (Motorola transistors) to drive the MOSFET gate? This is not a converter where we want to switch a transistor in 50 ns, but a very slow linear application. The reaction... [Read more]
As far as I know, analog circuits have outlived many converters. As for the performance being bad, that's another story. Cooling needs to be calculated, choose the right radiator and determine the... [Read more]
The fact that a high-voltage transistor and a Zener diode in the gate is very good, because in such a system, for example, when the battery breaks, overvoltages may appear. A system with the ability to... [Read more]
Nothing will help. The battery will break, the generator will give high voltage, the transistor will take it on itself and start to heat up. High SOA is needed, not high voltage. And the gate can be secured,... [Read more]
And the impulse ones that give "several dozen" amperes do not have fans? Not completely. Here the fact is that the weight and dimensions are large, but believe me, you can build a good linear power... [Read more]
If there is such a need (and the customer is able to pay for it), they do not have it. Quick proof based on the contents of a drawer in the garage https://obrazki.elektroda.pl/1771377700_1667748098_thumb.jpg... [Read more]
But the dimensions also increase x2 with a power of only 250W. [Read more]
More often they don't have than they have :) You give @spec220 some real 250W 12V linear power supply on the hit. Data + photo please. Please show the forum what outstanding constructions I have overlooked... [Read more]
Just take into account the fact that the TL at full control (exceeding Vref) does not short-circuit like the transistor, there is only a KA drop of about 1.8 V, add to this a drop on the MOS gate divider. ... [Read more]
Let me criticize something. The capacitor loads the zener diode too much. However, I prefer Zener over resistors, because at high values, dust and moisture will affect the voltage of the divider. Why IRF640?... [Read more]
So if you prefer Zener, why do you need this TL431? In fact, it will be slightly better, but only "slightly". I don't understand this at all. You mean "kick" when power is turned on? This gives... [Read more]
That's how it should be. [Read more]
Ladies, do you read with comprehension? ;-) I gave the example of the power supply as proof that it is possible to make a system that dissipates a lot of heat and is reliable at the same time, and not... [Read more]
It will turn off without any problem in reverse, because the MOS works as a diode, while in the direction of conduction, if it is sufficiently driven, it will also turn off, because the channel is open.... [Read more]
The winding does not give more than 2 A, so there will be no heating. The original regulators died when the battery was disconnected. This one will be fine. [Read more]
This capacitor introduces a slight delay, but the zener diode also gets a slight kick when sparking. Some kind of limiting resistor would be useful there. [Read more]
In practice, it turned out that the capacitor is unnecessary, because it is one winding, so alternating current probably with 300 Hz, so why slow it down? (on the table I played from a linear regulated... [Read more]