# 370 V, 400 mA stablilized power supply with variable load

I need a power supply that can deliver stable 370 V DC voltage, while load resistance is variable.

Average load current(assuming stable 370 V supply) is 200 mA, however it may rise to 400 mA. Input voltage(DC) is 40 - 60 V higher than 370 V, but it's unstable.

I was expecting that I would simply buy something like LM317, but rated for 370V and 400mA. It quickly turned out that nobody sells such things. Closest to what I need was this stabilizer: Microchip LR8K4-G, 1.2 → 440 V Linear Voltage Regulator, 30mA, 1-Channel, Adjustable, ±5%.
It meets my voltage requirements, but it can only handle 30 mA.

This fragment from its datasheet caught my attention:

The output voltage can be adjusted by means of two external resistors, R1 and R2, as shown in the typical application circuits. LR8 regulates the voltage difference between VOUT and ADJ pins to a nominal value of 1.20V. The 1.20V is amplified by the external resistor ratio R1 and R2.

So if we can just amplify the voltage difference, then perhaps there is a way to use lower voltage stabilizers for higher voltage? Can somebody point me in right direction?

simulate this circuit – Schematic created using CircuitLab

On the one hand, the Q1 VBE is not well defined, but on the other, any error is negligible compared to the errors there will be in setting the LR8 output voltage.

This arrangement is not current limited, other than Q1 beta times LR8 limit, but it would be straightforward to add a few more components to make it so

• Technically, you can elimintate V_BE's effect by tying the top of R1 to Q1-E instead of the LR8's OUT. However this may introduce some instability. A 0.1uF cap across Q1 B-E would help, plus another resistor from LR8 OUT directly to ground (to compensate for the DC current draw change having removed R1's connection). May 10, 2020 at 21:26
• @Atomique That's a common misconception, it works for ground referenced regulators, but not for output referenced regulators. Draw a schematic of what you suggest, and calculate what happens, or simulate it. May 11, 2020 at 8:43
• Neil_UK You are absolutely right. Was not thinking. (The problem is the V_BE of the transistor now interferes with the internal (V_OUT - V_ADJ)=V_REF that the regulator "seeks"). Thank you. May 12, 2020 at 22:35

In a bit (but not very much) of simplification, a linear regulator like the LM317 is nothing but a differential amplifier, operated in negative feedback, trying to get the voltage on its inverting input as close as possible to that on its non-inverting one.

In fact, that's exactly what the TI LM317 datasheet shows:

Instead of the Darlington pair, one would probably use a N-Channel MOSFET in this day and age.

From the theoretical point of view you could use another, higher output current capability, three terminal regulator, paying attention to keep the dropout voltage $$\V_\mathrm{drop}=V_{in}-V_\mathrm{out}\$$ within the specified datasheet limit: however, you will surely incur in the following two problems

1. Three terminal regulators with $$\V_\mathrm{drop_\max}\ge 60\mathrm{V}\$$ are not easy to find, especially with the output current you require: apart from the difficulties of having a technological process which allows the joint integration of high voltage and low voltage transistors (now largely superseded), a problem is the high power dissipation needed by such kind of IC.
2. However, even if devices of such kind exists (e.g. the LM317HV), your input voltage range of $$\(410\div 440)\mathrm{V}\$$ is such that a minimal input spike and/or an accidental short circuit at the output will burn your voltage stabilizer almost immediately.

Therefore my advice is not to use a lower maximum dropout voltage three terminal voltage regulator IC in such an application.

However, You can still try to use the Microchip LR8K4-G in a current boost configuration jointly with a pnp transistor. A tentative circuit could be the following one:

simulate this circuit – Schematic created using CircuitLab

This was a circuit that was once proposed in the older datasheet of the 78XX series in order to overcome their former current output limitation.

• The current flowing in $$\R_{bb}\$$ is controlled by the LR8 IC, which in such a way controls the pnp BJT $$\Q_1\$$ via its $$\V_{BE_{Q_1}}\simeq V_{R_{bb}}\$$.
• The components $$\R_E\$$, $$\R_B\$$ and $$\C_\mathrm{comp}\$$ have to be bench optimized in order to avoid instabilities (read oscillations) of the circuit: and apart from lowering the overall stage gain, $$\R_E\$$ could also be used, in conjunction with other circuitry, to implement a short circuit protection.