[SOLVED] PMOS Regulator Soft Start

[Leethal]

I have been trying to design a linear regulator for the first time. The basic regulator appears to work regulating to 30v and the soft start seems ok at about 3 mS on LT Spice. When combined it appears to take around 400mS to come up to 30V. Can anyone with some experience in these circuits suggest why this is happening and perhaps a work around.

 

[crutschow]

I would suggest eliminating the separate circuit and do the soft-start in the voltage control stage. That way you eliminate one of the power MOSFETs and don't reduce the regulation accuracy by the soft-start MOSFET ON resistance .

Post your .asc file and I'll look at the simulation.

 

[Leethal]

Hi,

I was thinking that would be a better way to do it with just the one FET however I couldn't come up with a circuit that did both functions, regulate and soft start.

 

[crutschow]

There is an error in your circuit. You need a pull-down resistor from the M1 MOSFET gate to ground so the transistor can reliably turn on. The diodes at the op amp outputs prevent a reliable pull-down of the gate voltage, leaving that to leakage current. 10kΩ to ground should be fine.

You can connect a capacitor between Vref and ground to slow the startup. The startup time depends upon the capacitor value. 1µF gives about 15ms.

To see the startup delay in the simulation you need to check the "Start external DC supply voltages at 0V" option in the Simulation Command window.

 

[Leethal]

Thanks for that Zapper. The diodes are my attempt at oring with the vgs min supply to get around the VGS min of the Mosfet. Not sure how this is usually done but most seem to be +/- 15 or 20V and I'm working up to 36V.

 

[crutschow]

During further simulation I noticed some instability and oscillations (noticed particularly at U1's output) in your regulator circuit due to the added loop gain provided by M1. After some experimenting I made some modifications to stabilize the circuit and provide a soft start as shown in the attached circuit.

The voltage limit op amp and diodes were removed. Instead R1 and R2 are used to limit the maximum Vgs to 1/3 the supply voltage or 12V, well below the maximum Vgs rating.

Transistor M2 (can be any small P-MOSFET) was added to inhibit the startup spike at the output by keeping M1's Vgs initially near 0V, which keeps M1 off during the initial start-up transient. This, along with C1, provides the soft start.

Capacitors C3 and C4 were added to provide a stable loop response. (V2 was added for test purposes to provide a small step input at 100ms to check the transient response of the circuit). A smooth response at the op amp output with no ringing or significant overshoot indicates a reasonably stable loop response.

A 10µF output filter capacitor, C5 was added.

Note: I did not have a model for the voltage reference you used so I replaced it with a 4.7V zener and changed the value of R3 for simulation purposes.

View attachment Basic Voltage reg Mod.zip

 

[Leethal]

Thanks Crutschow. That is a lot more stability than I tried although I have just come across a document on simulation of LDO Regulators. I found I had to add a 22uF 1ohm esr COUT cap to get it stable. The 1 ohm was interesting as instability would appear if too low or too high. I was going to ask about the spike at start up so thanks for covering that also.
Is there any way of turning it on harder as when the input drops down to say 30V there is a voltage drop across M1 of 225mV. I did manage to have it lower ( 70mV) with the attached version 2 after following your earlier suggestions. In my LT Spice IV the reference is in the references folder (not that there is anything against the zener)

 

[crutschow]

If you want a lower drop across M1, you can use a MOSFET with a lower ON resistance.

I can't open a .rar file. Can you post it as a .zip file like the first one?

 

[Leethal]

Hi, It is the same device in both circuits.

 

[crutschow]

I updated my LTspice which had the LT1634-5 model. Due to the lower impedance of the LT1634-5 as compared to the Zener, the circuit again showed some instability. That was eliminated by adding a 2k ohm resistor in series from the junction of R5 and U2 to the (-) input of the op amp.

 

[Leethal]

So the zener was somehow keeping VG a little high therefore not allowing M1 to turn on fully whereas the voltage reference wasn't. Not sure I understand why however I now have a circuit I can work with. I just noticed the PMOS I selected only handles 1.5W. Ahhhhh. I will change to a SUP75P05-08 which should handle more than enough. Thanks again for your help.

 

[crutschow]

Note that with a 33V supply and a 7 ohm load the power dissipated in M1 is about 13W so it will need to be mounted on a decent heat sink.

To plot the power dissipation, place the cursor over the component, then press the Alt key and, when the thermometer appears, click the left mouse key.

- - - Updated - - -

So the zener was somehow keeping VG a little high therefore not allowing M1 to turn on fully whereas the voltage reference wasn't. Not sure I understand why .............................

The voltage set resistors with the zener regulator were not quite the optimum value to give 30V and therefore the regulated output voltage was set to slightly less then 30V. Thus the MOSFET was not fully turned on.

 

[Leethal]

Is there some way of adding output short circuit protection/current limiting utilizing M2 perhaps or adding futher components?
Actually, the problem I'm having is that the output is to a large NI PXI DC model's inrush current. It has a large inrush current which my Li-Ion batteries protection cct can't handle. They turn themselves off for a minute due to current requirements of the inrush which is maybe 20 amps. They can only handle maybe 5A. The ramp up of this cct doesn't appear to handle the inrush current well enough in practice . I think I need some further current limiting.

Any ideas on how I can improve the cct to handle this.