Power Supply Voltage Regulator Problem

[jackand08]

Hi Everyone,

I have designed and built a basic power supply for a speaker system project of mine but am having trouble in finding the cause of this problem.

It is set up as most PSU's are as a dual polarity supply, but with the main power rails (around +/-35V from a 25V dual secondary transformer) regulated down to +/-15V (using L78/7915 regs) for the input circuitry. There are also dropper resistors to reduce the voltage going to the regulators to something that is in their input voltage range.

The positive reg works fine, regulating whatever voltage is input into it down to +15V all the time, whilst the negative regulator will only drop some of the voltage (say 35V down to only around 23V), regardless of whether the input voltage is in or out of its range.

For example, I connect the board up to my lab supply and raise the voltage up past +/-15V to something that is comfortably in the range of the input voltage of the regulators. The positive reg does its job and drops it down to +15V whilst the negative reg never produces -15V, and its output voltage only rises when the main rail voltage rises.

I have checked the layout and there is nothing wrong there, and all the regs have bypass capacitors right next to them, so I am wondering what is wrong.

Below I have uploaded a (very much) simplified version of how everything is connected.

[Elvee]

Most monolithic voltage regulators have their regulation characteristics specified between a min and max current.
In this case, the min current is 5mA, which means that below this value the regulation is unspecified.
It doesn't mean all regulators will cease to function below 5mA, but some might, and the 79xx series is more prone to problems at high input voltages and/or low loads than others.
The solution is to use a bleeding resistor <3K at the output.

BTW, using dropping resistors at the input is a very bad idea: they will not reduce the input voltage at low currents, and regulators hate seeing a high impedance at their input. It is an excellent recipe for troubles, like unwanted oscillations.

[PullaEmu]

Maybe you have killed the 7915 with overvoltage? Datasheet lists -35V as absolute maximum for 7915, which means -36V even for a fraction of a second might cause permament damage, even at no current through chip. If you have left the PSU unloaded, the resistor doesn't drop voltage for chip (Voltage over resistor is current times resistance, so at I = 0A U = 0V). Also some trafos give a bit higher than rated voltage when unloaded, 10 % is not uncommon but well enough to push regulator over limit (resistance of wiring eats some voltage, due to same law of U = I * R, voltage rating is often given with rated load). A bit unrelated, but 20mF of filtering capacitance also seems awful high: assuming you're running off 50Hz mains, you get unregulated peak voltage once every 10 ms (twice per cycle). As dU = I/C, assuming I is maximum rated to 7yxx (1,5A) capacitors are going have voltage drop of dU * dt = I/C * dt = (1,5A / 20mF) * 10ms = 0,75 V. While low voltage ripple has obviously been a design goal, having huge input filter capacitors isn't the way to go with it: As an exercise left to reader, calculate how long (in milliseconds) sine wave-formed voltage of transformer is within top 0,75V of cycle, and assuming constant current throughout that period, calculate current going through transformer into capacitors (hint: you can use time current is coming from capacitors and compare that to time current is coming). To get better regulation of voltage, maybe you could consider cascading regulators (21V feeding 15V), but I've never used cascaded regulation so don't take my word for it.

[Osvaldo de Banfield]

It is better to add a zener in each input of the regulator, of a slowly higher voltage than the minimum needed to overcome the drop out of it, say 5-6V over output voltages of each regulator. Or use HV versions of some regulators available, LM317 has a HV version (60V max input).

[jackand08]

Thanks for all the advice, will definitely try loading the outputs before measuring the voltage this time.

As for killing the reg, I replaced it once, and then increased the voltage on my lab supply below the maximum input voltage for the regulator (the supply only goes up to 30V) and still, it did not regulate.

Therefore, I am thinking that it is purely to do with the loading of the outputs. I also agree about the high input impedance-although it probably wouldn't cause too much oscillation with the bypass caps being so close, I understand that it is not a good thing for power supplies. At least I have space on the board (where the dropper resistors are) to add a zener if I need to go down that route.

[AndrewT]

I see you have the PCB traces different for the PINs of the two regs. This is normal for +ve & -ve regs, but check again you have this correct.

[jackand08]

Thanks for the suggestion Andrew, that was one of the first things I checked. For the 7915 it is Gnd - Input - Output with the part number facing you.

[ChrisA]

Originally Posted by jackand08

Hi Everyone,

I have designed and built a basic power supply for a speaker system project of mine but am having trouble in finding the cause of this problem.

It is set up as most PSU's are as a dual polarity supply, but with the main power rails (around +/-35V from a 25V dual secondary transformer) regulated down to +/-15V (using L78/7915 regs) for the input circuitry. There are also dropper resistors to reduce the voltage going to the regulators to something that is in their input voltage range.

This is going to be very unreliable. +/- 35 volts to simply to high for a 78xx or 79xx regulator. Your dropping resisters are useless until you have a load going.

What you need is a transistor, rated for at least 50V and use the LM79/78 to control the base of the pass transistor. Look in the data LM78/79 data sheets for an example

Remember about "transients" and start up and shut down. Put in some zeniers to shunt over voltage around the pass transistor.

[sofaspud]

There was a load on the circuit, but it was the ~5mA quiescent current through the regulators. At that minimum current, you'd want the series resistors to be >200 ohms (for >1V drop). But I'd do without them. 25V (transformer secondary) times 1.414 equals 35.35 volts. With bridge diode drops you should be under the 35V max reg input. It's close so I would definitely measure to see exactly where it's at. The circuit should otherwise be good for about 100mA with only the smallest of heat sinks (20V x 100mA = 2W). For your average input circuitry that should be plenty.

[morinix]

terminate the neg regulator with a 1k resistor.