High Voltage at no load

Hi JQ,


It's not uncommon to see an increase of voltage at light loads on flyback supplies, because the output capacitors tend to peak charge when there is no load to absorb some of the power. In order to diagnose this problem with your specific design, please post (or send via PM) a PDF of your schematic for this design.


Cheers!

Thanks for your answer PI-Jordi

As requested I attached the Schmatic I have assembled and the project report from PI Expert. To give you a better idea this supply will be used to power gasoline dispensers we manufacture. The 13 Volta will power display back lights and general electronics and will have 400mA min. load always so no problem.
The 27 Volt output will power relays and solenoid valves. When the pump is in stand-by the load will be only a few mA (Only a power on LED).

Thanks in advance for your help.

Thanks for the info. The first test you should run is to diode-or the Bias Winding with a 12 Volt bench DC power supply. (Please see schematic attached.) You may use a 1N4007 diode or equivalent diode to protect your bench power supply in case the Bias Winding voltage increases above 12 V. Please let me know if this changes the behavior of the power supply at no-load. Thanks

Hi Again PI-Jordi

I was testing in more detail the power supply. When I looked at the outputs with a scope and a high frequency and voltage +25 to -5 Volts related to 0V output apears on the 13 V output. Similar on the 27V output.

I decided to ground (Earth) the 0V output of the supply the oscilation decreased considerably but was still there with few volts.

After a few minutes at half load the circuit fail to work.

Any hints on what I have done wrong.

Thanks

Hi JQ,


It sounds like the chip could be in auto-restart protection mode. When the TOPSwitch does not receive feedback for a certain period of time, referred to as the auto-restart on-time, it will assume that the feedback loop has been broken and will shut down the supply. This prevents supply components from being damaged by limiting the power delivered by the supply while it's running open loop. After the auto-restart off-time, the supply will restart and try to reach regulation again. If it again does not receive feedback, it will shut down and try again after a short period of time. While the chip is stuck in Auto-restart, you will see the output voltage "hiccup" every 1 second or so and then fall back. (Please refer to the device datasheet for the exact timings involved in auto-restart). This sounds like what you're describing, but I can't completely tell from your description. Please post a scope waveform of your output voltage waveforms if you can.


If this is your problem, the issue is that the TOPSwitch is not receiving feedback. Check to make sure that the proper feedback components are installed on your board (with values as recommended by PI Expert), and that no debris on the back of the board has caused the loop to short circuit. Also try the experiment I told you about last week and see if there is any improvement. If there is, it's possible that the Bias Winding is not providing enough voltage to drive the feedback circuit, and you will need to add a couple of turns to the winding.

Hi PI Jordi

I send you a scope of the power supply output with low load (10mA) on each output. I used a Tektroniks MSOn4024. It looks much better now. I followed your recomendations. In fact the use of the 12V supply (I used a batery for isolation)improved the low loads. So I decided to measure bias voltage.

Power up no load -> Bias Output 10,1 V -> Output Voltages are too High.
Connect the 12 V Battery _> 13 V output OK 27 V Dropped from 40 to 29-30V.
Load the 13V 1A and 27V output 2A -> Bias output 19V -> Output Voltages are OK.

As you suggest I think I should increase the turns on the bias winding.
This raises me some questions.
Is this "the more the better" ?
It is now at 19V with 65% load. If I add more turns and more load for sure it will rise even more.
Shall I do some limitation in this voltage ?

Please see the attachment related with these questions.

If I take the load from the 27 V some time the voltage raises back to 30 - 31 V.
I have seen a design on your data sheet with regulation from two outputs. See atttachment. Will this improve regutation ?

The output has some spikes as you may se in scope. They increase in amplitude if I increase the load. I had to ground (esrth) the 0V DC output to remove a lot of noise from the scope.
Is this OK for the circuit behaviour?
What to do to to get reed of these spikes?

Thanks

The downside to increasing your bias winding voltage is that it will significantly increase your no-load consumption. Because the TOPSwitch-GX is a current controlled feedback device, it will draw the same supply current regardless of the voltage at which it is provided. Thus, increasing the voltage on the bias winding will increase the power delivered to the device. PI Expert usually specifies the correct number of turns for a bias winding, however sometimes the leakage inductance of a transformer is higher than expected, and it's necessary in increase the bias winding voltage slightly by adding an additional turn. But increasing the voltage further won't help you in this situation.


For the Y capacitor you pointed out, this component is used to reduce conducted EMI. It provides a path for leakage current to close around the transformer rather than traveling through the input cable. This capacitor is typically placed between HV+ and output return. However, in some cases better results are seen when it's placed between HV- and output return. The only way to determine which is most effective is to try both configurations and note the results. Unless you have specific EMI issues, adding one between HV+ and output return should be good enough. The size of this capacitor is typically limited by a safety specification to about 2.2 nF for most universal input designs where the output is not earth grounded.


As you also point out, it is possible to provide a weighted regulation of multiple outputs. The advantage of doing this is that your high voltage output won't peak charge as much when it's unloaded. The disadvantage is that the regulation of your lower voltage output will suffer. (When the higher voltage output is unloaded, the voltage on the lower voltage output may begin to decrease as the supply tries to regulate both outputs at the same time.) Typically, your lower voltage output will drive a digital circuit which requires very tight voltage tolerances and the higher voltage will be used to power some analog circuitry. If this is the case, a slight increase in voltage on the high voltage output won't damage your equipment, and it's generally better to regulate only from the lower voltage.


As for the switching noise seen, I'm not entirely sure this is real. Because you mentioned that it decreases when you grounded the output, I suspect that the noise your seeing is being picked up by the ground loop on your scope. Try remeasuring this waveform with a ripple probe (see document attached) see if this ripple decreases.


Cheers!

Hello PI Jordi

I have decided to redesign my power supply and I used now PIExpert 7.1. I needed to increase the power. I have now 24V 4,5 A and 14V 2,5 A. Becuase of power increase I quit the universal and I have made it now with 180 to 250 AC Input.
I attach the output of the design.

The 14V output has always load and it works OK.
The 24V output powers solenoid valves and relays and depending on what the machine is doing it may be not loaded (machine in stand by).

If the 24 V has some load .3 to .5 A it works OK.
If it has light or no load the voltage rises a lot depending on the load of the 14V output. Ex.

Bot outputs open -> 14V OK 24V output 26,5 V
If I load the 14V output to 1A and leave 24V output open the 24 output rises to 34 V (more or less).

If I load the 14V output to 2A and leave 24V output open the 24 output rises above 40V !!!!!.
If I Load the 24V output > .5A the output get normal to 24V.

How can I overcome this situation.

Thnaks in advance.

Without looking at your exact schematic for this supply, I'm assuming that your circuit is only receiving feedback from one of the two outputs. Particularly, the 14 Volt output. In this case, it's normal to see some peak charging of the unregulated output when you load the primary output of the supply.




There are two ways to solve this problem. The first, and most straight-forward way, is to place a small pre-load resistor across the 24 Volt output. As you found, this problem only occurs when the 24 Volt output is unloaded, and as soon as you begin loading the voltage comes back down into regulation. One of the ways to solve the peak charging problem is to make sure the output is never unloaded by placing a small resistor sized to draw enough current to bleed the extra charge on the output capacitors. The downside to this approach is that the small resistor will burn power constantly, and may severely reduce the efficiency of your supply.




The better way to improve cross-regulation at no-load without harming efficiency is to divide the feedback signal between both outputs. To do this, you will need to construct your feedback circuit to sample the voltages of both outputs and combine the two into a single signal. This will greatly improve the regulation of the 24 V output, but will also decrease the performance of the 14 V rail. When the 24 V output begins to rise as the 14 V rail is loaded, the controller will reduce the duty cycle slightly to compensate. The result is that both voltages will begin to droop. Unfortunately, there's no way to prevent this cross-regulation. The thing you can control is the weighting of each supply in the feedback signal.




For an example on how to 'and' feedback signals together, look at the main power supply schematic here: http://www.powerint.com/sites/default/files/PDFFiles/der204.pdf
Notice that R14 and R15 are tied to each respective rail, and that by varying their value with respect to each other, you can control how much feedback is taken from each output. If the currents are equal, each output will have 50% of the signal. If they are off balance, one output will have better regulation than the other. It's up to you to decide which output (if either) needs tighter regulation.




If you are afraid about protecting your output load from an over voltage condition, consider enabling the Over-Voltage Protection feature of the TOPSwitch-HX device. See the product datasheet for more information on this feature.




Cheers