Switch mode v Traditional isolated Power supplies help :(

[pigman]

Hi guys,

I've been working on a project that powers a device which averages about 600mA at 12V but fluctuates and peaks at about 4A but for very short periods of time especially when starting up. While it runs on batteries most of the time I am wanting it to run from a 240v source from time to time and as such I've devised a 240v / battery power module with a basic relay to switch between the two automatically.

The issue
I have a 12V 5A switchmode PSU in play and while it works fine 90% of the time, sometimes (only when a cold start situation) the PSU just cant handle it and instead of firing up it will start flickering the power and putting my multimeter on it I'm getting random hz from 1 to 200Hz output and low voltage. and unless I dissconnect it, it will never get to providing the good 12V source my device requires. Sometimes it will work if I turn off the load device and let the PSU get its act together then turn them on, but even then only half the time that will work.

When switching from battery to power it seems to be fine I guess as the Caps in the device are already charged or something.

My question is, would a traditional coil based transformer stop this from happening? I am at my wits end.

I did design a new schematic as I thought the original cause was the relay chatter from the sudden voltage drop, so I designed a different schematic to route the power from the transformer via a big diode (stopping battery feedback during battery operation) and used the relay to simply isolate the battery when on Transformer power. This worked in testing and theory but as the issue isn't about the relay chatter it is still a major problem.

Any help is greatly appreciated!

Thanks

 

[MikeMl]

Put a large electrolytic capacitor on the output of the PSU. Devise a means (delay relay, or toggle switch) to defer connecting your load ONLY after the PSU has started up and charged the capacitor. When you connect your load, the energy stored in the capacitor will "start" your load, meaning that the PSU only needs to supply the average power instead of the instantaneous peak power required by your load.

 

[pigman]

Mike,

Perfect! Basically using a Cap to power the initial inrush to the device and a "soft start" resistor or thermistor to trigger the relay over to Transformer power? And shoves in the right direction on the delay circuit? I've heard of the resistor in line / thermistors being used but this is the first I've had to do this.

thanks mate.

 

[MikeMl]

You could use a PTC thermistor for the delay, but it may be better to use a FET as a series switch between the capacitor and the load.

 

[pigman]

Wow, you're one smart cookie. Take a look at the attached, I just threw it together quickly to see what you thought about this one, here I'm using a zener (don't know what rating yet) to delay operation of a NPN transistor to operate the relay. Would this work in your mind? or should I use a FET? I have never used a FET before so a shove in the right direction there would be greatly appreciated.

Thanks heaps!

 

[MikeMl]

The posted circuit is a voltage detector; it effectively delays pulling-in the relay until after the output voltage from the supply/capacitor reaches a predetermined threshold. Can you tolerate the wasted power in the relay? If not, the FET circuit would be better because it takes almost no power to keep the FET turned on after the initial delay.

 

[pigman]

ohh, sorry you mean replace the relay with a FET? So using the schematic I posted earlier but replacing the Relay with a FET? And the zener + NPN to opperate the FET? What about isolating of the Battery input while the FET is operated? Thanks for your help.

Cheers

 

[MikeMl]

I would use a bistable to switch the gate of the FET so that the FET turns on with a "snap", rather than just the simpler Zener circuit which has rather a soft knee. That will prevent the FET from heating, and minimize its heatsinking requirements.

 

[pigman]

Mike,

Sounds good, but I'm a bit lost are you referring to this style of setup? https://en.wikipedia.org/wiki/File:Transistor_Bistable.svg

I haven't used this setup before and am a bit lost with it, any help would be greatly appreciated

 

[MikeMl]

OK, here is a circuit I call the "electronic relay". It allows the PSU to slowly come up, charging the capacitor as it goes. No current is flowing through the 3Ω simulated load. When the capacitor voltage reaches ~10V, the bistable switch turns on with a nice snap, dumping the capacitor into the load. From that moment on, the PSU delivers the running current. When the PSU turns off, the bistable switch turns back off as the voltage decays below ~8V.

You can use this a place to start. Look at your local suppliers to see what PFETs you can buy there. Write back with the number of something you can find. You are looking for one that has a Vds of >25V, an Id >10A, Vth of -3 to -5V with an On resistance of a few tens of milliOhms.

 

[pigman]

Mike, wow thats a neat circuit. I did however play with a quick Idea I had its very basic ~3second delay. I've uploaded a pic, let me know your thoughts. It should provide a 3 second delay upon startup so the transformer can get its act together and the 2200uF cap can be charged before the transistor fires the SSR and then in turn the Mechanical Relay. I had to put in a basic SSR to get a solid on / off. Would this fix the issue I'm having?
Sorry for the dodgey pic, just a quick sketch, I did breadboard this and it works. Well it works with a 12v zener running 13.8v I need to get a 10v zener or similar to get a good 12v source to switch the transistor.

 

[MikeMl]

Now I'm confused. If the battery is always present, and is wired in parallel with the PSU, then it should be able to supply the peak current to start your load without adding either the capacitor or the other circuitry?????

How do you prevent the PSU from overcharging the battery? Shouldn't there be a diode between the PSU and the battery to prevent the PSU from backfeeding into the battery?

 

[pigman]

Mike, There isn't any tie in from battery to PSU, I have the relay there as a mechanical unit to completely isolate the power sources. I have tied the GND together, is that the concern?

You are correct in that during cut over to PSU from battery when battery is present there isn't any issues with the load on the circuit, the issue is when the battery power is not present, a rare occurrence but does happen and ends up messing stuff up badly.

Do you think I should isolate the GND circuit with a diode?

 

[MikeMl]

No, about the GND

Here is my Final OFFER!

Since you are using the relay, you do not need the SSR; let the relay become the "bistable". Put a variable resistor in series with the coil, and adjust it to set the pull-in voltage of the relay to be about 11V.

The first circuit is the one I prefer. I use diodes to hold up the load voltage while the relay contacts are moving from the NC to the NO side.

The second circuit is what happens when the battery is missing.

The third circuit is what happens if you do it without the diodes, even with the battery in the circuit.

 

[pigman]

Mike, thanks heaps for your help mate! I know this may be a little tedious, my understanding of this stuff is getting better .

In you're schematics there you are also tying GND together, is this an issue? You have me worried now I generally just tie GND together... Also As my device is voltage critical (has a 11.5v cut off for low battery protection) I'm concerned using a diode will take the voltage of the PSU (which is about 12.3-12.4v) down too low in some situations to put the device into low power mode. Will your first diagram with the Diodes fix the GND over batt charge issue you where concerned about? And lastly would using a variable resistor cause a chatter on the relay when there is a voltage dip considering I can't use a 10,000uF cap? I tested out the PSU and from cold start it struggles to charge anything larger than about 4000uF.

Your help is greatly appreciated mate! Stick with me

Cheers

 

[MikeMl]

All of the GND symbols in my (and most) schematics are tied together. Think of the GND as short hand for "tie this to to the common ground point"

Cant you just tweak the PSU so that it puts out 12.6V instead of 12.0V? Use a Schottky in place of the Silicon diode, and that will about halve the forward voltage drop.

I added the diodes to prevent the voltage from sagging as the relay switches. If you dont care about that, then leave them out.

Putting the resistor in series with the relay coil will not cause it to chatter.

If the PSU is having trouble charging a big capacitor, you are not waiting long enough, or it is a really crappy PSU.

 

[pigman]

Mike, Probably a mixture of me not waiting long enough and a shitty PSU... I am limited in space and as such am throwing in a Switchmode PSU, I was using those caged versions that you can tweak the voltage up a bit, but the 2.1A version that fit the location wasn't handling the load. Nominal load is about 600mA but peaks are up to 4A so I ditched it for a sealed 5A switchmode unit that I don't have a "tweak" adjuster on any more. Do you know of a good PSU brand/model/maker? I will give your suggested circuit a go, it seems simpler and less likely to break than mine .

Something like this for the diode? doesn't really need to be 20A but at least 4 or so yea? **broken link removed**

And what Variable Resistor rating would you recommend? I guess it only has to be rated to opperate the coil on the relay 1W or something?

And lastly just to be certain, you're concern about me tying the battery and psu voltages together was because on my sketched schematic you mistook a overlap in the two +'s as a join? And Tying the GND's together isn't an issue? My sketched circuit was a bit crude and I usually use a circle/dot to show a join... Just want to be certain I'm not F'ing it up

Thanks mate

 

[MikeMl]

G'day, Mate

No, there is no problem tying the PSU negative to the Battery negative. Forget about the problem I though I saw...

R2 in the schematic only dissipates <100 mW, so almost any resistor will work. I suggest trying 150Ω +- 100Ω. A 12V relay will usually pull in on like 9V, so by putting some resistance in series with it will desensitize it so it takes more voltage to pull it in. If you add too much, it wont pull at all... Find a resistor where it takes about 11V...

 

[pigman]

Mike, exo! Thanks for that, I have bought a few components I didn't already have to give this a go tonight. What do you think of that Schottky diode I posted a link to? it is a bit of overkill but should do the trick? In fact I should be able to use just one of them too to tie the outputs together and remove that dip in power. Also what is that program you are using to do those charts? looks very cool!

 

[pigman]

Mike, I was just looking at the schematic here about to build it and I'm wondering about the diodes D1 and D3 you have in there. I can understand that D3 is there to prevent the flow back from the battery into the PSU while in PSU mode. The issue is with D1, it is there to minimise the dip in voltage during the change over to PSU mode from Battery mode but the way D1 is configured it would potentially always be powering the load from Battery? While it levels out the voltage change over it also would simply provide Battery power as well as PSU power to the load preventing feedback from either to the other. This defeats the purpose of the physical dissconection?

I like the idea of feathering out that dip in voltage during the relay cut over but is there any other way you can think of that won't cause the Battery to end up powering the device in either mode? I'll give it a go without the diodes.

Cheers mate!