Power Supply Ground Confusion

[Dwiel]

Hey guys, the project I am currently working on only slightly has to do with audio. The goal is to syncronize lights to music. I am currently having a PIC control everything. I didn't have any triac's handy and was having a hard time finding good schematics for them for what I am trying to do, so instead decided to use MOSFETs to switch the mains voltage. The problem is that mains voltage is AC and MOSFETs can't switch AC. So I simply put a bridge rectifier on the mains source giving me a 120V 120hz pulsed DC. Since I am merely switching a light bulb, this is no big deal.

Enough background. I found that the MOSFET only worked when I tied the - side of the bridge rectifier to my +5v power supply ground. That supply is what is being used to switch the MOSFET. This causes a different problem though. Now, my ground is 60V higher than the ground in my wall and in all of my other electronics. I measured with a voltmeter my ground and the ground on a few of my amps as well as the ground in an outlet to confirm this. So now I can't connect my circuit to any others because the grounds are way off and would cause a short... I know there must be something simple I am missing, but I can't figure out what it is.

Thanks!

Zach Dwiel

 

[HDTVman]

You need to use opto isolators to drive your FETS and a seperate isolated +5VDC supply for the HOT side of the opto isolator. BE VERY CAREFULL. This could be a shocking experince.

 

[Dwiel]

Thanks for the info!

Here is an attached picture of the schematic I am thinking of. It is slightly different than what I have built and was using which prduced the wrong and unsafe grounds. Does what I haev drawn look correct? I am not sure however which of my grounds can all safely be connected. My biggest worry is 'gnd1'... I doubt that I can connect it to 'gnd2' because of the said 60V difference when they are not connected. I have found that the only way to get the FET to switch is to connect gnd2 and the gnd associated with +5v#1. I also understand that +5v#2 and its associated gnd3 must be isolated from all of the others in the circuit, but I'm still not 100% sure how isolated they need to be. Do I need 2 transformer/bridge/cap/vreg sets for each? Or can some of those be shared? Maybe only have a single transformer and 2 bridge/cap/vreg sets comming off of that or something like that to cut down on size and cost.

Thanks abunch for the help!

 

[imix500]

Also keep in mind that you are actually supplying ~162v DC after full wave rectification.

 

[HDTVman]

Thanks for the info!

Here is an attached picture of the schematic I am thinking of. It is slightly different than what I have built and was using which prduced the wrong and unsafe grounds. Does what I haev drawn look correct? I am not sure however which of my grounds can all safely be connected. My biggest worry is 'gnd1'... I doubt that I can connect it to 'gnd2' because of the said 60V difference when they are not connected. I have found that the only way to get the FET to switch is to connect gnd2 and the gnd associated with +5v#1. I also understand that +5v#2 and its associated gnd3 must be isolated from all of the others in the circuit, but I'm still not 100% sure how isolated they need to be. Do I need 2 transformer/bridge/cap/vreg sets for each? Or can some of those be shared? Maybe only have a single transformer and 2 bridge/cap/vreg sets comming off of that or something like that to cut down on size and cost.

Thanks abunch for the help!

OK, let's see here.

Grd 1 is AC ground and should be connected to the chassis.
Grd 3 is circuit Grd for +5v#2 and the PWM circuits.
Grd 2 should not connected to ground. It is the - connection for +5v#1 and must be isolated from all grounds.

Do I need 2 transformer/bridge/cap/vreg sets for each? Or can some of those be shared?

You need a seperate isolated power supply for +5v #1. It does not need to be regulated. It can be just a transformer / bridge / cap. It can be very low current, less than 100 ma is OK.

I think I would put the load on the other side of the bridge rectifier in the neutral side of the AC line as shown in you drawing.

I say again BE CAREFULL.

 

[SY]

Use an isolation transformer! This is a potentially deadly circuit.

 

[TwoSpoons]

Ditch the MOSFETs. Go buy some TRIACs. Its really not that hard.

Use something like the MOC3021 (?) isolated triac driver.

Stay alive.

Are your lights going on/off or are you trying for dimming?

 

[jan.didden]

This is a dangerous circuit. Make ONE mistake and pouf, you'r dead meat.
If you don't have a good life insurance yet, this is probably an excellent reason to get one....

Jan Didden

 

[Dwiel]

Thanks for the suggestions guys! I guess I'll be switching to triacs rather than PWM controlled MOSFETs. I do have a few more questions though. I'd rather be safe than sorry in this case... I've already been shocked by this system once, am I lucky to be alive?

First, why does using a triac make the circuit safer? Is it that I don't have 120VDC? I assume thats the main concern.

Second, I definately want to be able to dim the lights, and hopefuly do so with a fair amount of control and speed. I assume I will still be able to do this with a triac. Am I correct that the triacs always switch at the same frequency as the mains? From what I've read, it looks like they simply shorten each cycle. How much extra electronics am I going to need in order to adapt my PWM signal into an in phase, correct frequency one. I guess making it the correct frequency shouldn't be a problem, but making it in phase might... Is that a necessity?

I have seen many triac controlled light dimmer schematics online. Would I be able to simply replace the pot with an optoisolator? I could then send my PWMsignal at any frequency and any phase to that in order to control the brightness. Would this work?

I guess the goal would be to have a simple, safe circuit which could take a high frequency (higher than audible) PWM signal and control the brightness of a lamp off the mains power with it.

Thanks agian for all of the help! It might be advice which saved my life.

 

[imix500]

If you already have the pwm signal up and running, you can drive solid state relays with it. That is what 99% of modern dimmer packs use. They are much cleaner and easy to drive.

 

[Dwiel]

A lot of solid state relays I am finding don't report the on/off time which I am slightly worried about. I can adjust my PWM frequency, but I'd rather not have to. Right now, the max frequency is around 20,000hz Can most solid state relays switch that fast? Or should I make sure I find specs on them before I buy them? I would need them to be able to turn on and off around 2.5 or maybe 5 microseconds. That would allow me to use a 20khz PWM and get a maximum of 90% full power to my lights.

Although after a little more looking, it's looking like even fast solid state relays take 0.05 ms to switch, which is still 50micro seconds which is way to long for a 20khz PWM. I think, please correct me if I am wrong.

Maybe I should just stick to triacs. I assume they switch plenty fast enough?

 

[imix500]

Yeah I think the designs I've seen are around 1khz. Don't think ssr's could switch that fast.