GFCI operation question

|> | 120/240 supplies switch the rectifier type depending on input voltge. |> In |> | the 240V position, they are a full wave rectifier. In the 120V position |> | they are configured as voltage doublers. Either way, the bus voltage is |> | supposed to be around 280V to 350V. In the 120V position, 240V will |> take |> | them out. Most PC supplies are this type. Some are autoswitching, but |> a |> | sudden increase from 120V while operating can take them out. |>

|> But autoswitching is NOT the same as autoranging, correct? |>

|> So the autoswitchers are best switched only from a power coming on state? | | Autoswitching and autoranging usually mean the same thing. | | Wide range is the term you are looking for.

The specifications of some power supplies say 100-250 volts. So shouldn't that mean they could operate just fine on 150 volts? These say autoranging.

|> Can the autorangers function for any voltage in between 120 and 240? | | Wide range can.

So are the makers like Antec just trying to pull a fast one on people?

| One place where autorange or manual switched 120 /240 designs can get into | trouble is 208V feed. The supplies that include 90V for Japan are OK. | Supplies for 120V or 240V only with a lower limit of 102V/204V get into | trouble becuase there is not enough margin on 208V.

So what is the term to look for on the power supplies to get the right ones? And which manufacturers make them?

How would you select a power supply for a PC if you needed to connect them to a 208 volt power source? I assume if it were a 240 volt power source (e.g. single phase in the USA, L-L) you wouldn't have any trouble.

|> The impression I got from the designs I saw for a switch mode power |> supply, |> which was not really complete info, is that on a higher voltage the time |> the current flowed was shorter, and for a lower voltage it was longer. |> But maybe that was just for some brownout handling. I have seem power go |> completely out for an instant, long enough for the lights to go all the |> way out and even for clocks to lose time, and yet, my computers stayed up. |> I guess the capacitors have enough reserve and I'm not loading them up on |> the DC side too much. | | Wide range supplies under 90W will have just a straight rectifier on the | input. When supplied with 240V, they have a large ride through because the | ride through is based on minimum voltage. At 2.5 X the minimum input, the | cap has 6.25 times as much energy. | | Over 90W, and the supplies have power factor correction. They boost the AC | to 380V DC, regardless of input voltage. This makes them invariant of input | voltage for ride though time.

How do they boost the AC voltage regardless of input? Sounds like a nice trick a lot of things might like to do.

What do you mean by "ride though time"? Does that imply or mean there is a time limit?

|> | Consumer electronics are often 120V only. This is because their high |> volume |> | and NA unique requirements (NTSC, ATSC, frequency alocation) make it |> cost |> | effective to use 120V only supplies. With these also, 240V will wipe |> them |> | out. |>

|> As long as a single voltage supply has a cost lower than a dual voltage |> by an amount greater than the savings of having only one type of supply |> being manufactured and in inventory, then I guess we wil continue to see |> them, and have damages. |>

|> So why are PC supplies pretty much all dual voltage or autoranging? I can |> see that for laptop external power converters, as they may be used in some |> other country. But for towers that don't move around much? | | One PC design can be shipped to anywhere in the world, with the only change | being the power cord, keyboard, manual, software, and shipping carton. The | tower doesn't change. So it makes sense to have a power supply that will | work anywhere in the world. | | Consumer electronics are either NA or ROW because of NTSC,ATSC,PAL,SECAM & | frequency allocation. Here the box itself is specific to end user location, | so a unique supply is cheaper.

But the power supply is still a component. They either have to stock two different power supplies, or one, at the place where the whole electronics is manufactured. Why is it cheaper to have only one power supply design for computers but not so for TV sets?

Another term used is universal range - I was grasping for the other term earlier and didn't remember until now.

If it's not speced as 2 descrete ranges, then it's likely wide range / universal range and should work anywhere in the whole range.

I think it may be a case of marketing types writing the spec sheet.

PC's are usually OK. They are designed for 90-140 on the low range and

180-280 on the high range. 180V is plenty of margin on a 208V supply.

The equipment I had trouble with were industrial power supplies.

Look Here:

TI is trying to sell the chip so it's written towards that aim, but it is one of the better (free) articles on PFC design.

Yep.

Look here:

That is the current CBEMA standard. It once was 8.3ms. Way to many crashes from dips, so now it's 20ms.

Often the entire product design is a ROW of NA only, so there really isn't a case of two power supplies.

| Another term used is universal range - I was grasping for the other term | earlier and didn't remember until now. | | If it's not speced as 2 descrete ranges, then it's likely wide range / | universal range and should work anywhere in the whole range.

I've seen them spec'd at "100-250", presumably the lowest (Japan) and highest (China) nominal home/office building distribution voltages. Hopefully there is some undervoltage and overvoltage margin in that.

|> |> Can the autorangers function for any voltage in between 120 and 240? |> | |> | Wide range can. |>

|> So are the makers like Antec just trying to pull a fast one on people? | | I think it may be a case of marketing types writing the spec sheet.

Now there's an area where we could use some government intrusion and require some licensing. As it is now we have to depend on land sharks.

|> | One place where autorange or manual switched 120 /240 designs can get |> into |> | trouble is 208V feed. The supplies that include 90V for Japan are OK. |> | Supplies for 120V or 240V only with a lower limit of 102V/204V get into |> | trouble becuase there is not enough margin on 208V. |>

|> So what is the term to look for on the power supplies to get the right |> ones? |> And which manufacturers make them? |>

|> How would you select a power supply for a PC if you needed to connect them |> to a 208 volt power source? I assume if it were a 240 volt power source |> (e.g. single phase in the USA, L-L) you wouldn't have any trouble. | | PC's are usually OK. They are designed for 90-140 on the low range and | 180-280 on the high range. | 180V is plenty of margin on a 208V supply. | | The equipment I had trouble with were industrial power supplies.

The old "some places have 240 and some places have 208" for single phase supply (and even three phase) is a big annoyance sometimes.

|> How do they boost the AC voltage regardless of input? Sounds like a nice |> trick a lot of things might like to do. | | Look Here: |

| | | TI is trying to sell the chip so it's written towards that aim, but it is | one of the better (free) articles on PFC design.

Could this chip be used to control a device that regulates the AC voltage (AC in, AC out) for a building on the order of 200 amps or more?

|> What do you mean by "ride though time"? Does that imply or mean there is |> a time limit? | | Yep. | | Look here: |

| | That is the current CBEMA standard. It once was 8.3ms. Way to many crashes | from dips, so now it's 20ms.

OK, 5 cycles to 12 cycles in 60 Hz systems.

I've seen a PC stay up in around 400 to 500 millisecond outage. I saw an old Apple II (circa late 1970's) stay up in a 1.5 second long outage.

|> But the power supply is still a component. They either have to stock two |> different power supplies, or one, at the place where the whole electronics |> is manufactured. Why is it cheaper to have only one power supply design |> for computers but not so for TV sets? |>

| | Often the entire product design is a ROW of NA only, so there really isn't a | case of two power supplies.

If the power supplies are built on the same floor as the end product, then I can see that being the case. But if the power supplies are sourced in, then I believe there would be economic advantage to have one model instead of two.

And if we start seeing single DC output voltage designs for PCs in the next generation of PC architecture, I think that will help to bring P/S costs even further down.

All GFCIs today must detect a NG connection downstream, regardless of whether or not a load is connected. They inject current on BOTH the hot & neutral so things stay balanced with any load, as long as there is no additional return path that bypasses the neutral.

Ben Miller

It is more likely that the rf is getting into the GFCI electronics directly, rather than through the toroid. I suspect the input to the chip is fairly high impedance, and there may even be some non-linearity that acts as a detector.

I know you have mentioned this before, and although I have never experienced it, I can easily believe that it can happen. I might try some experiments to replicate it. Do you have any idea of how strong and at what frequency the rf was?

Ben Miller