Power sensor?

Google for 'high side current measurement'. There are specific devices, basically differential opamps with very wide common mode operating range that exceeds the power supply, that can be used to measure current. Knowing the high side voltage you can then calculate power.

Mark.

What do you mean by "calculate power", are you referring to using a calculator?

If so, that's not what I was looking for, I can surely use a calculator and multimeter to multiply voltage by current, but I need some device to make real time signal that is proportional to V*I.

Let me explain: I made a power supply. High power, 0-300 VDC, 0-200 amp, about 10 kW or so. (a welder actually).

It is controlled by SCRs and uses a controller that takes a voltage input and "current input". The controller regulates SCR firing phase angle to match the voltage and current.

It works. Does welding with stick already.

It can be made into a constant current or constant voltage source based on potentiometer settings.

Since the controller does not know what is the current input, I could as well use a "watt transducer" to input _power_ instead of getting input from a _current_ shunt.

That would make a constant power welder.

For this, I need a power transducer that would perform multiplication of I by V.

An example of such a device is here:

I called them and they quoted $600 for it, which is beyond my means.

Are there any cheaper ways to achieve what I want (make a watt transducer).

i

Do the calculation in any microcontroller. I do 'per cycle' switch mode PWM modulation over 40KHz, taking voltage and current measurements with a chip like this. A modern micro able to do this should be easily under $5.

Best Wishes

I found something called "analog multiplier", AD633.

``The AD633 is a functionally complete, four quadrant, analog multiplier. It includes high impedance, differentlay X and Y inputs and a high impedance summing input (Z). ...''

Output W = (X1-X2)*(Y1-Y2)/10v + Z

Would it work if I feed it with current input and voltage input, would it produce output equal to product of them?

I could use

- a divider to produce 0 to 10 volt voltage input from bus voltage

- input voltage across 50 mV current shunt.

Then the output would vary from 0 to 50 mV, based on input values. In reality, I would never be able to do 300 amps at 300 volts output, so possibly the output would need to be rescaled to work as feedback for my SCR firing system, but that should not be a huge problem.

i

Hello Roger,

That would probably include a healthy profit margin. After all, the payments on the Ferrari need to be made on time :-)

The MSP430F2013 would be ideally suited for that. 16bit SAR conversion on multiple channels, can be hooked up differentially. Built-in programmable gain amplifier, on-board reference, temperature watch and zero-calibration. It doesn't get much better than that.

Regards, Joerg

Why not? But pay close attention to its common mode range. IIRC it's

+/-10V if your supplies are +/-15V.

Also, its 2% total full scale error can be a lot higher if your true full scale is a fraction of what the chip can do.

Regards, Joerg

What I meant was use a high side current monitor, i.e. differential amplifier, that produces an output voltage proportional to current, then use a resitor divider to produce a voltage proportional to the supply voltage, then use a microprocessor such as an ATmega8515 to read both of these two values using its built in analogue to digital converters. Use a C compiler such as CodeVision to write a program that multiplies the two figures together and produces a pulse width modulated signal whose mark/space is proportional to the product, then apply an external filter to that to produce a voltage proportional to the mark/space ratio. Hey presto, an output voltage proportional to power without using expensive quadrant multipliers. The microprocessor should be a few dollars or so.

Mark.

Yes, so I could get both inputs to be on 10V scale. In case of the voltage sensing, it is not a problem, in case of current sensing (50 mV), I would have to do something with a opamp to get it to 0-10V, nothing terribly complicated, right?

i

It's not complicated. Again, you'd have to mind the common mode range of the opams. Many such as the old LM324 allow the input to go to the negative rail but not to the positive. If you need Vcm up to the positive rail you'll have to go through the lists at National, TI or others.

Regards, Joerg

The analog multiplier is a natural for this application. However, instead of a current shunt, I'd recommend a LEM current transducer

Price is quite reasonable for the units. Offers total isolation for the current signal, and the output is easily scaled. That will relieve you of the problems associated with opamp common mode ranges.

Cheers

--
Dave M
MasonDG44 at comcast dot net  (Just substitute the appropriate characters in the 
address)

Some days you\'re the dog, some days the hydrant.

In my case, both voltage and current signals will be referenced to the negative side of the DC bus.

i

That will make life a whole lot easier. Besides the AD633 you can probably find all you need right in your parts bins.

Regards, Joerg

a

scaled.

To my huge surprise, these transducers are very inexpensive! Thanks!

Let me sit on this for a few days. I am deciding whether I should ground the negative DC side of the welder, or let everything float (and what would be the implications for the PCTI SCR controller). I think that from a safety standpoint, floating is better, plus I could weld grounded objects (like fence). On the other hand, this behemoth of a welder is not portable and can hardly reach anything grounded. It will likely be used for welding various assemblies.

i

Well, I do not think that I have suitable opamps.

Joerg, this is great. I am currently wrapping up the welding part of this, just want to screw down everything that needs to be screwed down, make a nice little temporary control panel, etc, etc etc. After a little while, when I am done with CC and CV modes, I will work at my leisure on providing power feedback.

I will definitely implement this feature. I am kind of hopeful that CP will provide me with the most idiot proof way to weld.

i

of a

scaled.

ranges.

Iggy, all the nifty analog multiplier suggestions aside, you seem to have missed Genome's idea: you can measure *input* power instead. Since the supply voltage is fixed, all you need is the input current (LEM). Just scale that reading by the input voltage, correcting for conversion efficiency, and you're done.

Or,

Cheers, James Arthur

of a

scaled.

ranges.

First of all, I know 10 times less than you or Genome.

But let me share my concern: this SCR fired welder has an awful power factor. The input current is not commensurate with output power at PF of 1. That's why I am not sure if I can measure input current and make proper conclusions from it.

If you are skeptical, I could try to prove myself by measuring input currents and calculating power at the arc approximately.

i

Not even an LM324? Radio Shack usually has them. Buck forty-nine or so. Less at the usual distributors.

Just be careful :-)

Regards, Joerg

If you can read current and voltage with a microprocessor then you could measure the phase angle between voltage and current and work out the real power from that.

Mark.

instead of a

scaled.

ranges.

No need for that. Not paying enough attention, I'm afraid I assumed an architecture with constant power factor, or at least power factor correction, ala a big switcher.

The link I gave for Pease's analog wattmeter still applies, as do the other, nifty multiplier schemes.

James Arthur