wanted: device to monitor AC power via sound card

You could put a voltage transformer on the line for some step down and isolation and then put a pot across the transformer to aide in calibration. Send the attenuated waveform to a PC based sound recorder application. A cheap USB mic or line card should have plenty of bandwidth and may even come with basic DAW software. Something like this recording one mono channel should run for a week. Two channels (240 VAC) half a long etc. Three channels three phase one third.

If you need a really cheap recorder to leave on site or spread around you could use a portable recorder like the Zoom H2 and leave it recording 8K MP3 for quite some time.

Then you could mix the recorded power line signal with a synthesized out of phase sine wave of the appropriate frequency (50-60 Hz) and all that should be left on the mixdown are noise and voltage changes. All mixing would be non destructive. You could use hitpoint markers ( how we find drum whacks to replace them with samples) to find all your perturbation points.

| You could put a voltage transformer on the line for some step | down and isolation and then put a pot across the transformer | to aide in calibration. Send the attenuated waveform to a PC | based sound recorder application. A cheap USB mic or line card | should have plenty of bandwidth and may even come with basic | DAW software. Something like this recording one mono channel | should run for a week. Two channels (240 VAC) half a long etc. | Three channels three phase one third.

What kind of frequency bandwidth would that transformer have?

I plan to do this with my own software. I don't know how many computer programmers prefer to run their own, by I do. It makes it easier to innovate. Back when devices used serial ports to interface with computers, they would just document the serial data stream and let peoplw write their own. Now days too many things use USB, keep the data stream secret/proprietary, and require the use of an unstable operating system.

I am hoping for a resistive and/or optically isolated method of interfacing. But if a voltage transformer has a wide audio grade bandwidth, that might be a good part to use. It would be easier to work with the resistor part at 12V instead of 240V.

| If you need a really cheap recorder to leave on site or spread | around you could use a portable recorder like the Zoom H2 and | leave it recording 8K MP3 for quite some time.

I want to have alerts for certain power abberations sent to me when they happen. So offboard recording won't do. But that would be a nice way to record a week to a month of audio portably.

| Then you could mix the recorded power line signal with a | synthesized out of phase sine wave of the appropriate | frequency (50-60 Hz) and all that should be left on the | mixdown are noise and voltage changes. All mixing would be non | destructive. You could use hitpoint markers ( how we find | drum whacks to replace them with samples) to find all your | perturbation points.

I planned to do both something like that, as well as mixing the waveform with itself from N cycles before as a means to detect changes.

If your want UL approval and 600 volts you would have to get a purpose built voltage instrumentation transformer from someone like ABB and see what bandwidth they have.

Audio transformers with 100K bandwidth are "high end" but available. A transformer from Hammond that is made as a tube amp interstage coupling transformer or output transformer should handle the voltage and bandwidth. 20 KHz is a piece of cake for this type of device.

A Hammond power transformer should get you to 600Hz easy.

Optically the line would have to be brought down to the devices level before applying it to the device.

As far as open architecture for the USB/ADC devices, look at VST interfaces. The VST spec is from Steinberg and Charlie may share it with you.

Good luck

Peace dawg

Are you sure you want to put a transformer in the middle? The recorded waveform I guess is going to be more regular than the real AC source. What about a voltage partitor with pure resistors? (Obviously high AC spikes are going to risk the life of your sound card, it is more likely in this way than with the transformer)

Only to the extent that frequency components of the waveform exceed the response of the transformer. Its a tradeoff, but just how much data does the OP really need? Transformers can be had with better frequ responses than typical sound cards.

That will work. Add some transient suppression on the input side, plus some peak limiting circuitry (diodes) on the card input and there shouldn't be a problem. With this approach, the voltage under test needs to be treated as a floating signal. Don't depend on the neutral being close to ground potential. Wiring could be backwards or the neutral voltage could jump due to switching transients or a fault. Of course, this goes without saying for line to line measurements.

Yes agreed, I am thinking that he could also measure the line after a surge-suppressed power strip like an APC one

A capacitive divider will maintain its ratio at all frequencies, so it will be more accurate than a resistive one. I have used one to connect a scope to the power line, and it works well. Of course, I doubt he will be looking at frequencies high enough for it to matter.

None of this solves the problem of isolation, which is essential for safety. I think a voltage transformer will have plenty of bandwidth. Of course, since phil seems to like experimenting, he could go the route of professional equipment. Do all of the analog processing and A/D conversion on the line side, then couple the digital data through a 50 cent opto-isolator to the serial or parallel port of the PC. This would not use the sound card.

Ben Miller

Only true if you assume you can find perfect capacitors but not perfect resistors.

I want the waveform AND I want to do it on

Lateral. Webcam looking at an Oscilloscope?

It has nothing to do with "perfect" components. Resistive dividers, by necessity, are limited to fairly large values of resistance. The parasitic capacitance that shunts each resistor is somewhat random and uncontrolled, and is unlikely to be in the same ratio as the resistors. This has the effect of changing the division ratio as the frequency goes up.

A capacitive divider can use large enough capacitors that the parasitic capacitance effects are negligible, and all frequencies of interest are divided by the same ratio. Granted, the accuracy may be sufficient for Phil's purpose either way, but it is a factor if you are looking at high frequency noise, carrier frequency signals such as X10, etc.

Ben Miller

Capacitors don't have internal resistance? ...not to mention tolerance issues.

*ALL* frequencies? Large capacitors at line voltage? X10 is "high frequency"? Sorry, while capacitive dividers have a place, you're dreaming.

Tolerances can be measured and then compensated in the calculations, or adjusted out with a trimmer (as is done in the capacitive divider in every x10 scope probe). Internal resistance is ngligible for this purpose.

"All frequencies of interest" does not mean "all frequencies". It means within acceptable tolerance *over the range of interest*.

Large enough to swamp the parisitics. We are still talking about fractions of a uF. I don't know what your definition of "large" is.

It depends on your perspective. From the standpoint of a cell phone operating at 1+ GHz, no it is not high frequency. From a power system standpoint, 100 kHz vs 60 Hz. is high frequency.

We appear to be debating semantics here. Virtually every 10X scope probe uses a capacitive divider in parallel with the resistive one. He will need more like a x100 divider, making it even more important, as the series impedance will be higher. Since he doesn't need to measure DC, he can eliminate the resistive divider. It doesn't get much simpler than two capacitors, and the power dissipation is negligible. As I pointed out, however, he may get satisfactory results for his purpose with a resistive divider. I am just proposing a tried and tested alternate method that has been in commercial use for years.

Scope probes have no resistors?

DC is often of interest.

Yes, the capacitor is there to *COMPENSATE* for the amplifier's input capacitance. It is a resistive divider.

No, it doesn't get much simpler than two resistors.

Yes they do. Read my entier post befiore you reply.

Not typically for AC power measurements, which is the context of this thread. I do a fair amount of power measurements, with equipment that shows all harmonics down to DC. I have never seen anything measurable in the DC component.

It is both. How does the series capacitor in the probe compensate for the input capacitance? By creating a capacitive divider with the input capacitance that has the same ratio as the resistive divider:

(scroll down to number 5)

I will grant you that they are equally simple in terms of component count.

No, it is most assuredly not. The capacitor in the probe is there

*only* because of the stray capacitance of the amplifier input. Notice that the capacitor is adjustable and the resistor not.

And far more simple to get right.

On Wed, 03 Dec 2008 20:04:25 +0100 qmu wrote: | Paul Hovnanian P.E. wrote: |> That will work. Add some transient suppression on the input side, plus |> some peak limiting circuitry (diodes) on the card input and there |> shouldn't be a problem. | | Yes agreed, I am thinking that he could also measure the line after a | surge-suppressed power strip like an APC one

That might possible. Part of what I want to measure are the surges, at least to 800V peak. Maybe I would later decide to make a more reduced level device to also measure more extreme peaks, if I find peaks happening that clamp at the 800V level.

| We appear to be debating semantics here. Virtually every 10X scope probe | uses a capacitive divider in parallel with the resistive one. He will need | more like a x100 divider, making it even more important, as the series | impedance will be higher. Since he doesn't need to measure DC, he can | eliminate the resistive divider. It doesn't get much simpler than two | capacitors, and the power dissipation is negligible. As I pointed out, | however, he may get satisfactory results for his purpose with a resistive | divider. I am just proposing a tried and tested alternate method that has | been in commercial use for years.

It seems like a viable idea to me. What farad values do you guess might be a good starting choice? Obviously I'd need a high working voltage.

| Not typically for AC power measurements, which is the context of this | thread. I do a fair amount of power measurements, with equipment that shows | all harmonics down to DC. I have never seen anything measurable in the DC | component.

I don't expect any DC. But there might be systematic noise _below_ 60 Hz, such as from motors. I'd want to see that. As a guess, I'd like to go down to 5Hz on the bandpass, and up to 20 kHz. I've also found sound cards that can do even higher (beyond 100 kHz) which I might try some day if it works in my OS.

|> No, it doesn't get much simpler than two resistors. | | I will grant you that they are equally simple in terms of component count.

It seems like it would be just as simple with capacitors. But if capacitance is more of a nuisance with resistors than resistance is with capacitors, then the latter sounds like the way to go. I had never thought about capacitors to do this, before, but it makes sense. It's not like I'm trying to measure current via a voltage drop. I'm trying to measure voltage via a voltage drop.

Am I missing something here? Are we considering a sound card with infinite input resistance? If not, it seems that a resistive divider would be necessary to obtain any reasonable low frequency response with any reasonable capacitance values, especially if the input resistance is relatively low. In all the discussion I have not seen any reason not to use a resistive divider, compensated for high frequency response by capacitors if necessary, despite the fact that capacitive dividers are possible. Resistive dividers are commonly used at any desired degree of accuracy and stability at moderate prices.

Don Young

Don Young