Volt Hertz limitation

[jeffjmr]

Could someone please elaborate and give an example of the volt hertz limit for DMMs?

My Keithley 2015, like many others, lists a limit of 8x10e7 (dot) Hz. I have also seen others list it as 8x10e7 - Hz. When described as a product, I assume the “dot” means “times”, which doesn’t make sense since 80 million volts times hz is an astronomical number of volts, even at 1 hz! Subtracting the hertz from 80 million makes much more sense, but then it is not a “product”.

I ask because I noticed the 2015 will accurately measure frequencies and period FAR in excess of the specifications, with accurate ACV readings as well. I want to be sure I am not exceeding a difficult to understand limitation in the process.

Thanks for any help with this.

Jeff

[alm]

The meter will likely state both a voltage and volt*Hertz limit. You need to stay under both. So if the voltage limit is 750 Vrms and the volth*Hertz limit is 1e7 V*Hz, then at 1 Hz the limit is 750 Vrms, and at 1 MHz the limit is 10 Vrms. So this means that the voltage is derated with higher frequencies (above 13kHz in this example).

Exceeding this means running the meter above its maximum ratings, and could result in degraded accuracy and damage to the meter in the worst case.

One way to think this is imagine the DMMs input impedance as a parallel R-C combination. At low frequencies the resistive part dominates, and hence the impedance is pretty much independent from frequency, while for higher frequencies the capacitance starts to dominate, so the impedance will drop as the frequency increases, and hence the current through the meter.

[jeffjmr]

Thanks, alm.

What you say makes sense. But there still remains the meaning and validity of the limit as described:

≤8 × 10e7V·Hz.

But if the dot means multiply, and it should since the limitation is described as a product, ignoring the hertz for a moment, 8x10e7 equals 80 megavolts!!! There is no frequency you can multiply by 80 million to derive any reasonable voltage limitation for any measuring device I am aware of.
 
If the dot is a misprint, and should either be a / for divide or as I see in manuals for other meters, a - for subtract, the result makes more sense, but then is not a product.

Divide 8x10e7 by a megahertz, and you get 80 volts. Reasonable.

Subtract 1 megahertz and you get 79 megavolts. Not reasonable.

What am I missing here?

Jeff

[xrunner]

Subtract 1 megahertz and you get 79 megavolts. Not reasonable.

I think two things are going on. Yes in that circumstance you are exceeding the absolute voltage limit of the meter, which would take care of that case. If you were working with lower voltages than the maximum input voltage, yet at an increasing frequency, you'd run up against the volt / Hz limit. But yea your example is taken care of because you'd have far exceeded the maximum input voltage.

[jeffjmr]

Ok, looking closer at alm’s example, it looks like v-hz is shown as a unit, not a math operand.

If that is the case, then besides the obvious ultimate limit of 750VAC, the formula

≤8 × 10e7V·Hz.

means a limit of 8VAC or less at 10Mhz, or 80VAC at 1Mhz, 4VAC at 20Mhz, 2VAC at 40Mhz, etc.

Could someone confirm?

Thanks,
Jeff

[CatalinaWOW]

The units of the limit are in V*Hz.  You can't just ignore the Hz in this and call it a voltage than you can ignore one of the meters in an area given in meters^2 and say the length is less than the area.

Let's take a little liberty with the numbers and round the 750V voltage limit to 800 volts (8E2 volts).  The voltage*frequency limit then says that the meter is good for the full 800 volts until frequency is equal to 10E5 Hz, and above that frequency inputs should be reduced proportional to frequency.

While the input impedance explanation is likely the dominant reason for this limitation, only Fluke knows all of the reasons.  And it is un-necessary to know. 

Doing a graph of the input limitations (Voltage on the vertical axis and frequency on the horizontal) may make this more palatable.  There are two lines on the graph.  One at the 750 V voltage limit of the meter, constant over frequency.  The other is drawn by dividing the Voltage-Frequency limit by the frequency and plotting the points to get a line slanting down and to the right.  The safe operating area for the meter is the region below both lines.

It is somewhat analogous to the safe operating area of transistors and fets.  Both voltage and power limits must be observed.  And are frequently plotted on a voltage vs current chart where the power is not directly identified, but is there by the product of voltage and current.

[jeffjmr]

Got it. Then it looks like my last calculations are correct.

Also seems I was safe exploring the limits of the 2015. Despite specifications stating frequency range of 3hz to 500khz, my 2015 accurately displays frequency and VAC to just over 5Mhz if Rate is set to Fast. Period is also speced at 333ms to 2us, and at 4Mhz mine accurately reports .25us. All these with sine wave inputs. Square waves produce similar results at lower voltages.

Nice bonus performance.

Thanks for your help.
Jeff

[David Hess]

High impedance oscilloscope inputs and high impedance probes require voltage derating as the frequency increases for the same reason.  For flat frequency response, the resistive input divider is capacitively compensated.  The current through the shunt capacitance is proportional to the frequency and voltage so the product has to be limited.

There can be other limits like the ability of active stages to slew or bootstrap stages to follow the input signal.