So I wish to measure the current of a 20kHz signal.
The specifications for my multimeter on the AC range are not listed, but 5kHz is listed as +- 0.1% of reading.
Does any one have a rule of thumb as to how inaccurate this reading may be at 20kHz? Do you think I can trust it to 5%, I can't imagine that it will be attenuated too hard?
Also, how important is probe choice for 20kHz... I have it hooked up using alligator clips, I figure that 20kHz is similar enough to DC that I should not worry about it too much..
34401A? Even the
datasheet says the AC bandwidth: 300 kHz.
34401A? Even the datasheet says the AC bandwidth: 300 kHz.
but the ACI function only has listed up to 5kHz, the ACV function is 300KHz.
I was curious why. the manual mentions inductance being the main factor for ACI error, does the meter have a inductance that makes it unusable above 5kHz? Are the spec writers just lazy?
And is a 3 foot long alligator clip a serious error for 20kHz?
True. Interestingly, it lists the current accuracy spec to 5 kHz, but that line has a note:
For sinewave input > 5% of range. For inputs from 1% to 5% of range and < 50 kHz, add 0.1% of range additional error
So they seem to be implying that it's usable up to at least 50 kHz. Seems a bit unclear. In the manual, perhaps?
I can't really find anything specific in the manual relating to the ACI mode. I don't know if those little footnote tips are applicable to ACI mode... I guess I should email agilent or wait longer.
Specifications are for 1-hour warm-up at 61
?2digits,
Slow ac filter, sinewave input.
[ 2 ] Relative to calibration standards.
[ 3 ] 20% overrange on all ranges, except 750 Vac, 3 A range.
[ 4 ] Specifications are for sinewave input >5% of range.
For inputs from 1% to 5% of range and <50 kHz,
add 0.1% of range additional error. For 50 kHz to 100 kHz,
add 0.13% of range.
[ 5 ] 750 Vac range limited to 100 kHz or 8x10
7
Volt-Hz.
[ 6 ] Typically 30% of reading error at 1 MHz.
[ 7 ] For frequencies below 100 Hz, slow AC filter specified
for sinewave input only.
[ 8 ] For 1 k?unbalance in LO lead.
[ 9 ] Maximum reading rates for 0.01% of ac step
additional error. Additional settling delay required
when input dc level varies.
[ 10 ]For External Trigger or remote operation using default
settling delay ( Delay Auto ).
[ 11 ] Maximum useful limit with default settling delays defeated.
[ 12 ]Speeds are for 41
?2digits, Delay 0, Display OFF, and
Fast A
Yeah, the manual doesn't seem to be written with the idea that anyone would ever try it above 5 kHz.
And is a 3 foot long alligator clip a serious error for 20kHz?
You'll have around a microhenry of inductance, so the burden impedance will increase by around 125 milliohm.
So they seem to be implying that it's usable up to at least 50 kHz. Seems a bit unclear. In the manual, perhaps?
Since this footnote also applies to ACV, I think you can safely assume the 50 kHz limit only applies to ACV, and that ACI is limited to 5 kHz. I can't see why 1-5% of range would have a higher frequency limit than 5-100% of range.
Series resistance/inductance won't cause an error with current measurements (it will only increase the burden voltage), it's the parasitic inductance of the shunt and PCB traces within the meter you'd have to worry about. The input of a similar DMM I had on hand is about 0.3 uH in series with 0.2 Ohm, but I would expect the majority of this to be outside the sensing circuit.
Since operation beyond 5 kHz is not specified, there's no way to know except testing it. It may even vary depending on the age of the instrument (they could have switched shunt type, as long as the inductance is within spec for 5 kHz).
I tried the same source on a keithley meter which is specified to 10khz and I found a discrepancy of 5mA. (98 vs 93).
But I think the keithley has a worse crest factor performance and what I am measuring appears to be a triangle wave.
The specifications only apply to sinusoidal signals. It's 5 kHz bandwidth, not a 5 kHz fundamental frequency. Non-sinusoidal signals contain harmonics with higher frequencies (eg. 10 kHz and 15 kHz). This will introduce an additional error beyond the error for a 5 kHz sine.
So I conducted a test... I hooked up the meter in series with a function generator and a resistor set to ACI mode. (3kohm resistor, 5Vpp sine wave)
I measure 60uA at 1kHz and I pump up the frequency. I notice little change from 1-50kHz, past 50 things drop (50uA).
Is this an acceptable test ?
with a triangle wave stays around 40uA from 1-40kHz then drops to 30uA at like 100kHz.
done with a film resistor. I suppose the resistor can be the problem too. and the wires.
Film resistors are not going to be the problem, most likely it is a gain bandwidth limited opamp inside the unit that is used to amplify the current range voltage across the shunt. Low noise and high gain with a high bandwidth are all going to need a different opamp. Most likely the bandwidth is limited to reduce internally generated noise on the current ranges.
I measure 60uA at 1kHz and I pump up the frequency. I notice little change from 1-50kHz, past 50 things drop (50uA).
Is this an acceptable test ?
You try to compare a 100mA value from a 60uA value (probably in the same 200mA range?
I would try a 50 Ohms load or even smaller to get currents in the right range.
With best regards
Andreas
I measure 60uA at 1kHz and I pump up the frequency. I notice little change from 1-50kHz, past 50 things drop (50uA).
Is this an acceptable test ?
You try to compare a 100mA value from a 60uA value (probably in the same 200mA range?
I would try a 50 Ohms load or even smaller to get currents in the right range.
With best regards
Andreas
It is a 1A range.
I chose a low value of current because I was afraid that the function generator is not built with amplitude stability in mind and loading it down may cause drift due to heating of components.
I will also try a high current test.
Film resistors are not going to be the problem, most likely it is a gain bandwidth limited opamp inside the unit that is used to amplify the current range voltage across the shunt. Low noise and high gain with a high bandwidth are all going to need a different opamp. Most likely the bandwidth is limited to reduce internally generated noise on the current ranges.
Is the 34401A shunt resistor a film resistor? It could also be a precision wire wound. The 34401A has a 0.1 Ohm shunt and its lowest ACI range is 1 A, i.e. a 100 mV drop full scale. The lowest ACV range is also 100 mV, specified up to 300 kHz, with a typical -3 dB point of 1 MHz. So I see no reason why the AC current range would involve an extra (bandwidth-limited) gain stage. My guess is that they just don't feel comfortable with the parasitics of the shunt to specify it beyond 5 kHz.
I would also be worried about the amplitude stability of the function generator, but in general the amplitude would go down with frequency. You can of course check this by measuring the voltage across the series resistor. The meter is specified from 5% of full scale (1% with reduced accuracy), or 10 mA. So it's likely quite non-linear in this range (do 30 uA and 60 uA give the expected readings?). Assuming the function generator has a 50 Ohm output impedance, it should be able to source 10 mA in a 50 Ohm series resistor. Eventually the 5 kOhm film resistor would become capacitive, but this is unlikely to be an issue at audio frequencies.
I would monitor the voltage across the series resistor to check if the current is indeed constant: connect the lo terminal to the series resistor and connect the current input back to the function generator. To measure voltage, you can connect the hi terminal to the other side of the series resistor so you measure the voltage across it. To check for the effect of the input capacitance of the meter on the impedance of the series resistor, you can test if the current changes as you attach the hi terminal to the resistor.
1kHz - 33.9 mA
20 kHz - 33.88 mA
50 kHz - 33.8 mA
150 kHz - 33.15 mA
750 kHz - 27.1 mA
2.5 MHz - 53mA
I am pretty confident it reads ok now!
I don't understand why they did not specify...