Hi all,
I am looking for a decent True RMS Millivoltmter for measuring low level audio and noise signals. New preferred.
I want to measure:
- Amplifier output noise
( need adjustable high- and lowpass filters, A-weighting, CCIR)
- Flicker Noise of FETs (should be able to measure down to uV)
- PSRR of amplifiers
- exact RMS value of low level signals
- filter transfer function measurements with a signal generator (up to 100 Volt)
- balanced or floating input
- better than 0.1%stability and quality housing
- bandwidth 10Hz - 100 kHz at least
- low noise input (< 2uV rms)
I looked at Agilent.com but did not find anything at the first glance...
What do you use for low level Audio/Noise measurements?
Cheers,
Udo
I am looking for a decent True RMS Millivoltmter for measuring low level audio and noise signals. New preferred.
I want to measure:
- Amplifier output noise
( need adjustable high- and lowpass filters, A-weighting, CCIR)
- Flicker Noise of FETs (should be able to measure down to uV)
- PSRR of amplifiers
- exact RMS value of low level signals
- filter transfer function measurements with a signal generator (up to 100 Volt)
- balanced or floating input
- better than 0.1%stability and quality housing
- bandwidth 10Hz - 100 kHz at least
- low noise input (< 2uV rms)
I looked at Agilent.com but did not find anything at the first glance...
What do you use for low level Audio/Noise measurements?
Cheers,
Udo
One of our Members is selling PCBs for a very comprehensive very low noise, uVac meter that is RMS scaled and does much else in the way of filtering etc.
There are a few others that do a 40dB or 60dB wideband amplifier (LNA) to boost the signals prior to measurement with a less sensitive rms scaled instrument.
There are a few others that do a 40dB or 60dB wideband amplifier (LNA) to boost the signals prior to measurement with a less sensitive rms scaled instrument.
Thanks Thimios!
I want something which is tested and specified...
Something similar to MN-447A 2ch Noise Meter / MEGURO ELECTRONICS CORPORATION
or HP3410a
HP 3410A AC Microvoltmeter Restoration ** In operation added** - Page 1
There is even a description in HP journal 1967-05 how it works
Measure Magazine
EDIT: The Link is wrong - correct one is: http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/hpjindex.html#1967 (click on May-1967).
It should have a decent modern display and an true RMS cirucit based on modern 16 bit ADC chip
Cheers,
Udo
I want something which is tested and specified...
Something similar to MN-447A 2ch Noise Meter / MEGURO ELECTRONICS CORPORATION
or HP3410a
HP 3410A AC Microvoltmeter Restoration ** In operation added** - Page 1
There is even a description in HP journal 1967-05 how it works
Measure Magazine
EDIT: The Link is wrong - correct one is: http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/hpjindex.html#1967 (click on May-1967).
It should have a decent modern display and an true RMS cirucit based on modern 16 bit ADC chip
Cheers,
Udo
Last edited:
Most hand-held DMM's RMS detectors are only accurate to a few kHZ -- you will get much better results with a legacy instrument like the HP 3478, a bench top unit which now sells for less than $100. The HP3478 uses a true-rms detector from Analog Devices.
I have an HP403B which wasn't working so I fixed it up with a very low noise front-end using a THAT1510 (like the SSM2019) -- noise is less than 1nV/Rt Hz. The amplifier feeds a Linear Tech LTC1968 true rms converter and another op-amp to drive the meter.
If you're going to measure noise, remember that you will need a filter -- A-Weight, CCIR etc so that you can compare your results to another persons. The Audio Precision has narrow bandpass filtering in both digital and analog modes.
If you're interested in measuring semiconductor noise -- take a look at this paper on measuring below 1Hz -- it uses the TI OPA4140 quad JFET amplifier:
http://jci2012.sciencesconf.org/conference/icnf2013/cache/export/12266.pdf
I have an HP403B which wasn't working so I fixed it up with a very low noise front-end using a THAT1510 (like the SSM2019) -- noise is less than 1nV/Rt Hz. The amplifier feeds a Linear Tech LTC1968 true rms converter and another op-amp to drive the meter.
If you're going to measure noise, remember that you will need a filter -- A-Weight, CCIR etc so that you can compare your results to another persons. The Audio Precision has narrow bandpass filtering in both digital and analog modes.
If you're interested in measuring semiconductor noise -- take a look at this paper on measuring below 1Hz -- it uses the TI OPA4140 quad JFET amplifier:
http://jci2012.sciencesconf.org/conference/icnf2013/cache/export/12266.pdf
Gotta love a Robrenz restoration. That man is a GOD at restoration, he's a master machinist and as such has some amazing precision manufacturing skills. Check out his youtube videos sometimes to see some amazing work.
The 3478A uses the AD536 which has a -3dB bandwidth of 450kHz. Most decent handheld meters now use AT LEAST the AD636 or AD637 which are 600kHz and 900kHz respectively. Some budget meters use the AD736 or AD737, which at worst have over 100kHz of bandwidth. Not to hate on the 3478A which is a solid meter for a variety of reasons. Just that its not for this reason.
ERMSDCV2 LNA+RMS/LOG converter box projects
Hello udok,
I'm one of designers that AndrewT speak about in it's first answer(hello to him !). I have designed for noise measurements purpose a small box that include :
- A 10-100kHz bandwidth very low noise amplifier (En<500nV).
- A DC-1MHz RMS/DC converter with DC and LOG output (direct in dBv).
(I sell the bare PCB, some are available).
I recommend you to read the first pages of this thread, you will see how it looks. As you want a fully tested/specified device, you will see in page 2 of the thread a very comprehensive measurements results.
Note that many low cost RMS voltmeter are not able to measure precisely noise, because crest factor of the measured signal significantly affect the precision.
(For the MN447A, crest factor <2, precision is +/-10% for 10Hz-50kHz bandwidth in scale above 30µV !)
So, you must look many parameter to be sure that the intrument will be suitable for your need.
Regards.
Frex
Hello udok,
I'm one of designers that AndrewT speak about in it's first answer(hello to him !). I have designed for noise measurements purpose a small box that include :
- A 10-100kHz bandwidth very low noise amplifier (En<500nV).
- A DC-1MHz RMS/DC converter with DC and LOG output (direct in dBv).
(I sell the bare PCB, some are available).
I recommend you to read the first pages of this thread, you will see how it looks. As you want a fully tested/specified device, you will see in page 2 of the thread a very comprehensive measurements results.
Note that many low cost RMS voltmeter are not able to measure precisely noise, because crest factor of the measured signal significantly affect the precision.
(For the MN447A, crest factor <2, precision is +/-10% for 10Hz-50kHz bandwidth in scale above 30µV !)
So, you must look many parameter to be sure that the intrument will be suitable for your need.
Regards.
Frex
Hello All,
Thanks to all for the great links and advices
Something slightly offtopic: I looked at the input circuit of some distortion analyzers (AP Sys1 and VN7721A).
Unfortunately they are all not very low noise...
They all have 100 kOhm input impedance and a instrumentation amplifier.
The instrumentation amp does not hurt the noise performance very mutch,
but if the input attenuator is on (> 5 Volt on AP Sys1) the input signal is taken from the attenuator and has a lot of noise because the source impedance is getting very high.
So AP Sys1, HP8903 and VP7721a are not very useful for lowest noise measurements without a preamplifier in front...
If someone is interested i can post the LTspice file of the AP Sys1 input structure.
Cheers,
Udo
Thanks to all for the great links and advices
Something slightly offtopic: I looked at the input circuit of some distortion analyzers (AP Sys1 and VN7721A).
Unfortunately they are all not very low noise...
They all have 100 kOhm input impedance and a instrumentation amplifier.
The instrumentation amp does not hurt the noise performance very mutch,
but if the input attenuator is on (> 5 Volt on AP Sys1) the input signal is taken from the attenuator and has a lot of noise because the source impedance is getting very high.
So AP Sys1, HP8903 and VP7721a are not very useful for lowest noise measurements without a preamplifier in front...
If someone is interested i can post the LTspice file of the AP Sys1 input structure.
Cheers,
Udo
YES! I have never before seen such a wunderful restoration project before.
The LT1968 seems to be good too. When i have time i will see what IC is in my HP34401a.
Do you know meters which do the RMS calculation digitally with an ADC? Should be mutch more precise.
The 33401A uses the AD637. This chip is Analog Devices' best RMS converter and it is used in many, many other higher-end meters for the past several decades. Keithley 199, 2001 (7.5 digit), 2002 (8.5 digit) to name a few. Mid range meters are more likely to use less expensive chips; the AD637 costs >$30, more than any other IC in the meter including the voltage reference.
The HP 8903B audio analyzer uses several AD637, but not in the measurement circuit per se; they are used in various auto-ranging circuits throughout the machine. The RMS converter for measurements is a discrete design, using op-amps and transistors to do log and anti-log conversion steps. I can only guess, but I suppose this is used instead of the AD637 to get better crest factor response (AD637 handles only up to 3 or so, good enough for most waveforms, but not good enough for noise measurements).
The HP 33410A/33411A (6.5 digit) and the 3458A (8.5 digit) use high speed sampling and calculation to measure RMS. The 3458A also has a traditional RMS converter IC that can be selected instead. The 3458A can measure the RMS value of signals up to 10 MHz by first determining the frequency, then sub-sampling the signal so it can "build up" a view of the waveform over many periods, similar to how a digital sampling oscilloscope can do many GSps "effective" sampling rate rather than real time. Google "5989-4038EN" for 33410A and "hp journal volume 40 number 2" for 3458A.
I have an old (mid 80's) Keithley 194A "high speed voltmeter" which can acquire 32768 of 16 bit samples at 100 kHz (or 8 bit at 1 MHz) and calculate RMS and other values from the acquired data. It does this for two input channels concurrently. The input is a BNC connector instead of banana jacks.
The sigma-delta true RMS converters are more accurate for measuring noise than the root-squaring converters. I compared both with an HP 8403C thermopile True-RMS Meter (0.2%)
The ADI parts are easier to implement -- you can still get them in DIP! The Linear parts are MSSOP.
ADI did introduce a part to compete with the Linear line, the AD8436. Haven't used it yet.
Don't run out and buy an HP 8403C or Fluke 8920A -- quite often the thermopile is damaged. I was fortunate to get one with a current cal sticker. If you see one at a hamfest, make sure to check it out first.
The HP 403 I modified (mentioned earlier) -- I used the development board from Linear because it's all soldered up and you can choose the settling time. The board is something like $50 and was easier to get going --
The ADI parts are easier to implement -- you can still get them in DIP! The Linear parts are MSSOP.
ADI did introduce a part to compete with the Linear line, the AD8436. Haven't used it yet.
Don't run out and buy an HP 8403C or Fluke 8920A -- quite often the thermopile is damaged. I was fortunate to get one with a current cal sticker. If you see one at a hamfest, make sure to check it out first.
The HP 403 I modified (mentioned earlier) -- I used the development board from Linear because it's all soldered up and you can choose the settling time. The board is something like $50 and was easier to get going --
Last edited:
The problem with all the delta-sigma types is bandwidth. The AD8436 part you mention has -3 dB bandwidth of just 1 MHz. The decades-old AD637 is 8 MHz (and in my Keithley 2001's, this is fully utilized, only 1 dB down at 5 MHz, and 3 dB down above 7 MHz). The thermal converters are even better, limited mainly by the front-end attenuator/amplifier that drives the thermopile.
An HP403B does .001V rms f.s. AND has rechargable internal battery power to float the instrument. You can still buy the internal NiCad battery (the battery will last approx 10 years).
Then add a good low noise amp..... DIY or commercial... I use a Kiethley Instruments 823 nanovolt amplifier.
THx-RNMarsh
Then add a good low noise amp..... DIY or commercial... I use a Kiethley Instruments 823 nanovolt amplifier.
THx-RNMarsh
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.