Hi!
I own a "standard" oscilloscope that is capable of measuring down to mV.
But I want to measure uV or even better nV. Who doesn't, right?
... and I do not want to buy an expensive equipment. Typical DIY situation. What to do?
I guess I need to make an low noise, high precision amplifier, i.e. use such Op-amp w/ good PSU (e.g. battery).
My question is how have you solved this problem?
Thanks,
Matej
I own a "standard" oscilloscope that is capable of measuring down to mV.
But I want to measure uV or even better nV. Who doesn't, right?
... and I do not want to buy an expensive equipment. Typical DIY situation. What to do?
I guess I need to make an low noise, high precision amplifier, i.e. use such Op-amp w/ good PSU (e.g. battery).
My question is how have you solved this problem?
Thanks,
Matej
search this Forum.
Here is the link to a 40dB/60dB gain stage that is battery driven (isolated) using low noise opamps and as an alternative the relatively cheap opa228
http://tangentsoft.net/elec/lnmp/
There's also an Audio Express article from 2007 by Dennis Colin doing similar.
Here is the link to a 40dB/60dB gain stage that is battery driven (isolated) using low noise opamps and as an alternative the relatively cheap opa228
http://tangentsoft.net/elec/lnmp/
There's also an Audio Express article from 2007 by Dennis Colin doing similar.
There is also a mistake in Dennis Colin's article -- in the original the opamp feedback resistor was specified as 1k5 -- the actual value is 51k1.
You want low noise, a.c. coupled and low input capacitance -- the correction appears on page 50 of the June '07 issue of AX.
You should also check out the data sheet for the TL431 from Texas Instruments which has some noise measurement techniques, and the following application notes from Linear Tech:
AN83 -- "Performance Verification of Low Noise, Low Dropout Regulators" and
AN85 -- "Low Noise Varactor Biasing"
If you need to measure really, really low voltages and currents, then there are a bunch of papers in the journal "Superconducting Science and Technology -- which you can access via JSTOR (or at a University Library).
Bob Pease has some interesting stuff on the foibles of teflon amongst his whitepapers at National Semi.
I keep an old analog scope (5223) with a differential amplifier on hand for looking at noise.
You want low noise, a.c. coupled and low input capacitance -- the correction appears on page 50 of the June '07 issue of AX.
You should also check out the data sheet for the TL431 from Texas Instruments which has some noise measurement techniques, and the following application notes from Linear Tech:
AN83 -- "Performance Verification of Low Noise, Low Dropout Regulators" and
AN85 -- "Low Noise Varactor Biasing"
If you need to measure really, really low voltages and currents, then there are a bunch of papers in the journal "Superconducting Science and Technology -- which you can access via JSTOR (or at a University Library).
Bob Pease has some interesting stuff on the foibles of teflon amongst his whitepapers at National Semi.
I keep an old analog scope (5223) with a differential amplifier on hand for looking at noise.
I might comment a little about low noise measurement. There are 2 important factors:
1. a low noise input gain stage
2. a defined bandwidth, so that only the audible noise is measured and defined.
The 797 is an excellent choice, but be careful of the feedback resistors. They should be as low value as possible. Even 100 ohms is too much.
The reason oscilloscopes generally don't have too much sensitivity, is that they have such high bandwidth. This increases with the square root of the total bandwidth. Therefore a 100MHz scope will be 100 times noisier than a 10,000 Hz bandwidth amp, with the same input stage.
1. a low noise input gain stage
2. a defined bandwidth, so that only the audible noise is measured and defined.
The 797 is an excellent choice, but be careful of the feedback resistors. They should be as low value as possible. Even 100 ohms is too much.
The reason oscilloscopes generally don't have too much sensitivity, is that they have such high bandwidth. This increases with the square root of the total bandwidth. Therefore a 100MHz scope will be 100 times noisier than a 10,000 Hz bandwidth amp, with the same input stage.
The Tektronix 7a22n and 5a22n have adjustable bandwidth for this reason. 5a22's go for less than the cost of a couple Budweisers...problem is finding a mainframe with a decent phosphor.
Another good paper -- from Burr Brown (on TI's website) is "Noise Analysis of FET Transimpedance Amplifiers"
Another good paper -- from Burr Brown (on TI's website) is "Noise Analysis of FET Transimpedance Amplifiers"
Ah yes, the 7a22 my most missed lab instrument.
Current record (I've seen) for low noise .065nV/rt-Hz. I have not read the entire article, but if the input transformer has to be tuned for a constant 3dB noise figure then this is less impressive. In any case this example is built for 3dB noise figure on a .5 Ohm source (SQUID)! Probably nice for ribbon mics too.
http://www.picovolt.com/win/elec/articles/Lepaisant_preamp-xfmrs_RSI.pdf
Current record (I've seen) for low noise .065nV/rt-Hz. I have not read the entire article, but if the input transformer has to be tuned for a constant 3dB noise figure then this is less impressive. In any case this example is built for 3dB noise figure on a .5 Ohm source (SQUID)! Probably nice for ribbon mics too.
http://www.picovolt.com/win/elec/articles/Lepaisant_preamp-xfmrs_RSI.pdf
scott wurcer said:
I worked with somebody on a project for NQR measurement, and we built a pre-amp that delivered about 0.3nV /rtHz, using very large geometry custom JFETs that were left over from somebody else's project. The frontend JFET ran at about 50mA. The current noise was pretty good too.
The circuit shown reminds me a lot to the ones found in European Patent # EP0157187 by Studer (of Switzerland) from 1985, although the goals seem to be a little different, aiming to lowest distortion and best performance from smaller xformers running in current mode (known as "zero field technology"). But good noise performance was also a design goal, obviously.scott wurcer said:
- Klaus
The problem here is a matter of communication.
IF a serious audio amateur wants to make a gain step-up device that is low noise and practical, we can't play with exotic transformers, and even cap dividers are a bit exotic.
Any resistance that is in SERIES with the input, will add its noise.
Now, let us say that Scott's AD797 has an equivalent noise of 60 ohms. Then another 60 ohms in series with the input will add 3dB to the noise. 100 ohms will do even worse, which is the case in one of the examples.
Another factor is that a non-inverting input can be just as quiet as an inverting input, AND you can have high input impedance as well. This is more practical.
IF a serious audio amateur wants to make a gain step-up device that is low noise and practical, we can't play with exotic transformers, and even cap dividers are a bit exotic.
Any resistance that is in SERIES with the input, will add its noise.
Now, let us say that Scott's AD797 has an equivalent noise of 60 ohms. Then another 60 ohms in series with the input will add 3dB to the noise. 100 ohms will do even worse, which is the case in one of the examples.
Another factor is that a non-inverting input can be just as quiet as an inverting input, AND you can have high input impedance as well. This is more practical.
For maximum versatility you could use an instrumentation amplifier. There are many already available or you could build this little circuit. The FET's need to be matched or you need a big AC coupling cap in series with the gain resistor which goes between the sources of the input FET's.
Attachments
In case you were wondering:
* BF862 SPICE MODEL MARCH 2007 NXP SEMICONDUCTORS
* ENVELOPE SOT23
* JBF862: 1, Drain, 2,Gate, 3,Source
Ld 1 4 L= 1.1nH
Ls 3 6 L= 1.25nH
Lg 2 5 L= 0.78nH
Rg 5 7 R= 0.535 Ohm
Cds 1 3 C= 0.0001pF
Cgs 2 3 C= 1.05pF
Cgd 1 2 C= 0.201pF
Co 4 6 C= 0.35092pF
JBF862 model parameters:
.model JBF862 NJF(Beta=47.800E-3 Betatce=-.5 Rd=.8 Rs=7.5000 Lambda=37.300E-3 Vto=-.57093
+ Vtotc=-2.0000E-3 Is=424.60E-12 Isr=2.995p N=1 Nr=2 Xti=3 Alpha=-1.0000E-3
+ Vk=59.97 Cgd=7.4002E-12 M=.6015 Pb=.5 Fc=.5 Cgs=8.2890E-12 Kf=87.5E-18
+ Af=1)
* BF862 SPICE MODEL MARCH 2007 NXP SEMICONDUCTORS
* ENVELOPE SOT23
* JBF862: 1, Drain, 2,Gate, 3,Source
Ld 1 4 L= 1.1nH
Ls 3 6 L= 1.25nH
Lg 2 5 L= 0.78nH
Rg 5 7 R= 0.535 Ohm
Cds 1 3 C= 0.0001pF
Cgs 2 3 C= 1.05pF
Cgd 1 2 C= 0.201pF
Co 4 6 C= 0.35092pF
JBF862 model parameters:
.model JBF862 NJF(Beta=47.800E-3 Betatce=-.5 Rd=.8 Rs=7.5000 Lambda=37.300E-3 Vto=-.57093
+ Vtotc=-2.0000E-3 Is=424.60E-12 Isr=2.995p N=1 Nr=2 Xti=3 Alpha=-1.0000E-3
+ Vk=59.97 Cgd=7.4002E-12 M=.6015 Pb=.5 Fc=.5 Cgs=8.2890E-12 Kf=87.5E-18
+ Af=1)
Hi, all!
Thanks for the input, I really appreciate it.
Simple and very effective way for me (as you have suggested) is to go with the low-noise op-amp (AD797 or LT1028), non-inverting, keeping Rs low (< 60ohm), +/-9V battery.
This requirement of low input impedance annoys me a little bit (right now I cannot tell what I will measure in the future; I might be over-concerned). Should a discrete JFET (nx 2SK170) input stage + op-amp solve this problem or not?
Want to know out of curiosity - always want to learn. My thinking: FET inputs helps (if low noise due to 1/f), but how much?
I would like to avoid instrumental amplifiers since usually I will do single ended measurements (=> always connecting one input to a GND).
ps I would like to have good 1/f behavior; 0.1Hz to 10Hz Vpp < 0.1uF (clock supplies require good 1/f PSUs; 50nV as I've learned).
But again, everything more than AD797/LT1028 adds complexity - is it worth for my purpose?
I want to keep this a one day build project (also useful for many other hobbyists).
(Although, I've already spent more than couple of days learning lots of 1/f stuff
)
Thanks again,
Matej
Thanks for the input, I really appreciate it.
Simple and very effective way for me (as you have suggested) is to go with the low-noise op-amp (AD797 or LT1028), non-inverting, keeping Rs low (< 60ohm), +/-9V battery.
This requirement of low input impedance annoys me a little bit (right now I cannot tell what I will measure in the future; I might be over-concerned). Should a discrete JFET (nx 2SK170) input stage + op-amp solve this problem or not?
Want to know out of curiosity - always want to learn. My thinking: FET inputs helps (if low noise due to 1/f), but how much?
I would like to avoid instrumental amplifiers since usually I will do single ended measurements (=> always connecting one input to a GND).
ps I would like to have good 1/f behavior; 0.1Hz to 10Hz Vpp < 0.1uF (clock supplies require good 1/f PSUs; 50nV as I've learned).
But again, everything more than AD797/LT1028 adds complexity - is it worth for my purpose?
I want to keep this a one day build project (also useful for many other hobbyists).
(Although, I've already spent more than couple of days learning lots of 1/f stuff
)Thanks again,
Matej
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.