DIY low frequency noise meter and some measurement result of voltage references

[MiDi]

The LNA noise floor looks like a little on the high side. Is this measured open circuit or short circuit ? The open circuit noise is expected to be higher and does not apply to the later use.

The scope adds too, my DS1054Z @ 20MHz BW & HiRes adds ~35nV (referred to x10k LNA input) - without BW limit and normal aq mode this value is higher.

Without proper shielding of LNA I got ~100nV extra noise with weird artefacts.

[Andreas]

Hello Midi,

is it possible on the 1054z to add a 1kHz software filter additionally to the 20 MHz BW limiter?
On my scope I can either select HiRes (generating 4 bits more resolution) or a programmable low pass filter.

With best regards

Andreas

[MiDi]

AFAIK nope, only option is HiRes and 20MHz BW-limit.
I think the 20MHz are software limit, like the 50MHz software limit with option for 100MHz.

[RandallMcRee]

Hello,

for the results:
how often did you repeat each measurement?
how was the standard deviation of the measurement?
how long was each measurement (10 s or 100 s)?

For 10s measurements I typically have a stray or +/-20 .. 30%.
And this under ideal conditions (DUT + amplifier in cookies box).

I usually do 15-20 repeated measurements and calculate average + standard deviation.

with best regards

Andreas

I should have said something in the original post--I took Vp-p readings from Rigol 1054Z set to 1s/div for a 12 second measurement, total. I then cherry-picked measurements  that did not have me walking about. So very unscientific, but when near the gear everything goes up by a factor of two to three. I took seven to nine Vp-p measurements by hand and averaged them.
The capacitor forming did not seem to be an issue (It does take quite a while to switch from one voltage to another). For example, this morning I am repeating the 10V measurement and the LNA has been connected since yesterday. The readings are perhaps slightly lower. The major uncertainty is me being in the room and air drafts, as well as the LNA noise floor, which as has been pointed out is definitely a factor of two too high.

A scope shot of this mornings readings of 10V output. DC value is 9.9999973 volts (maybe--not cal'ed). Second scope shot is "drafty". I don't see those three-five second excursions from across the room. If I take the cover off the enclosure--oh my, threw the roof! As expected I guess.

[RandallMcRee]

Very interesting results Randall

Could you please elaborate on your 7xLTZ-Ensemble and the 5V/10V-ensemble? Did you connect the 7 LTZ-references via the 2k-input-resistors to get the 5V/10V?
Nobody answered yet how much the soakage time of the LNA-input-caps matters/how big the noise-difference is; did you wait long enough after connecting the 7.1/5V/10V-ensemble to the LNA?
. . .

Schematic for the ensemble part of the circuit attached. It ties in to what was previously posted. This shows the OPA140 low-pass filter I am using. The reason for the low-pass filter (optimized for low noise using Analog Filter Designer tool) is because I have this idea to compensate for LTZ anomalies e.g. long-term drift by detecting the difference between the LTZ input at the filter and the filter output. Pretty crazy. Not sure if this will ever work, but I'm including this in the interest of full disclosure because the OPA140 is not a low-tempco creature. The inputs to the filters are six PX (jason's PX LTZ1000 reference) and one KX (TiN's board). Two LTZ1000A and five non-A.

The Pipelie/Zlymex LNA has a "ready-to-go" signal. It appears to work. Yes, it takes quite some time.

[RandallMcRee]

The LNA noise floor looks like a little on the high side. Is this measured open circuit or short circuit ? The open circuit noise is expected to be higher and does not apply to the later use.

The scope adds too, my DS1054Z @ 20MHz BW & HiRes adds ~35nV (referred to x10k LNA input) - without BW limit and normal aq mode this value is higher.

Without proper shielding of LNA I got ~100nV extra noise with weird artefacts.

I measured closed circuit. As you both imply, seems like I need to put the LNA into the enclosure. Sigh.

[MiDi]

I measured closed circuit. As you both imply, seems like I need to put the LNA into the enclosure. Sigh.

Think this large unshielded loop area with clips from LNA input/BNC to DUT would introduce a lot of pick up noise.

LNA input to DUT should be as short as possible and fully shielded, best would be shielded twisted pair (I use cat 5e patch cable).

For your setup you could first try:
If possible put bare BNC direct onto jack of DUT, if not twist the clip and remaining wires as tight as possible together - should improve pick up noise and nearby sensitivity.

[Andreas]

I measured closed circuit. As you both imply, seems like I need to put the LNA into the enclosure. Sigh.

Hello,

your signal on the scope above looks a bit fuzzy for a 10 Hz bandwidth limited signal.
The last time I had this there was a lot of mains hum on the signal by a transformer about 0.5 m away.
When I did a FFT of the noise, the 50Hz mains frequency was relatively dominant.

So it is always a good idea to make a noise floor measurement and a FFT of it before starting measurements.

with best regards

Andreas

[branadic]

Can agree that noise can improve if reference is well shielded from ambient hum during measurement. Attached a comparison between reference at LNA on the lab table, but connections as short as possible (a) and reference with LNA inside a cookie box (b). In both cases there is no 50Hz hum visible. However, there is a clear difference.

-branadic-

[RandallMcRee]

Folks,

Thanks for the suggestions, they have helped immensely.

I put the LNA scope cable into the enclosure and simply exited through hole in enclosure. This decreased the noise floor and air drafts ceased to be a problem. Simple and effective. I also found that the scope was triggering on the wrong channel, and this was causing fuzzy pics and slightly incorrect readings, so new pics and new measurements...

This time, to be more rigorous I took seven to nine sequential readings and averaged them, no cherry-picking  (No air drafts, as I mentioned).  Representative pics of each output attached, all on 2mV/div scale

Noise floor: 0.16uV p-p
7V output: 0.23 uV p-p
5V output: 0.34 uV p-p
10V output: 0.59 uV p-p

Thanks again for the help,
Randy

[branadic]

Looks better now, but there is still some variation in noise visible. However, a raw estimation of the LTZ1000 noise: 1,2µVpp/sqrt(7) = 0,454µVpp proves that you are in the right ball park.
In general, activate 20MHz bandwidth limit on your scope, decrease sampling rate, no need for 1MSps, set the timebase to 10s per divison and watch noise for several 10 seconds. This way you can see if it's really noise or noise plus some additional thermal variation and/or hum.

-branadic-

[Andreas]

Noise floor: 0.16uV p-p
7V output: 0.23 uV p-p
5V output: 0.34 uV p-p
10V output: 0.59 uV p-p

Ups.

The noise floor of the LNA should be at least a factor 3-5 below the values that you want to measure to get a error below 10%.
Note that noise adds by square-law. (sqrt of squares)

with best regards

Andreas

[RandallMcRee]

Branadic, here are one minute noise measurements for the 10 volt output...

If I read these correctly, these show that, yes, it is noise.

Randy

[David Hess]

One of the tricks I used when I last designed and built a very low noise DC to low frequency amplifier was to measure the output noise with the input shorted and then with the input connected across a known resistance which produces a known noise.  This provided a sanity check on my noise measurement methods.

[Gerhard_dk4xp]

Yes. You don't even need to know the exact gain.
60 Ohms equals 1nV/rtHz absolute at room temperature.

Gerhard

[splin]

Trollocks; I thought this had posted a couple of days ago but just realised it didn't.  It's been a bit overtaken by events, but since I went to the trouble I may as well post it anyway.

Presumably all the later measurements are with the extra caps added which limit the bandwidth to around 1 to 2 Hz - which will significantly reduce the resistor, op-amp voltage and current noise (none of which should have much 1/f noise), but rather less of the 1/f noise from the LTZ1000 references. But I can't be bothered to try and calculate the impact.

Recently made some measurements of my equipment using a Pipelie LNA (s/n 1157A). Values below are RTTI. Should be comparable to Zlymex results column 6.

Fluke 731B #1 avg = 2.5 uV p-p
Fluke 731B #2 avg = 1.7 uV p-p

Malone DMMCheck 5V output avg = 26.8 uV p-p

Ensemble of 7 LTZ1000 7.115V output = .39 uV p-p
Ensemble 5V output (custom resistive divider) = .35 uV p-p
Ensemble 10V output (opamp X2) = .87 uV p-p

My measurement setup is pretty basic. In particular, I have not shielded everything very well and I notice that readings increase when I am nearby in the room. Still, the Fluke 731B measurements suggest that I am not wildly wrong.

However it seems that my circuits do not preserve the low noise of the ensemble very well. Straightforward ratios would suggest
7V output = 1.2uV/sqrt(7) = .45 uV <ideal>    <--did ok here!
5V output = 0.39/1.43 = 0.27uV <ideal>        <-- marginal
10V output = 5V*2 = 0.55 uV    <ideal>        <-- very marginal, but good compared to zlymex table right?

Assuming ensemble measurment is at output of U23A, noise over .1 to 10Hz:

1) 7.1 to 5V stage:
a) .39/1.43 = .27uV
b) U23B, Vn = 117nV pp x noise gain (1x)
c) 2112//5000 = 1484 ohms = 102nV pp thermal noise
d) U23B In- 16pA x 2k ohms = 32nV pp
e) U23B In- 16pA x 1484 ohms = 24nV pp

RSS noise = .32uV pp. Not too far from .35uV from your earlier test - it's not clear if your revised measurements are including C27 and C28, so I ignored those.

2) 5 to 10V stage:
a) 5V stage noise .35uV pp
b) U24 Vn 117 nV pp x gain (2x) = 234nV pp
c) 10k//10k = 5K ohms = 188nV pp x gain (2x) = 376nV pp
d) U24 In+ x 10 ohms - negligable
e) U24 In- 16pA x 5k x gain (2x) = 160nV p

RSS noise = .84uV; again not far from your .87uV.

Better, from a noise perspective, would be to generate the 10V from the 7.1V avoiding the added noise of the 7.1V to 5V stage. The drift of the 10V output relative to the 5V would likely be greater as the two stages can drift in opposite directions, but the 10V drift would be lower than the current scheme (given similar performance scaling resistors) as a) the ratio is lower and b) the 7.1 to 5V stage drift is eliminated.

Using lower value resistors would help - but DSMZs don't come cheap so filtering with added caps is a pragmatic solution. But when you are looking at low level, very low frequency noise signals, should you worry about dielectric absorption and temperature dependence of those caps adding their own 'noise'? (ie. not actually noise, but hard to distinguish changes from the low frequency noise and drift of the references).

[Kleinstein]

For the reference circuit one does not have to worry about dielectric absorption (DA) very much, as the voltage the caps see does not change very much. The DA in film caps has a time constant in the 10-100 s range - so waiting that long is reasonably easy. It is quite a bit longer for electrolytic caps however. But still those high accuracy ref usually needs some warm up anyway.

The DA of an electrolytic cap at the LNA input however can be at least annoying, as it would take longer to settle when changing the voltage to test. If there are different long time constants involved (I don't know, but well possible) it can be even difficult to see when settling has finished.

The other point can be a temperature or mechanical effect on capacitors that can cause a voltage change, a little like thermal EMF fluctuations. I would expect this mainly with the electrolytic caps.

For the 5 to 10 V amplifier, there could be some extra noise from interaction between the AZ OPs.  Especially those ADA4522 in separate cases can have a clock the is close, but usually not in sync.  As it is a similar frequency residual spikes from one AZ OP can disturb other AZ OPs of similar type.

[RandallMcRee]

So I built a second 7-to-10V circuit using Vishay dividers and an ADA4522-1. The input is directly from the LTZ1000 ensemble.

The noise is lower, around 0.3uVp-p.  Pics attached.

[3roomlab]

has anyone noticed this blog? (fresh new JFET kids to make more LNA anyone?)
http://tech-blog.sblo.jp/article/183433346.html



Mr tzukasa yasui's noise test on the new N Jfet (June 2018).
the 5908 are duals in a SOT26, it is going to be a very hot JFET, the singles version is apparently 2SK3557

*bonus*
http://tech-blog.sblo.jp/article/180081088.html

[pelule]

Here the link to the english version: http://tech-blog-en.sblo.jp/article/183391570.html

[chuckb]

It appears some choppers do not completely eliminate the basic flicker noise. The below link shows the limits of my favorite op amp, the ADA4522. I will need to find a new favorite. The ADA4528 looks good but it's a 5V part. This results are similar to what I experienced when using 16 ADA4522s in parallel for a 60dB DSA preamp. I had a nice 1.5nV / rt Hz white noise but I still had flicker noise below 1Hz.
https://ez.analog.com/amplifiers/operational-amplifiers/f/q-a/102479/ada4522-1-f-noise

This other paper by David Hoyland (2016) documents the flicker noise performance of several other chopper opamps below 0.1Hz. It also has the current noise spectrums.
https://dcc.ligo.org/public/0126/T1600206/001/Opamp%20Noise%20Test%20Results.pdf

If you need performance below 0.1Hz don't assume just any chopper will remove the internal flicker noise.

So far -
flicker noise free
ADA4528 (5v)

Low Frequency Flicker noise corner, less than 0.1Hz
CS3002
ADA4522
OPA188
OPA180

Has anyone tested other choppers below 0.1Hz?

[branadic]

Thanks for sharing. I wonder how LTC2057 competes in this respect.

EDIT: Maybe similar to the latest LTC2058, with the input voltage noise spectrum given from 0.1Hz - 10MHz on page 7, which by the looks of it appears to be flat.

-branadic-

[David Hess]

The ADA4528 link identifies the likely problem, noise from thermocouple junctions.  Besides using dummy parts to equalize the number of junctions and their potential, drafts need to be eliminated to maintain an isothermal environment.

Even the precision of amplifiers like the OP-07 can be compromised by drafts so at least air baffles should be used where this is important.

Care should also be used to not load the amplifier output because this will limit open loop gain and cause drift from die heating; if necessary, an external transistor or integrated buffer is very useful for this.  This should be less of a problem with chopper stabilized amplifiers but I would do it anyway.

[chuckb]

I have reached out to TPeter through AD Engineer Zone and invited him here to discuss the results.

My impression was that the test board had an ADA4522 and also an ADA4528 on it that were powered from the same power supplies. The amplifier outputs went to two channels of a spectrum analyzer. A thermal EMF issue should be an equal opportunity noise source.

I will set up a test myself in a few week when I'm back from vacation. I have also started to do some Chopper input current spike evaluations with a 100MHz BW transimpedance monitor. I definitely can see the current spikes in detail. More later.

[Kleinstein]

Thermal EMF combined with temperature fluctuations is one source of 1/f like noise. Here the often higher power dissipation in the higher supply OPs can be a factor.
I don't think that dummy junctions will help much, as it would not be clear they see the same temperature. It is more about using a good thermal layout.

For the current noise, one can expect some flicker part, as there is some internal charge injection, that does depend on the supply voltage or internally regulate supplies. So there is some kind of supply voltage to bias/offset current coupling.
For the current noise, also details of the circuit could have an effect, like parasitic capacitance at the inputs. Another point is that the current noise at the 2 inputs is likely partially correlated. So it depends where and how to measure.

The flicker noise shown in the links so far still looks relatively low. So in most application it is still not such a concern.

With several AZ OP in parallel I would be slightly concerned of possible interaction / inter-modulation effects.