220pv/sqrt(Hz) preamp for measurements

[gerhard]

Oooohps, pV, not nV. Why can't I edit the header???
-------

I have designed a very low noise preamplifier that can be used
to characterize regulators etc built around LT1028/AD797 & friends
without obscuring their true performance.

The description is at:

< http://www.hoffmann-hochfrequenz.de/downloads/lono.pdf >

(3.8MB pdf)

There won't be kits but Gerber files and maybe some leftover FR4 boards.


regards, Gerhard

 

[1audio]

Good effort, and pure brute force. I saw a germ of the idea in a Yamaha tuner where they paralleled two haves of an opamp to get lower noise. Since the opamp was on a common substrate the literally tied the pins together.

To get similar results I had to resort to an input transformer (ebay sourced geoformer) and an opamp. It works but the transformer would be elusive to find and has a high frequency limit.

Have you been able to characterize the low frequency noise? (100 Hz, 10 Hz etc.)

Did you consider a servo for the DC offset/low frequency limit?

I like the 60 Ohm calibration idea. It does not take into account the internal noise but the difference is enough that the error is inconsequential.

 

[Speedskater]

If you click on that red triangle under your name. You can request that the moderator edit the title.

 

[gmorita]

Very nice. I had to design a 44dB pre-amp with a discrete low noise JFET in order to measure noise from 10Hz to 10MHz. I get nearly the same performance (300pV/rt-Hz) at 1KHz but my 1/f corner is around 100Hz and the noise increases to 100nV/rt-Hz at 1Hz. A colleague of mine built one almost exactly the same as yours and got similar numbers.

 

[jackinnj]

Thanks for sharing!

 

[geraldfryjr]

Very Cool !!!

I am just now getting into some of these very low noise devices!!!
They sure are a far cry from the ole' TL072 and NE5532/34 standards !!!

I wish I had the tools to properly measure them with, But they do still look nice using my 30year old Hitachi 40Mhz scope though. 🙂

Cheers !!!

jer 🙂

 

[iko]

Thread title fixed

 

[jackinnj]

They sure are a far cry from the ole' TL072 and NE5532/34 standards !!!
jer 🙂

NE5534 remains a standout.

 

[gerhard]

Good effort, and pure brute force.

Thanks, I like electronics with predictable performance. No Voodoo parts,
just Digi-Key, design the board, populate it, and it must do. No trimmers,
no hand-selected parts etc.

> Have you been able to characterize the low frequency noise? (100 Hz, 10 Hz etc.)

Not yet. From the op amp data sheet, the 1/f corner should be 30 Hz or so.

> Did you consider a servo for the DC offset/low frequency limit?

No. I want to characterize reference voltages, supplies for quality
oscillators and tuning voltages. When I had a servo that could remove
5V, then I would shift the noise problem to that servo. Even op amp
offset & bias are much larger than the noise voltage to investigate.

> I like the 60 Ohm calibration idea. It does not take into account the internal noise but the difference is enough that the error is inconsequential.

It even shows exactly what we need to know: When we applied 1 nV/sqrt(Hz)
at the input of the preamp: Where would the horizontal line on the
spectrum analyzer be?

regards, Gerhard

 

[1audio]

I want to characterize reference voltages, supplies for quality
oscillators and tuning voltages. When I had a servo that could remove
5V, then I would shift the noise problem to that servo. Even op amp
offset & bias are much larger than the noise voltage to investigate.

I was thinking of the second stage and driving a 50 Ohm load. You either need to cap couple or use a magic reference supply for the first stage. Few will be using a GHz capable spectrum analyzer for this so the 50 Ohm requirement is less of an issue using a sound card or low frequency spectrum analyzer.

This is my transformer variation. Simple but based on an unobtainable transformer. The input cap needs to be much bigger than shown. The equivalent input noise resistance is around 5 Ohms.

 

[jackinnj]

This is my transformer variation. Simple but based on an unobtainable transformer. The input cap needs to be much bigger than shown. The equivalent input noise resistance is around 5 Ohms.

Geophysical transformer from Triad perhaps?

Commercial version from Stanford Research Systems:

SR554 - Transformer Preamplifier

 

[agdr]

gerhard - impressive! I hadn't heard of the ADA4898 before. I've just been comparing it to the LME49990 datasheet. The voltage and current noise graphs on the two are remarkably similar, but you get a dual package with the ADA part vs. solo for the LME. Looks like the ADA wins the DC offset numbers but the LME has higher open loop gain.

This will be a fun build. 🙂

 

[Rax]

Gerhard - is a PCB available for purchase?
Thanks!

 

[31697B]

I would like a pcb also.

-RM