EMI problems

Jun 14, 2017 50 Replies

I designed this little photodiode amp and I'm trying to measure its output noise. All we can see on a spectrum analyser is radio and TV stations.



formatting link



The *output* of the amp needs to be around 10 nV/rootHz.



The DUT output is 100 ohm differential, which makes it harder. So we took one of our standard products, an x100 1GHz amplifier, and hacked a wideband transformer into the front end. Then I stuffed the DUT and the X100 amp into a custom-machined shielded enclosure.



formatting link



The DUT is the smaller extruded box. There's enough gain that I can use a scope to measure RMS noise voltage directly.



Those are really nice shield cans, but before you can use them you have to find a way to dispose of the filler materials that they come with.



John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com

You can still get the 'Charles Chips' 1 lb tin. It's a tad bigger if you need more room.

Cheer

A square can would be better. It's a little weird bolting connectors into the round one. This needs research.

John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com

You can get squares... even without filler material!

formatting link

Pere

For smaller stuff, I have a bin full of unused 70-mm nickel-plated steel film cans that I got from Surplus Shed. You put the connectors in the lid. ;)

Clean paint cans work well too--there's a tea place in Maryland that ships their teas in them.

Cheers

Phil Hobbs

Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net

But I want cookies!

John Larkin Highland Technology, Inc lunatic fringe electronics

10nV/Hz sounds hard.

How about short bread.... (searching for cookies in square tins.)

formatting link

George H.

Well, it depends on the gain, the bandwidth, and the photodiode capacitance. Your average 20-dB amplifier with a 50-ohm input is better than that, for instance, and some are _much_ better.

Cheers

Phil Hobbs

Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net

Well in my rather slow (opamp) world Fet opamps start off at ~10 nV/rtHz and noise only gets worse in a TIA. (of course you know all that.)

And right, there are some better Fet opamps these days.

George H.

The overall voltage gain is only 3. I used two parallel gain paths, resistively summed at the output, to reduce noise by sqrt(2).

Shortbread is good; I'll try some. I hope that can is solderable.

John Larkin Highland Technology, Inc lunatic fringe electronics

Yikes!

formatting link

OK, you might have to add another box and build a PI filter at the power entry. With those DAMN transmitting antennas so close, you need SERIOUS filtering. There's another power entry that is powering the X100 amp that looks to NOT have a feedthrough cap. That likely is the real culprit.

Jon

Who installed those vents? Did they smoke some funny-smelling cigarettes?

If you want a nice-looking smaller box there is also Fishermen's Friend. You can get cans with a more international flair:

formatting link
's_Friend#/media/File:Fisherman%27s_Friend_Tin_Taiwan.jpg

Be careful. The dad of a friend popped one and then fell asleep on the couch. When he woke up the cheek pointing towards the pillow had become numb like after a major dentist session.

Regards, Joerg http://www.analogconsultants.com/

That's the roof of the San Francisco Ballet School. Artsy types. Sometimes I can hear them practising. One of my engineers dances Flamenco there, conveniently next door.

Some people like the Altoids cans for small stuff.

John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com

Den torsdag den 15. juni 2017 kl. 21.47.56 UTC+2 skrev John Larkin:

formatting link

formatting link

I wish we had a nice dance hall here because my wife and I did ballroom dancing in Europe. The only one we had for a few years was in a simple industrial building with fluorescent lighting, linoleum floor, having the romantic level of a Soviet stairwell. In Europe they had wet bars, tables, seats, mood lighting and most of all a polished wooden dance floor.

Once in a while we go to Cafe Marcos in Lotus when they have live country music. There we can dance. The food is IMHO subpar but they always have very good beers on tap.

Some folks consider those too pedestrian because too many people carry them :-)

Regards, Joerg http://www.analogconsultants.com/

Much cheaper in regular retail packaging and then you get the mints in there "for free":

formatting link

Regards, Joerg http://www.analogconsultants.com/

We are staring up at 22 megawatts that pack the entire spectrum. In a wooden building. I bet I could light an LED with a small antenna.

I have a cap across the amp power connector on the inside. It's from a switching wall wart! The DUT power (+-6V) comes through 3.3nF feedthru caps, ebay bag-o from Lithuania.

It seems to work! I'm seeing 15 mV RMS reported by the scope. Correcting for the X100 amp, and the 200 MHz scope bandwidth, the 100 MHz noise is 106 uV RMS, 10.6 nV/rtHz. The noise spectrum looks flat up to the scope's 200 MHz rolloff, so the 100 MHz correction is mostly honest.

I had to bolt both boxes to the bottom of the cookie can, for good grounding.

Should we ship a Danish Butter Cookie can with each production unit?

John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com

Den torsdag den 15. juni 2017 kl. 22.56.01 UTC+2 skrev John Larkin:

in gold?

formatting link

What is the power consumption of the devices. Would an internal battery be an option, thus no need for power entry ?

Of course a good RF contact between the cookie can and the lid is critical. Put finger stock around the edge. Alternatively secure the lid to the can with screws every few centimeter around the perimeter of the lid, this should be good to a few GHz.

Am 15.06.2017 um 21:27 schrieb Joerg:

formatting link

;-)

Jorgen

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required