Wenzel Low Frequency Circulator

[uofmrob]

Hi Everyone,
I've been following the EEVBlog channel for ages. It has taught me a lot about what I know about electronics. I'm a PhD chemist by training, but my position is an instrumentation position where I need to be able to test and repair a lot of fancy expensive equipment. I'm currently trying to build my own electron paramagnetic resonance spectrometer at home. Part of the challenge is getting an inexpensive low-frequency circulator. I found this really cool design and I'm trying to learn about op-amps and build my own. I think that it could make a really cool YouTube video!

http://www.qrp.gr/circulator/RFDesign3.pdf

Thoughts?
Rob

[SilverSolder]

Sounds ambitious!  -  what does an electron paramagnetic resonance spectrometer actually do?

[KE5FX]

Hi Everyone,
I've been following the EEVBlog channel for ages. It has taught me a lot about what I know about electronics. I'm a PhD chemist by training, but my position is an instrumentation position where I need to be able to test and repair a lot of fancy expensive equipment. I'm currently trying to build my own electron paramagnetic resonance spectrometer at home. Part of the challenge is getting an inexpensive low-frequency circulator. I found this really cool design and I'm trying to learn about op-amps and build my own. I think that it could make a really cool YouTube video!

http://www.qrp.gr/circulator/RFDesign3.pdf

Thoughts?
Rob

It's a nifty circuit, all right.   An LMH6733 or AD8003 is a good fit.  The same technique gets you a single-ended directional bridge that works down to DC, which occasionally comes in handy around the house.

Sounds ambitious!  -  what does an electron paramagnetic resonance spectrometer actually do?

See page 335 ("A simple magnetic-resonance spectrometer") here.  Same idea works with proton precession.

Edit: actually they both use proton precession.  For some reason I was misremembering the Scientific American project article in connection with paramagnetic electron resonance (EPR).

[uofmrob]

Yup! Same physics behind both. EPRs look at unpaired electron spin while NMR spectrometers look at nuclear spin. Electrons are a bigger magnet so they require higher excitation energy, but smaller magnetic fields to split the quantum spin states. The big research spectrometers that cost $1.6 million dollars can do this stuff among many other things (I wrote the introduction and sections on naturally occurring radicals and metals).

https://scholar.google.com/scholar?q=epr+study+biological+systems+mccarrick&hl=en&as_sdt=0&as_vis=1&oi=scholart#d=gs_qabs&u=%23p%3DdaE_5BGMGPMJ

Mine is just for the sake of seeing anything and learning in the process.

[uofmrob]

I made a mock up on a breadboard and it didn't seem to work. I think I may have had op amps that aren't what I should have used. I have this one now and I soldered it to a 14 pin IC breakout board from Adafruit. It's a quad amp as I'm going for a 4 port circulator. I'm very much not an electrical engineer. I'll probably be back for lots of help from all of you experts!

https://www.mouser.com/ProductDetail/Texas-Instruments/LMH6722MA-NOPB?qs=7lkVKPoqpbZgnjs%252BcyvqkQ%3D%3D

[KE5FX]

Yeah, I don't think an LMH67xx opamp is going to work very well on a breadboard.  With 1 GHz-class parts, your breadboards need to look more like

Yup! Same physics behind both. EPRs look at unpaired electron spin while NMR spectrometers look at nuclear spin. Electrons are a bigger magnet so they require higher excitation energy, but smaller magnetic fields to split the quantum spin states. The big research spectrometers that cost $1.6 million dollars can do this stuff among many other things (I wrote the introduction and sections on naturally occurring radicals and metals).  Mine is just for the sake of seeing anything and learning in the process.

Very cool.  I always wanted to build either the SciAm project or Joe Geller's version but have never seemed to get around to it.  If I had, I'd presumably remember what sort of particles it uses...

[uofmrob]

This was my first prototype using these op amps  (with most of the bare wire replaced with insulated wire before I powered it up once I got the connectivity right).

https://www.mouser.com/ProductDetail/NJR/NJM2137D?qs=54bfbthyeuO6TvPIDHWJog%3D%3D

 Initially, I just powered it directly from my bench power supply, then I watched a bunch of videos and learned more about op amps and figured out that I needed to use my dual tracking supply in serial mode. I'm only shooting for a 25 MHz spectrometer. I've never built something like what you had in the photo. Is the copper plate the ground and then everything else is wired directly above the plate?

[uofmrob]

I built one of the cheap 1-30 MHz antenna tuners off of eBay for the coil matching and it works well. The basic layout will be the source going into port one, then the matching network and saddle coil with the sample in the middle of the coil in port 2, then an rf detector on port 3 and a 50 Ohm load on port 4. When the coil is well matched, there should be hardly any reflection to the detector in port 3. Then, when the magnet field causes the spins to go into resonance, the Q changes and you should see the signal as an increase in the reflected power.

[uofmrob]

At least that's how our big fancy EPRs work. But ours use microwaves (either 9 GHz or 34 GHz) and magnetic fields of 340 mT to 1.2 T.

[KE5FX]

I've never built something like what you had in the photo. Is the copper plate the ground and then everything else is wired directly above the plate?

Yep. It's just a piece of unetched/undeveloped copper clad PC board material.  Do a Google image search for ugly construction or dead-bug construction for inspiration. Some people like to carve islands and 'traces' in the foil with a Dremel tool, or employ additional bits of PCB material as 3D islands or standoffs, but none of that stuff is generally useful or necessary.

ICs built on modern semiconductor processes often contain transistors with frequency limits of 10 GHz or more, even when the IC itself is rated for use at much lower frequencies.  So it doesn't matter what the intended frequency of operation for your project is, stray capacitance and inductance is almost always going to be a concern.

[helius]

The most common key word is "Manhattan construction", I think because the Trinity test electronics were done that way.

[Brumby]

The most common key word is "Manhattan construction", I think because the Trinity test electronics were done that way.

I thought it was because such construction ended up looking like Manhattan in miniature.

[David Hess]

I made a mock up on a breadboard and it didn't seem to work. I think I may have had op amps that aren't what I should have used. I have this one now and I soldered it to a 14 pin IC breakout board from Adafruit. It's a quad amp as I'm going for a 4 port circulator. I'm very much not an electrical engineer. I'll probably be back for lots of help from all of you experts!

Start out with a lower frequency version using lower frequency operational amplifiers.  Fast operational amplifier will require ground plane construction and suitable decoupling.

[uofmrob]

Thanks for all the information everyone!

[uofmrob]

This individual did it this way. I'm not exactly familiar with this, either. Did he just hand drill this material?

http://www.qrp.gr/circulator/index.htm

[uofmrob]

Start out with a lower frequency version using lower frequency operational amplifiers.  Fast operational amplifier will require ground plane construction and suitable decoupling.

David, would you have a particular op amp that your would recommend for this project?

[David Hess]

Start out with a lower frequency version using lower frequency operational amplifiers.  Fast operational amplifier will require ground plane construction and suitable decoupling.

David, would you have a particular op amp that your would recommend for this project?

I would need some idea of the maximum operating frequency.  But even a few 10s of MHz should use ground plain construction as others mentioned; even if the application frequency is low, fast operational amplifiers are not.  Forget using solderless breadboard for this.  An expert might get away with point-to-point wiring on soldered breadboard at lower frequencies but Manhattan or dead-bug style construction will actually be easier to get working.

Maybe someone else has a better source, but pads for Manhattan construction are available here:  http://www.qrpme.com/?p=product&id=SAM

Another limitation is that the operational amplifiers will be required to drive low impedance loads which limits you to parts intended for 50 and 75 ohm applications.  (1) In this case they have to drive their own medium impedance feedback network, the low impedance output network, and a second medium impedance output network all in parallel; so general purpose parts are not suitable.

I would probably start with something low cost which meets the above requirements like the LT1395, which also happens to work at +/-5 volts where less expensive parts are limited to lower voltages.

(1) "Fast" operational amplifiers generally meet this requirement because they need to be able to drive transmission lines.

[uofmrob]

Thanks so much for all the advice!
Rob

[uofmrob]

Another quick question....since I'm looking at building it as a four port circulator, is there a reason to avoid just going with an IC that has four amps in one IC as opposed to four separate ICs? Is there more of a chance of signal leakage with the amps all being in one chip?

[KE5FX]

Another quick question....since I'm looking at building it as a four port circulator, is there a reason to avoid just going with an IC that has four amps in one IC as opposed to four separate ICs? Is there more of a chance of signal leakage with the amps all being in one chip?

With the LMH6722 or LMH6733, crosstalk at 25 MHz will be in the -70 dB to -80 dB range (see figure 9 of the data sheet), so it will not dominate the isolation. 

I'd favor the LMH6733 over the LMH6722 for this application, unless you have a use for the fourth opamp. (edit: I see you want four ports, so, yeah.) 

I'm not familiar with the LT139x family David mentions, but it looks fine as well.  Not a fan of whatever those 'Manhattan' pads are that he mentioned, though, unless you just want to add some stray capacitance in random places.  Current-feedback amps can be twitchy in that regard.

This bridge uses a very similar approach to Wenzel's, as described by Sam Wetterlin (link is a mirror, Sam's site seems to have vanished):





 If you use good opamps and good construction practices, the achievable performance from this class of circuit is pretty awesome.  With the AD8003 that I used, the directivity is still 20 dB at 900 MHz(!).  The LMH6733 performs similarly. 

[David Hess]

Another quick question....since I'm looking at building it as a four port circulator, is there a reason to avoid just going with an IC that has four amps in one IC as opposed to four separate ICs? Is there more of a chance of signal leakage with the amps all being in one chip?

Even if you ignore isolation between the operational amplifiers, the close confines around a dual or quad chip can make isolation between sections more difficult.  Using single parts allows designing one "section" and then replicating it.

Not a fan of whatever those 'Manhattan' pads are that he mentioned, though, unless you just want to add some stray capacitance in random places.  Current-feedback amps can be twitchy in that regard.

I have seen some better DIP and SO8 pads somewhere but the example I linked are what I was able to immediately find online.

[KE5FX]

Even if you ignore isolation between the operational amplifiers, the close confines around a dual or quad chip can make isolation between sections more difficult.  Using single parts allows designing one "section" and then replicating it.

That's for darned sure.

I have seen some better DIP and SO8 pads somewhere but the example I linked are what I was able to immediately find online.

Just solder.

[uofmrob]

 

[uofmrob]

So without any pads, do you just glue the ICs upside down on ground plate and solder to the connectors?I can 3D print supports for the BNC connectors pretty easily and glue them down.

[coppercone2]

yes, I have bought the parts for this circuit a while back (I chose a very high speed op-amp but still in SOIC or MSOP, something with some kind of leads)

I need to etch a PCB. It is a interesting one.

if you are a chemist etch the PCB