We are using RF radiation in between 0.3 and 100MHz for evaporative cooling of Rb87 atoms in vacuum. For example, in one application we ramp our frequncy in a three-part linear ramp from 80 to 1MHz during 8 seconds.

What would be the ideal antenna design for this frequency range?

Currently we are using a coil antenna with 3 windings of copper wire (1mm diam.) with a diameter of 4cm. We have no R or C present, our antenna is directly connected to an RF amplifier (Mini-Circuits ZHL-3A +).

The efficiency of our cooling depends on the RF power we can deliver to our atoms (Rb87). So far, we have measured a strong dependence of efficiency vs. distance (the closer the better, roughly linear dependence), but we are limited to ~3cm distance by our vacuum chamber. We have not observed any strong efficiency dependence on the angle of the antenna axis w.r.t. to the atoms. It seem the antenna is radiating, roughly speaking, homogenously in all directions (c.f. near field?).

  • \$\begingroup\$ Does the electrical field or the magnetic field initiate the "cooling"? In near-fields the orthogonality has not yet formed, the propagating component is much weaker than Efield or Hfield. Again, what drives the cooling? \$\endgroup\$ Jul 4, 2017 at 16:59
  • \$\begingroup\$ @Quasilattice any thoughts about the answers so far? \$\endgroup\$
    – Andy aka
    Jul 6, 2017 at 13:03

2 Answers 2


Yes it certainly sounds like a near-field transmission with the main energy being delivered as an alternating magnetic field and not a proper EM wave. The coil you describe sounds to me like it would have an inductance of about 2 uH and, to improve efficiency (at the low end) you could use parallel tuning.

I estimate that about 140 nF of tuning from good ceramic capacitors would tune it at 300 kHz but this doesn't help much at the higher frequency (100 MHz) where a tuning capacitance of barely 1pF would be needed and of course this is not practical because the self capacitance of your 3 turn coil would exceed this.

So that's the first point I note - at the higher end of the scale your antenna is likely to be quite problematic at delivering energy to the Rubidium isotope.

I would therefore suggest a much smaller inductance and a single turn coil. With 1 turn I estimate the inductance to be about 200 nH or thereabouts and this would parallel tune at 100 MHz with a capacitance of about 13 pF.

But, at the lower frequencies you end up with problem because you would need to use a bank of RF relays to progressively switch-in a bunch of capacitors up to about 1.5 uF. It can be done but it needs special care AND to achieve frequency alignment with the tuning I would want to build the antenna around a circuit so that it self oscillates i.e. it is always running on tune.

This is the approach I would take but, I would also consider splitting the energy delivery into smaller chunks so that the electronics could be tailored for two or more frequency bands i.e. you have two or more antennas.

Of course, I have no idea how convenient this would be.

  • \$\begingroup\$ Thanks Andy for your comprehensive answer! In fact, we also tried an antenna with just one winding but found no obvious improvement. I think the problem is indeed the large frequency range. Relay circuits are not feasible in close vicinity of the antenna. \$\endgroup\$ Apr 22, 2018 at 18:51
  • \$\begingroup\$ Happy now ? ;-) \$\endgroup\$ Apr 22, 2018 at 20:24
  • \$\begingroup\$ Of course!!!!!! \$\endgroup\$
    – Andy aka
    Apr 22, 2018 at 20:56

While some optimization of the antenna is possible, given the physical constraints, the effort may not yield a substantial improvement. Some optimization for a smaller range of frequencies is, however, quite plausible and worth pursuing.

Since your goal is to optimize the energy to which the 87Rb is subjected, I would recommend improving the match between the amplifier and the chosen antenna since this will have the most significant influence on the amount of energy delivered to the 87Rb. The MiniCircuits amplifier is looking for a 50 ohm load impedance in order for it to deliver maximum power. Your antenna is far from appearing as a 50 ohm ohm load. As a result, there is a substantial reduction of transmitted energy.

A solution to improving the load impedance that the antenna presents is to insert a matching network between the amplifier and the antenna to transform the impedance. Ideally, this should be located at the antenna in order to minimize feedline losses although with a short, low loss feedline, this requirement is mitigated. Due to the wide frequency range involved, several matching networks will be required to be switched in and out of circuit as the frequency changes. This is typically done with small relays that work well in an RF environment. These could be controlled by your lab system or they could be made to work autonomously by detecting the current operating frequency of your amplifier.

The matching circuits can be designed by modeling the chosen antenna(s) or by taking impedance measurements in situ. Given the close proximity to other lab apparatus, the latter is preferable.

Another possibility is to apply brute force to the problem by using a higher power amplifier. However, generally as the power increases the amplifier becomes more restrictive regarding the load in order to prevent thermal runaway and overvoltage destruction. But there may be some level of compromise that improves your application.

  • \$\begingroup\$ Cheers, Glenn. I understand the need of matching networks but this solution is not feasible for us, since we don't have any space (we have large amouts of optics around the vacuum chamber). So for now, we are simply turning up the power until our signal saturates and this works quite well I think. \$\endgroup\$ Apr 22, 2018 at 18:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.