RFDF R56

Moderators: cyao, michael_borland

Post Reply
daniel.marx
Posts: 9
Joined: 20 Dec 2016, 08:45

RFDF R56

Post by daniel.marx » 05 Apr 2017, 08:00

For an RFDF element with non-zero length, it was mentioned in another post that the element is split into many drift-cavity-drift slices, where the cavity slice is a zero length element represented by a second-order matrix.
As far as I can tell there is no R56 element included in the matrix used for the drift slices, so we do not see bunch length elongation in the TDS at non-ultrarelativistic energies despite the induced energy spread.
I am wondering whether what I've stated here is correct and, if so, if there is a reason this effect is not included.
Thanks!

michael_borland
Posts: 1796
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: RFDF R56

Post by michael_borland » 10 Apr 2017, 13:41

Daniel,

Unfortunately, this is a bug. I don't have any workarounds to offer beyond the rather tedious method of using many zero-length RFDF elements separated by drifts. You'll need to be careful about the relative phasing since (alas) the phase_reference feature is not implemented for this element.

--Michael

daniel.marx
Posts: 9
Joined: 20 Dec 2016, 08:45

Re: RFDF R56

Post by daniel.marx » 13 Apr 2017, 03:57

Thanks for confirming this, Michael. Hopefully it can be fixed in a future update.

I'm a little confused by your comments about the phase_reference feature as it is mentioned in the manual for this element.

michael_borland
Posts: 1796
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: RFDF R56

Post by michael_borland » 13 Apr 2017, 07:58

Daniel,

It will be fixed in the next release. Sorry for the inconvenience. If you build from source, I can provide the update this week.

You are correct that the PHASE_REFERENCE parameter is listed in the manual. However, it is not actually functional, which is another issue needing resolution.

--Michael

daniel.marx
Posts: 9
Joined: 20 Dec 2016, 08:45

Re: RFDF R56

Post by daniel.marx » 05 Dec 2018, 08:45

Michael,

I was just coming back to some RFDF simulations in elegant and I noticed again something funny, this time with version 33.1.1 so the problem below should have been fixed.

I am comparing a lattice with a zero-length TDS separated by drifts and a lattice with a 0.8-m long TDS. In both cases, N_KICKS=1, so these two lattices should be exactly equivalent I believe. However, we see different results. Any ideas what's going on?
Attachments
tracking_quadsoff_oldlattice_tdslong.ele
(271 Bytes) Downloaded 189 times
tracking_quadsoff_oldlattice.ele
(267 Bytes) Downloaded 183 times
noquadlat.lte
(444 Bytes) Downloaded 199 times

michael_borland
Posts: 1796
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: RFDF R56

Post by michael_borland » 05 Dec 2018, 10:10

Daniel,

The source of the discrepancy is that RFDF interprets the PHASE quantity to be the phase upon arrival at the entrance of the cavity. At the same time, the cavity is a speed-of-light traveling wave structure (STANDING_WAVE=0) and the beam is not fully relativistic. You can resolve the discrepancy with a slight change in the PHASE value for the long RFDF element. See attached.

By the way, I also noticed that the setting for the TILT value contained an error:

Code: Select all

TILT="pi -2/"
is not a valid RPN expression. It should be

Code: Select all

TILT="pi -2 /"
--Michael
Attachments
noquadlat.lte
(743 Bytes) Downloaded 196 times

simone.dimitri
Posts: 46
Joined: 09 Jun 2008, 01:19

Re: RFDF R56

Post by simone.dimitri » 25 Mar 2022, 02:59

Hello,
I am using the RFDF element for simulating crab cavities in an electron ring at 2.4 GeV.
As already observed and discussed by others, but with different parameters, I obtain a substantially different result for the following two sets (just a zero-length and a non-zero length standing wave):

RFDF: L=0, Voltage=0.8e6, phase=90, standing_wave=1, n_kicks=0, .....
RFDF: L=0.325, Voltage=0.8e6, phase=90, standing_wave=1, n_kicks=0, .....

The discrepancy is removed if I use a trevlling wave, i.e., standing_wave=0. So, what is the prescription for the correct modelling of a finite length standing wave? And, is recommended to increase n_kicks to a large value?
Thank you in advance. Best regards,
S.

simone.dimitri
Posts: 46
Joined: 09 Jun 2008, 01:19

Re: RFDF R56

Post by simone.dimitri » 28 Mar 2022, 07:08

I would like to correct my previous message.
I am able to get identical results from a SW with proper length (multiple of the RF wavelength) and from a TW. However, when using SW, I get the expected result if N_KICKS=1. For any other vaue of N_KICKS, I get something different and unexpected. This seems not to be a problem with a TW.

michael_borland
Posts: 1796
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: RFDF R56

Post by michael_borland » 01 Apr 2022, 16:18

Simone,

The reason is that with N_KICKS=1, the cavity is with a single impulse of V at the center, so you get what is naively expected. For N_KICKS>1, the standing wave is approximated using impulses at several locations, which approximates the transit time factor. You can check this by putting in a cavity with length c/f, which will not deflect the beam at all in reality. For N_KICKS=1, you see the deflection VOLT/E, which is wrong. For other values, you'll see a very small deflection.

--Michael

Post Reply