low-ish energy linacs

[astecpete]

Hi Michael,

I'm putting together a simulation for our new machine and have a few things to ask. We're using 2998MHz 2pi/3 travelling wave structures. I want to do magnetic compression at ~70 MeV as we want a small energy spread at our FEL, which will be at ~250 MeV.

I have elegant decks from other machines and they generally use RFCW elements.

First question - is it best to define each cell as an RFCW element, then string them together as a cavity, or use one RFCW element and specify the cell_length parameter? The wakes I have are for a single cell. If I use the former, I guess I should specify a phase_reference for the entire structure, if I use the latter is the phase slip automatically taken into account for the cells in the element?

Second question - I start using elegant at ~35 MeV. I'm starting to suspect that using the RFCW element will not cut the mustard for any part of our machine. Should I be using the TWLA element instead? If so, am I able to include wakes and LSC?

Third question - I'm confused about how one should define the gradients. In a RFCW I specify the peak voltage on crest, right. So if my peak gradient is 20MV/m and my structure is 4m long then VOLT = 20e6 * 4. And if I track a particle through on crest it will gain ~80 MeV. The actual peak electric field in the center of a cell will be ~ double the peak voltage i.e. 40 MV/m - integrating over the waveform in a pi-mode structure gives a factor of ~1/2. But in a TWLA element i specify the electric field EZ and in order to check the actual gain in energy of a particle on crest i need to multiply by the equivalent factor for a 2pi/3 structure.

Fourth question - focussing. In RFCW, you use the SRS standing wave model. What does focussing do in TWLA? Is this something to do with the fact that RFCW is first order matrix, but TWLA integrated?

Fifth question (sorry!) - The structures we may be getting are constant gradient, rather than constant impedance so the iris decreases through the structure. Is this what TWLA assumes?

Sixth question (really sorry)? - A radical one. Is elegant the right tool for this problem? As i'm compressing at so low an energy and the space charge will hurt me, am I better admitting that i need to change approach and use ASTRA or OPAL for the whole machine. I'm loath to do this, partly because of my own stupidity (I can't get dipoles to work in ASTRA for example), but partly because even If i get a nice setup in those codes they are more cumbersome and will make it difficult to do working point tuning, tolerance and jitter studies. I know this question has no simple answer, but I would value your thoughts!

Thanks in advance,

Peter

[michael_borland]

Peter,

First question - is it best to define each cell as an RFCW element, then string them together as a cavity, or use one RFCW element and specify the cell_length parameter? The wakes I have are for a single cell. If I use the former, I guess I should specify a phase_reference for the entire structure, if I use the latter is the phase slip automatically taken into account for the cells in the element?

The RFCW element puts in the wake at the end of each slice. So if you make the number of slices (N_KICKS parameter) equal to the number of cells, you'll get the same effect as splitting into many elements, with less work. However, you can certainly do it using RFCW and a phase_reference, which will ensure that all the cavities are phased consistently.

Second question - I start using elegant at ~35 MeV. I'm starting to suspect that using the RFCW element will not cut the mustard for any part of our machine. Should I be using the TWLA element instead? If so, am I able to include wakes and LSC?

I think RFCW should be fine for 35 MeV, since in that case beta=0.99989...

Third question - I'm confused about how one should define the gradients. In a RFCW I specify the peak voltage on crest, right. So if my peak gradient is 20MV/m and my structure is 4m long then VOLT = 20e6 * 4. And if I track a particle through on crest it will gain ~80 MeV. The actual peak electric field in the center of a cell will be ~ double the peak voltage i.e. 40 MV/m - integrating over the waveform in a pi-mode structure gives a factor of ~1/2. But in a TWLA element i specify the electric field EZ and in order to check the actual gain in energy of a particle on crest i need to multiply by the equivalent factor for a 2pi/3 structure.

The energy gain for TWLA should be EZ*L for the on-crest particle. It's just another way to specify the total voltage.

Fourth question - focussing. In RFCW, you use the SRS standing wave model. What does focussing do in TWLA? Is this something to do with the fact that RFCW is first order matrix, but TWLA integrated?

If you set FOCUSING=1, it includes the radial field for the first space harmonic in the integration. For RFCW, it is an analytical model only, even if you break the RFCW into many slices.

Fifth question (sorry!) - The structures we may be getting are constant gradient, rather than constant impedance so the iris decreases through the structure. Is this what TWLA assumes?

By default, it's constant gradient, but there's a field attenuation factor ALPHA that should allow controlling this. For RFCW, it's constant gradient.

Sixth question (really sorry)? - A radical one. Is elegant the right tool for this problem? As i'm compressing at so low an energy and the space charge will hurt me, am I better admitting that i need to change approach and use ASTRA or OPAL for the whole machine. I'm loath to do this, partly because of my own stupidity (I can't get dipoles to work in ASTRA for example), but partly because even If i get a nice setup in those codes they are more cumbersome and will make it difficult to do working point tuning, tolerance and jitter studies. I know this question has no simple answer, but I would value your thoughts!

If space charge is a concern, I'd probably use IMPACT or ASTRA, though I agree dipoles are not as easy in either code. In the limit when space charge is ignored, elegant should be just fine, even if the beam isn't fully relativistic.

--Michael

[astecpete]

Hi Again Michael,

Thanks for the hints on this one. I've been using both RFCW and TWLA since I started this thread.

I have come across what seems to be an inconsistency. I've swapped two cavity structures (accelerating from ~70 MeV) and done a phase scan in both. In the first picture all looks ok, the red points are TWLA and the blue points RFCW.

But a closer inspection reveals a puzzler - in the second plot I just take the difference between the two. As you can see it looks like there is a difference depending linearly on phase - up until a phase of 110 degrees. Then it goes wild. Any ideas?

Cheers,

Pete