Hi,
Following the recent license server issue, I re-installed Toutatis, and now finding strange results (at least not as in the past). Did you make any recent change in the code.
My point is about the longitudinal distribution that changes at the first calculation step.
The context: I am generating a dst distribution with large X-X' ellipse, very small in Y-Y' and Phi-E to compute the transverse acceptance of an RFQ. Looking at survivor particule at the end, I use PlotWin to look are their distribution at the input. Unfortunately, It looks like, at the first step of calculation, the longitudinal distribution is changed to accommodate for the large divergence particles... I am a bit lost there, and looks like the "acceptance" I obtain is not in line with the one I got in the past. Of course, doing all that with no space charge...
I attach a plot of the input and the distribution after the first calculation step.
Cheers,
JB
Longitudinal distribution change in first step [SOLVED]
-
- Novice
- Posts: 4
- Joined: Thu 25 Nov 2021 12:33
- Country:
France (fr)
Longitudinal distribution change in first step
- Attachments
-
- Atstep1.PNG (155.65 KiB) Viewed 37449 times
-
- input.PNG (118.93 KiB) Viewed 37449 times
-
- Global moderators
- Posts: 80
- Joined: Fri 28 Aug 2020 14:59
- Location: Paris-Saclay
- Country:
France (fr)
Re: Longitudinal distribution change in first step
Hi,
indeed, Toutatis includes the transverse divergence in the calculation of the energy such that this energy corresponds to the "total" energy and is not only longitudinal related. Also this way of doing is like this since the creation of the code, so I am surprised you observed a different behaviour with previous versions of the code.
The change in between *.dst and first step is probably due to the fact the code assumes that the *.dst file follows the same convention, meaning transverse divergence is included in energy values. If it is not the case, a gap will appear and the component Vz considered by Toutatis will be lower than anticipated by the user. All of that is related to the fact that Toutatis does not apply the paraxial approximation. Such tiny aspects are not visible at higher energy where Vz >> Vx,Vy.
I hope it helps.
Best,
Romuald
indeed, Toutatis includes the transverse divergence in the calculation of the energy such that this energy corresponds to the "total" energy and is not only longitudinal related. Also this way of doing is like this since the creation of the code, so I am surprised you observed a different behaviour with previous versions of the code.
The change in between *.dst and first step is probably due to the fact the code assumes that the *.dst file follows the same convention, meaning transverse divergence is included in energy values. If it is not the case, a gap will appear and the component Vz considered by Toutatis will be lower than anticipated by the user. All of that is related to the fact that Toutatis does not apply the paraxial approximation. Such tiny aspects are not visible at higher energy where Vz >> Vx,Vy.
I hope it helps.
Best,
Romuald
-
- Novice
- Posts: 4
- Joined: Thu 25 Nov 2021 12:33
- Country:
France (fr)
Re: Longitudinal distribution change in first step
Thank you Romuald,
I will check what I did in the past, maybe was doing XX' YY' at the same time resulting in a smaller acceptance.
Your explanations make sense. Still, I am afraid I need more clarifications, as at the first step, we also have the effect of the RMS that adds up and mix-up things.
I assume that the particles from "my ion source" get all the same total energy (45 keV). In the acceptance study case, with large divergence particles, the ones with large divergence should still get a total energy of 45 keV. Should I then feed in Toutatis a dst distribution with lower "energy" for those ?
When generating a beam distribution with Toutatis, I indeed see that 1 rad x' particules (I have been exaggerating a bit) are the one with higher total energy and the ones in the center, the one with "nominal energy" of 45 keV. (attached plot). Correct that they get 45 keV + 45 keV
Should I understand TW dst and Toutatis *.dst are not compatible for very low energy and large emittance cases?
I will have a try, as we believe this can be something interesting in our case, and certainly come back to you.
Thanks for your time !
Cheers,
JB
I will check what I did in the past, maybe was doing XX' YY' at the same time resulting in a smaller acceptance.
Your explanations make sense. Still, I am afraid I need more clarifications, as at the first step, we also have the effect of the RMS that adds up and mix-up things.
I assume that the particles from "my ion source" get all the same total energy (45 keV). In the acceptance study case, with large divergence particles, the ones with large divergence should still get a total energy of 45 keV. Should I then feed in Toutatis a dst distribution with lower "energy" for those ?
When generating a beam distribution with Toutatis, I indeed see that 1 rad x' particules (I have been exaggerating a bit) are the one with higher total energy and the ones in the center, the one with "nominal energy" of 45 keV. (attached plot). Correct that they get 45 keV + 45 keV
Should I understand TW dst and Toutatis *.dst are not compatible for very low energy and large emittance cases?
I will have a try, as we believe this can be something interesting in our case, and certainly come back to you.
Thanks for your time !
Cheers,
JB
- Attachments
-
- 1radcase.PNG (187.3 KiB) Viewed 37443 times
Last edited by JB-Lallement on Thu 25 Nov 2021 15:18, edited 1 time in total.
-
- Novice
- Posts: 4
- Joined: Thu 25 Nov 2021 12:33
- Country:
France (fr)
Re: Longitudinal distribution change in first step
Sorry... My shame for the sqrt(2).... All good
-
- Global moderators
- Posts: 80
- Joined: Fri 28 Aug 2020 14:59
- Location: Paris-Saclay
- Country:
France (fr)
Re: Longitudinal distribution change in first step
OK. These are tricky things. Let me share some thoughts about this. First, for Toutatis, the energy in "linac ..." line is for the reference particle not the average particle. This reference particle has no transverse divergence and then gets 45 keV. If you have Xp = 1 rad, then Vz = Vx and then, with non relativistic mechanics, you get E = 0.5 m (Vz^2+Vx^2) = 2 x 0.5 m Vz^2 = 2 x 45 = 90 keV (=> dW = 45 keV). Your plot is consistent with that.
By convention, the linac line points the reference particle as designed for your ion source. It is for consistency with the beta evolution of the vanes. Also, it is true to remark that, thus, the average energy of the particles is greater than this reference energy as we take into account the transverse divergence.
If we forget about codes for a while, the physics tells me that a particle that remains on the axis when space charge is negligible will arrive at the RFQ entrance with an energy equal to the ion source voltage. But now if we consider space charge, the potential on the axis will vary with beam size evolution, the wall pipe distance, space charge compensation evolution and the losses (that result in a modified residual space charge). Just the last aspect (losses of the pollutants) makes that the value of the potential on axis at departure and arrival are different even if the other parameters remain constant. This to say that the beam energy is changed in your LEBT. And the top of this, magnetic fields will transfer energy from one plane to another (rotation) such that you can also get a reduced Vz component and increased Vx.
In conclusion, I recommend to LEBT user when they tune the line and try to improve RFQ transmission to play a bit with the ion source voltage. It might be that your transmission is improved by setting a slightly increased ion source voltage compared to design.
Best,
Romuald
By convention, the linac line points the reference particle as designed for your ion source. It is for consistency with the beta evolution of the vanes. Also, it is true to remark that, thus, the average energy of the particles is greater than this reference energy as we take into account the transverse divergence.
If we forget about codes for a while, the physics tells me that a particle that remains on the axis when space charge is negligible will arrive at the RFQ entrance with an energy equal to the ion source voltage. But now if we consider space charge, the potential on the axis will vary with beam size evolution, the wall pipe distance, space charge compensation evolution and the losses (that result in a modified residual space charge). Just the last aspect (losses of the pollutants) makes that the value of the potential on axis at departure and arrival are different even if the other parameters remain constant. This to say that the beam energy is changed in your LEBT. And the top of this, magnetic fields will transfer energy from one plane to another (rotation) such that you can also get a reduced Vz component and increased Vx.
In conclusion, I recommend to LEBT user when they tune the line and try to improve RFQ transmission to play a bit with the ion source voltage. It might be that your transmission is improved by setting a slightly increased ion source voltage compared to design.
Best,
Romuald
-
- Novice
- Posts: 4
- Joined: Thu 25 Nov 2021 12:33
- Country:
France (fr)
Re: Longitudinal distribution change in first step [SOLVED]
OK.
In fact, we do observe a better transmission when going with the source to 46 kV instead of 45 kV.
Just for numbers: In our case, the acceptance goes up to +/- 200 mrad: It could represent changes in Vz up to 1.7 keV (for one plane) for unlucky particles, what we thought could play a role.
Thanks for your help, it is now clear to me what to do handling those *.dst files going form TW to Toutatis.
All the best,
JB
In fact, we do observe a better transmission when going with the source to 46 kV instead of 45 kV.
Just for numbers: In our case, the acceptance goes up to +/- 200 mrad: It could represent changes in Vz up to 1.7 keV (for one plane) for unlucky particles, what we thought could play a role.
Thanks for your help, it is now clear to me what to do handling those *.dst files going form TW to Toutatis.
All the best,
JB