Re: Muon-On-Request

[Peter Kammel <kammel@one.npl.uiuc.edu>]

Dear Mike,

On Mon, 17 Oct 2005, Michael Barnes wrote:

> Dear Peter,
>
> Thank you for the note. I read through the note and have the following main 
> comments and questions:
> d_switch is probably more like 250ns, rather than the original 200ns. The 
> majority of the ~250ns delay (~170ns) is attributable to the fiber optics, 
> MOSFET driver and MOSFET. The 170ns has increased slightly (~15ns), since the 
> original design, mainly due to for the inclusion of the two parallel 6.2R 
> resistors on the gate of the MOSFETs. In addition, to reduce "overlap" of the 
> FET stacks, during switching, the turn-on delay (with respect to turn-off) in 
> the controls has been increased to approximately 63ns. If deemed necessary we 
> can recover 18ns of the increase in the controls delay by modifying a setting 
> on the controls card. However, this will increase "overlap" current in the 
> stacks, during switching, and therefore increase power dissipation in both 
> resistors and MOSFET cards -- so I would prefer not to do this.
>

Clearly, we won't change during the first Muon-On-Request run. We can 
think about more risky optimization later, one we understand the system 
better.

> On the plus side, the TRIUMF TTL splitter box has a considerably shorter 
> delay than the original (Boston) splitter box: I can't recall the actual 
> numbers, which are in a logbook at TRIUMF [I'm on sabbatical at CERN], but 
> believe that the TRIUMF splitter has a delay of less than 20ns compared with 
> 60ns to 80ns delay, I believe, for the Boston splitter.
>
> Am I correct in interpreting  your note that the average frequency is 29kHz 
> for the kicker in Muon-On-Request mode? (compared with approximately 37kHz 
> for the present operating mode).
>

This depends highly on the Rm we can achieve. If the waiting time for the 
next muon would be small, the ultimate limit would come from 1/Toff ~40 
KHz, but the 29kHz is much more realistic.

> Peter Winter can confirm the overall delay through the TRIUMF TTL splitter 
> box, controls, etc to the pulse rise.  To do this I would recommend the 
> following measurement  procedure:
> 1) Use 2 high impedance (e.g. x10) probes and deskew them both (to have the 
> same timing) at the input to the TRIUMF TTL splitter box;
> 2) Use one of these high impedance probes as a trigger for the scope, connect 
> the other probe directly on the output of one of the capacitive pick-ups on 
> the beam tank (by directly I mean WITHOUT Anatoly's box --- this box includes 
> an "integrator" and delay cable so would distort the timing measurement);
> 3) Note the delay between the rise of the TTL signal and the rise of the 
> deflector plate voltage.

We will check the delays together before we switch to this mode.
>
> NOTE: we have deliberately set-up the timing of HV1 & HV2 to be 9ns delayed 
> with respect to the timing of  MV1 & MV2: this is to allow for the beam 
> propagation delay from the input of plates MV1 & MV2 to the input of plates 
> HV1 & HV2. The 9ns delay also ensures that the beam "sees" the same pulse 
> width at both sets of deflector plates. You should be able to see this 9ns 
> delay depending on which end of the beam pipe you make the timing measurement 
> at. The 9ns delay is achieved by adding 9ns worth of 50R coaxial cables to 
> the inputs of cabinets HV1 & HV2.
>

I see, we'll look for that.

We'll keep you up-to-date, how things progress. But we won't start this 
effort before 2 weeks from now.

Many thanks

Peter


> Best regards,
> Mike
>