1. HADES goes SIS-100* SIS-18 DAQ upgrade possibilities
Jan Michel - Goethe Universität Frankfurt
HADES goes SIS-100* SIS-18 DAQ upgrade possibilities
* not yet...

2. Jan Michel - Goethe Universität Frankfurt
HADES DAQ upgrades
Why another upgrade?
How to do it?
How much do we gain?
Who will work on it?
Jan Michel - Goethe Universität Frankfurt

3. Jan Michel - Goethe Universität Frankfurt
Current Limits
Slowest systems are
Shower, fixed read-out time of 11 us
MDC, 400 ns read-out time per hit, 5.5 us latency on link
Current maximum (for small events)
~ 66 kHz with all systems
AuAu, 10kHz
Jan Michel - Goethe Universität Frankfurt

4. Jan Michel - Goethe Universität Frankfurt
MDC front-ends
Slow data transport on MBO
Shaky connectors
Strongly fluctuating power consumption
Analog part quite good
but some more modern solutions to be investigated
Jan Michel - Goethe Universität Frankfurt

5. Jan Michel - Goethe Universität Frankfurt
MDC: Digital Part
Replace both TDC and OEP with FPGA-board
data transport / control: proven on OEP
time measurement in FPGA: proven in many experiments
The bad news: Cahit's TDC are just too good
resolution not worse than 50ps
only 500ps needed for MDC
Jan Michel - Goethe Universität Frankfurt

6. Jan Michel - Goethe Universität Frankfurt
Low-cost FPGA TDC
Sampling with phase shifted clocks
e.g. 400 MHz
sampling with rising/falling edge
4 clocks shifted by 90°
→ 312 ps binning
many channels!
Jan Michel - Goethe Universität Frankfurt

7. Jan Michel - Goethe Universität Frankfurt
MDC FEE Upgrade
Replacing digital part / TDC : “trivial”
Replacing analog part: challenging
PCB design: no critical points
Reducing read-out time substantially
Not reducing trigger-release latency!
maybe upgrade to faster serial links?
AuAu, 10kHz
Jan Michel - Goethe Universität Frankfurt

8. Jan Michel - Goethe Universität Frankfurt
Trigger Releases
Trigger – Busy release architecture not suited for high rates
rather long latency on optical links for transport
CBM, PANDA etc: Free running read-out
Jan Michel - Goethe Universität Frankfurt

9. Jan Michel - Goethe Universität Frankfurt
“Tagged” Read-out
Reference Time sent for each event
CTS assumes deadtime and free buffer sizes
trigger information / busy release only every N events
Handshake needed for buffer handling
Jan Michel - Goethe Universität Frankfurt

10. Tagged Read-out: Deadtime?
add 5us (MDC), 3us (other systems) for busy release
add 400 ns for reference time
4us assumed dead time seems reasonable
max. 250 kHz trigger rate
Jan Michel - Goethe Universität Frankfurt

11. Tagged Read-out – Reasonable?
We can gain a factor 2.5 in event rate
Only for light heavy-ions due to detector load
Much work – all trigger handling has to be rewritten
Code on TRB2 & RICH not touched since four years - experts “vanished”
Jan Michel - Goethe Universität Frankfurt

12. Jan Michel - Goethe Universität Frankfurt
Not only in HADES
TrbNet also to be used for CBM (here: RICH)
Jan Michel - Goethe Universität Frankfurt

13. Jan Michel - Goethe Universität Frankfurt
TrbNet for CBM
Several improvements to existing TrbNet planned for CBM/Panda
Jan Michel - Goethe Universität Frankfurt

14. Jan Michel - Goethe Universität Frankfurt
Summary
Upgrade of MDC FEE
should be done for data quality and read-out speed
gain 30% in max event rate
New read-out scheme
gain factor 2.5
huge cost in man-power
With new RICH read-out limits from load on detector might get less strict
Jan Michel - Goethe Universität Frankfurt