sPHENIX tracking discussion

["Yasuyuki Akiba" <akiba AT rcf.rhic.bnl.gov>]

Dear Tom,

 

> the momentum of the desired track is irrelevant.

This is true if we don’t have no information of the waveform. This is why I propose that the read-out is done by FADC with >200MHz.

 

> The statistics of ionization in a gas are VERY low. 

According to the review of particle detectors in the PDG book, number of primary ionization of Ar is 25/cm and that of C2H6 is 48/cm. So the number of primary ionization in Ar:C2H6 should be 36.5/cm or 3.6 per mm. This means that number of primary ionization for 2mm sample is 7.2. Not large, but I don’t think it is VERY small.

 

We can use a different gas if we want to have more primary ionization. For example, number of primary ionization of Ar:iC4H10 (70:30) is about 4.5/mm. If I remember correctly, this gas has advantage of smaller diffusion coefficient.

 

> When we rebuilt the east detector we enlarged the gap to 4mm between the gate (called

> channel in the older figures) wires so as to collect additional charge with an alley of 3mm.

This is your choice. There is trade-off between the sampling length and two particle separation. I suppose that you chose to enlarge the gas since you think that the worse 2-particle separation caused by the wider gap is a smaller problem.

 

The number of ionization, the geometry of the drift cell and wire structure, choice of the gas, etc are technical details we need to consider to achieve the best performance. We should consider what is the best combination to achieve the best performance of the detector.

 

> You cannot see a second track from the waveform as it is too lumpy. 

I disagree. This can be true when two low pT track overlap. But we are interested in the high pT track, which should produce a sharp and narrow pulse. Then we should be able to detect the leading edge of the pulse even if it  overlap with a wide pulse caused by a low pT junk. The fluctuation of the pulse from the low pT junk can detoriate the position resolution of the high pT track pulse. But it can still be used to confirm the track. In addition, the position resolution that is needed to achieve 100 MeV mass resolution for Upsilon is not so high – 300 microns of position resolution is sufficient.

 

In any case, we can simulate all of these effect. The simulation tells us how well we can reconstruct track with the proposed system, and now many silicon layers and how many DC layers are needed to achieve satisfactory performance.

 

> I also indicated in my slides that there _ARE_ viable alternatives to a TPC and in

>  particular you should Monte Carlo an interpolating pad chamber. 

 

I agree with you that Pad chamber is a possible alternative. What we need is an inexpensive detector that has sufficient position resolution just before the EMCAL.

 

Sincerely yours,

           Y. Akiba

.

From: tkhemmick AT gmail.com [mailto:tkhemmick AT gmail.com] On Behalf Of Thomas K Hemmick
Sent: Tuesday, July 12, 2016 9:30 AM
To: Yasuyuki Akiba <akiba AT rcf.rhic.bnl.gov>
Cc: EdwardOBrien <eobrien AT bnl.gov>; sphenix-tracking-l AT lists.bnl.gov; sphenix-tpc-l AT lists.bnl.gov; John Haggerty <haggerty AT bnl.gov>; mills, James A <mills AT bnl.gov>; Don Lynch <dlynch AT bnl.gov>; Sourikova, Irina V <irina AT bnl.gov>; Ernst, Robert E <ernst AT bnl.gov>
Subject: Re: [Sphenix-tracking-l] Comparison of PHENIX DC and proposed sPHENIX DC

 

Dear Yasuyuki

 

It is somewhat unfortunate that you missed the prior meeting.

 

1)  It does not matter at all what momentum track you intend to reconstruct.  Occupancy is driven by all tracks in the event.  As a result, when you calculate the probability that the area used by your desired track is already occupied, the momentum of the desired track is irrelevant.

 

2)  Radius = 200 MeV/c / (1e * 1.5 Tesla) = 44.4 cm; Diameter = 88.8 cm

(see https://www.google.com/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=200+MeV%2Fc%2F(electron+charge*1.5+Tesla) )

 

To not reach the drift chamber means that the *DIAMETER* of the circle should be less than 75 cm, not the radius.  This is why one CAN calculate the crossing angle for 200 MeV/c tracks, as I did correctly, and why it is so troublesome when using effective segments that are 1 meter long in zed and 1mm in the bend direction.  The key is to segment more in zed (see below) so that the occupancy boost due to track inclination is not an issue.

 

3)  The PHENIX DC had two variants of the drift cells.  When we rebuilt the east detector we enlarged the gap to 4mm between the gate (called channel in the older figures) wires so as to collect additional charge with an alley of 3mm.  The west chamber had 3 mm wire spacing and 2mm drift, but with limited electron statistics for the primary ionization (see below).

 

4)  The statistics of ionization in a gas are VERY low.  The number of primary electrons in 2 mm of gas is extremely small, with Poisson probability of zero being not insignificant.  It is also extremely "lumpy" since the variations are not a poisson fluctuation in the number of primary electrons, but rather Poisson in the number of clusters of electrons each of which has a broad distribution due to the energy distribution of the single electron that initiates the cluster.  You cannot see a second track from the waveform as it is too lumpy.  This effect has been the R&D of Craig Woody, Bob Azmoun, and Graham Smith of BNL instrumentation for years.  They have success using a mini-drift analysis to overcome resolution degradation of pad planes at large inclined angles driven by lumpy primary ionization trails, but the studies show that it is simply not possible to disentangle crossing tracks in the gas volume...you need to reduce the occupancy.  This result is part of R&D using up to 5 GHz sampling (CAEN module based upon the DRS4 chip).  However, the APV25 chip (25 nsec = 40 MHz) sampling is already

 

 

able to do a nice job for a minidrift device.  See the publication from eRD6 that came out this month:  

 

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=7497629&url="http%3A%2F%2Fieeexplore.ieee.org%2Fiel7%2F23%2F7497612%2F07497629.pdf%3Farnumber%3D7497629

 

 

I believe it is simple to see that using two z-segments is a non-starter.  As I also indicated in my slides that there _ARE_ viable alternatives to a TPC and in particular you should Monte Carlo an interpolating pad chamber.  These can obtain resolutions as a fraction of the pad size (we routinely do better than 100 micron in eRD6 with 2 mm pads) and this device can then be segmented in Z enough to have an acceptable occupancy.  This would be a reasonable investigation for providing the workable high resolution point at large radius that you desire.  Furthermore, since a normal pad chamber has severe position resolution degradation for inclined tracks, you can even consider a mini-drift configuration option as described the the eRD6 publication as a means of improving the resolution even for the low momentum tracks.  This would be a viable alternative to a TPC, while the jet-chamber is not.

 

All the best,

Tom

 

 

 

On Tue, Jul 12, 2016 at 8:33 AM, Yasuyuki Akiba <akiba AT rcf.rhic.bnl.gov> wrote:

Dear Tom,

 

 

I am aware of the tilt angle effect. This makes the separation of low pT tracks more difficult. However, this doesnot cause problem for high pT track that we are interested in.

 

High pT tracks are parallel to the anode wires. Therefore the electrons from the high pT track will make a sharp pluse, and we can deternine the rising edge of such pulse from the recorded wave form, even if it partially overlap with background hits from low pT tracks.

 

It is true low pT tracks makes a long pulse. But the worse case you describe – 200 MeV/c tracks – will not reach to active area of the DC. So they are not problem. A typical momentum of low pT track is half GeV and its tilt angle relative to the radial direction is about 30 degress. This enlarge the pulse width, but not a factor of two you wrote. Also, the sampling length of a track is one of the parameter we need to adjust to get the best two particle separation. My understanding is that the sampling length of a track by PHENIX DC is about 2mm meter. You adjusted the voltage configuration to get this.

 

I think you missed the most important point in my message. The message is that the proposed DC is not intended to reconstruct track by itself. It is used to confirm a track that is already reconstructed, and to measure the position of the track near EMC to improve the momentum measurement of the track. For this purpose, we need 3 or 4 isolated hits that are associated to the track. If a low pT junk track overlap with a hit of the track of interest, we can recognize such overlap easily from the recorded waveform. There should be a peak that is consistent with the hit from the track over a wide background. We can either assign a larger position error for the hit (taking into account the effect of the fluctuation of the background pulse) or ignore the hit. If we have a sufficient number of layers in the DC, we should have get 3 or 4 clean hits that is completely isolated from any background hits. These 3 or 4 hits are all I need to confirm the track and to measure its position at a large radius. The precise positon measurement from the DC then determine the momentum of the track at high accuracy.

 

In any case these effect can be simulated and evaluated. We can evaluated the number of layers needed from the simulation.

 

Sincerely yours,

              Y. Akiba

From: tkhemmick AT gmail.com [mailto:tkhemmick AT gmail.com] On Behalf Of Thomas K Hemmick
Sent: Tuesday, July 12, 2016 6:32 AM
To: Yasuyuki Akiba <akiba AT rcf.rhic.bnl.gov>
Cc: EdwardOBrien <eobrien AT bnl.gov>; sphenix-tracking-l AT lists.bnl.gov; sphenix-tpc-l AT lists.bnl.gov; John Haggerty <haggerty AT bnl.gov>; mills, James A <mills AT bnl.gov>; Don Lynch <dlynch AT bnl.gov>; Sourikova, Irina V <irina AT bnl.gov>; Ernst, Robert E <ernst AT bnl.gov>
Subject: Re: [Sphenix-tracking-l] Comparison of PHENIX DC and proposed sPHENIX DC

 

Dear Yasuyuki

 

Perhaps you missed my slides.  I have attached them here.

 

Because tracks are tilted by the magnetic field, they physically span a space that equals the DRIFT-ALLEY * SIN(ALPHA) as shown on slide 3.  It is humanly impossible to achieve 2-particle separation smaller than the limit set by the tilt angle.  For 200 MeV tracks (near the peak of what nature produces) with 3 mm drift alley, this will be 6 mm for the geometry that you suggest.

 

Although these "jet-chamber" designs intend to separate high momentum high density tracks, the occupancy figure is driven by the background tracks rather than the signal.  As a result of the tracks tilt, the occupancy is much too large for a jet chamber to function at 75cm radius and 1 meter wire length on each side.

 

Tom

 

 

 

On Tue, Jul 12, 2016 at 6:19 AM, Yasuyuki Akiba <akiba AT rcf.rhic.bnl.gov> wrote:

Dear Ed,

 

I see you have concern that the DC I proposed have difficulty to reconstruct tracks since it need to handle order of magnitude higher particle density than PHEINX DC and the occupany is high.

 

First, I have a mistaken in my slides a the general meeting. The maximum drift length of the proposed DC is the same as that of PHEIX DC, i.e. 2 cm. This still makes the number of track per anode reduced by a factor of two.

 

In any case, the main factor to determine reconstruction of tracks is not the average number of hit per anode, but the two particle separation. Here you calculate that the effective occupancy will be 20% assuming that dN/deta=750 and two track separation of 1.5mm. I assume that two track separation of 1mm can be achievable. Also, using the measured dN/deta = 620 (PPG019), I calculate the effective occupancy to be 10%. I agree that to achieve two track separation of 1 mm is not a trivial task. But since PHENIX DC can achieve two track separation of 1.5mm (a NIM paper of PHENIX DC wrote that the separation is 1.3mm at 50% efficiency) with TMC chip, I think 1mm separation with FADC readout is achievable. This is certainly an important R and D item.

 

One thing I want to point out is that the proposed DC is not intended to work as a stand-alone tracking device. The track reconstruction will be done by the entire tracking system consisting of “Pixel”, 4 layers of Si strips, EMCal and DC. A track is first reconstructed by EMC-Pixel-beam center. These three “pixel” points (3D points) define a circle for a track road. Then the track is confirmed by 4 layers of strips in R-phi plane. After that, DC hits are used to confirm the track again and to measure the position of the track near the EMC. With the precise position determined by the DC near EMCAL, and the precise direction vector information determined by “pixel”-strips, we can determine the momentum of the track at high precision. Gaku’s simulation has already demonstrated that this combination will have upsilon mass resolution better than 100 MeV with full sPHENIX GEANT4 simulation.

 

In this scheme, DC is only used

(1) to confirm a track that is already reconstructed, and

(2) to refine its momentum. It is not expected that DC can reconstruct tracks as a stand-alone tracker.

In fact, DC alone cannot reconstruct a track since DC doesn’t have any UV wire. So it cannot determine the z-coordinate of the track. But the z-coordinate of the track is provided by EMCAL, as a PAD chamber provide the z-coordinate in the PHENIX DC.

 

I agree with you that occupancy of 10% is high for a stand-alone tracker. However, I think the 10% occupancy is fine for a DC whose function is solely to confirm a track that is already reconstructed. If occupancy is 10%, some of the 6 DC hits will be masked by another background track. But some others are isolated. As long as there are sufficient number of those isolated hits remain, the DC can deliver its intended functionality. The probability of having 4 “isolated” hit out of 6 wires under 10% occupancy is more than 98%. We can then use only those 4 clean hits to confirm the track and to measure its position. This is why I think the proposed DC will work in central AuAu collisions.

 

If the occupancy is higher, say, 20%, this probably of having 4 isolated hits is reduced to 90% .But the probability of 3 “isolated hit” is more than 98%. So we can use these 3 isolated hits. We can also increase the number of wires. We can, say, increase the number of anode layers to 10 instead of 6. In this case, the probability to have more than 5 isolated hit out of 10 is more than 99% even if the occupancy is 20%. The number of read-out in this case is only about 5K channels so the cost of the DC is still modest.

 

We will make a realistic simulation with full HIJING event to evaluate the effect two track separation. This will answer the question how many layers we needs as a function of track separation. We can also evaluate how many strip layers are needed.

 

Please note that the proposed system is a MINIMUM system to make the cost as low as possible. If simulation shows that the efficiency or fake rate is not satisfactory, we can always increase the number or layers of strips and / or DC with a modest cost.

 

The present sPHENIX tracker program is not designed to work well for a projective detector like DC. We need a reconstruction program that is designed to work with the proposed tracking system. Gaku is working on it. Once the reconstruction code is working, we can quantitatively evaluate how many layers we really need with full GEANT4 simulation of HIJING events.

 

Sincerely yours,

              Y. Akiba

From: sphenix-tracking-l-bounces AT lists.bnl.gov [mailto:sphenix-tracking-l-bounces AT lists.bnl.gov] On Behalf Of EdwardOBrien
Sent: Friday, July 1, 2016 8:13 PM
To: sphenix-tracking-l AT lists.bnl.gov; sphenix-tpc-l AT lists.bnl.gov; John Haggerty <haggerty AT bnl.gov>; mills, James A <mills AT bnl.gov>; dlynch AT bnl.gov; Sourikova, Irina V <irina AT bnl.gov>; Ernst, Robert E <ernst AT bnl.gov>; eobrien AT bnl.gov
Subject: [Sphenix-tracking-l] Comparison of PHENIX DC and proposed sPHENIX DC

 

    Dear all,
     I have compared the design and performance parameters
    of the PHENIX Drift Chamber and the proposed sPHENIX Drift Chamber
    as described at the sPHENIX general meeting on June 24.
    Tom Hemmick and I have gone over these number and agree
    with the conclusion of this summary document.
    I'd be happy to answer any questions. Thanks.

    Ed

_______________________________________________
Sphenix-tracking-l mailing list
Sphenix-tracking-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-tracking-l