Horn current monitor

H. Kubo, A. K. Ichikawa (Kyoto university) E. D. Zimmerman (KEK & Colorado university) T. Sekiguchi (KEK)

History

• • • • K2K used CT’s for each striplines.

The stripline structure was different at the CT position. This looked that it is adding inductance and also that it is weak against the Lorentz force. (c.f. T2K Lorentz force is 1.7 times larger than that for K2K.) So we started a project to monitor the current flowing the stripline with pick-up coils, which is small enough that there is no need to change the stripline structure.

K2K has a analogue interlock module looking at the valance among four CT outputs. It’s maximum rating was 300kA. We have decided to make a new digital interlock system using PLC.

K2K CT’s K2K CT Interlock module

Present status

• The progress of the R&D was not so quick. This is partially because trial and debugging cannot be done by the power supply problem. We have only three opportunities to actually see signal before this March.

• So it is still debugging process.

• With the experiences in this April and June run, it should be modified if necessity found.

Purpose and requirement

• • • • • • Monitor the current of each stripline pulse-by-pulse during throughout experiment. Both magnitude and timing should be monitored.

From purely physics point of view, required precision for total current is 5%. See http://jnusrv01.kek.jp/internal/t2k/nubeam/report/Spitz.05/HornCurrent.spitz.ppt

http://www.t2k.org/docs/technotes/004 To monitor the stripline or cable status,1% sensitivity for stability monitoring is required.

So we set the requirement for the monitor as – A few tenth of 1% sensitivity for stability monitoring – <5% for the absolute current determination See p.? for other facility horn occurences. This monitor will be also used to tune the horn fire timing against beam. Required timing precision is 80micro seconds. (1% current drop from the peak) Note that this monitor was intended to be used to protect horn, striplines and cables, but not to protect the power supply. If <0.1% pulse-by-pulse sensitivity is required in order to protect the power supply, this monitor would not work for that.

Because it will be installed in radiation environment, it should be fully passive in the area.

Proportional to field change

Principle

Integration circuit Ground level control tent A-out Pick -up coil A-out Isolation amp Peak Hold module Copper FADC Interlock PLC ADC field stripline An integration circuit is necessary In order to deal the signal with the peak hold module and PLC.

Assumption of this method

• The field at the coil position is approximately uniform. Or the coil has to be small compared to the stripline width(40cm).

• The field at the coil position is dominantly determined by the current of the striplines sandwiching that coil.

• The magnitude of the current of the striplines sandwiching the coil is approximately same.

2cm 40cm

Field calculation by Poisson/Superfish

• • apply 250/4 kA for each pair.

m =1 is assumed for aluminum

• • • • •

Required Time constant of the integration circuit

Since the horn pulse width is 1~3ms, ideally t But large t should be >100ms.

means small output voltage and weak against noise.

Simulation study was done.

Still pulse shape with 20ms integrator dose not exactly reproduce the original pulse shape. But considering the purpose of this monitor, we decided to use t =~20ms in order to achieve sufficient pulse height for the production.

Optimization of t may still need some study. Peak timing should be also checked. 100 ms output ～ 13mV Blue : Horn current black: induced voltage Red: integrator output (Absolute pulse height depends on the coil inductance. Just see relative magnitude here.) 20 ms output ～ 55mV

1

st

option with compensation coil

field stripline Compensation coil using fine-met core

2

nd

option with RC circuit

A-out field stripline

Compensation coil v.s. RC • Both options look promising.

• But we had a some difficulty to produce good compensation coil, we decided to use the RC integrator.

(RC integrator is easy and cheap.)

What we learned in July and August, 08 operation at Tsukuba.

• Observed 5~10% field in the gap where field should be low.

• This is same whether the striplines are connected to the horn or just terminated. In the latter case, the current in both side of the stripline pair is exactly same.

Raw signal After offline integration

● ■ : at terminator : at horn2 connection black : coil A red : coil B green: coil C blue : coil D 4~5% 10% • • • • • • Difference among coils is small.

Pattern is same for the termination connection and horn2 connection. -> This phenomenon cannot be explained by unbalance among pairs.

Many current patterns are investigated using Poisson/Superfish, but failed. Nominally, the field should be 2~4%. (Current distribution in the stripline plate is assumed to be uniform in this study.) See http://jnusrv00.kek.jp/jnu/tgt-horn/horn/stripline/CT/HornCT.0808.kubo..ppt

So 7% is not understood.

In order to confirm that there is really 10% field, field probe measurement was carried out in December 08 and the result is consistent.

Possible origin may be dynamical effect? (Poisson/Superfish is static analysis.) Or non-uniform current distribution in the stripline plate?

What we learned in December, 08 operation at the target station ground level.

• pick-up coils are set and connected to coaxial cables. The expected signal size was ~1V before integration and ~15mV after integration.

• Big noise at the rising edge.

~200 mV, repetition cycle~ 5micro seconds – did not disappear even after integrator or isolation amp. – Seemed to be on GND Line of the coaxial cable.

Noise was reduced when the GND line of the coaxial cable was connected to the earth.

• Coaxial cable was replaced with twisted pair cables with shieldings. The shielding was connected to the earth in control tent. -> Big improvement.

• We have decided to adopt twisted pair cable with shielding.

• We have decided to enlarge the signal size by increasing the number of turns of the pick-up coil.

Pick-up coil specification

• 9ch/module • 0.25mm wire • 8mm thickness: 48 mm width: 40 mm wire region ・ (~140 turns expected) ・ wire resistance: 15m/ch, 370ohm/km(JIS) --> 5.5 ohm inductance: 1.4mH At ω~1000(3ms pulse) --> corr. 1.4 ohm calculated by http://emclab.mst.edu/inductance/rectgl.html

・ expected signal size – 30V (before the integrator) – 300mV (after the integrator)

Radiation Dose • Expected radiation dose is ~100Gy/5years. • Teflon should be avoided.

• Acrylic would be O.K.

• No active electrical component should be used in the area.

Calibration strategy

• • • • Use well-calibrated hall probe at close location to the pick-up coil.

Read simultaneously pick-up coil signal and hall probe signal.

Get absolute field value by hall probe and normalize the pick-up coil signal.

Use FEM calculation to link the field value and current of the stripline.

– 0.194T@250kA, 0.248T@320kA c.f. 0.251T@320kA by hand calculation • • • • See p.6 for the assumption for this method to be valid. Since we observed 7% discrepancy at the low-field gap, the obtainable accuracy for the absolute current would be ~ 7% unless we can understand this phenomenon.

– Field measurement along the stripline width may help.

Relative calibration accuracy between coils will be determined from the noise level. Another way of relative calibration is to move the pick-up coils to different gap position and measure. The absolute normalization can be improved by doing the field measurement by hall probe in side the horn conductors while pick-up coil is measuring the field in the stripline gap. We expect 2% level accuracy with this method. See p.21~25 * Rogowsky coil suggested by Koseki-san is also potent to calibrate the pick up coil.

Hall probe specifications

• SENIS three-axis magnetic field probe • Specially ordered model with 25 kHz bandwidth • Linear range +/- 2 T • Output coefficient is 0.2 T/V • Field accuracy of probe 0.1% • Temperature sensitivity 0.01%/ ° C • Noise <2 mV (4 gauss) • Calibration stability <1% over 10 years

Field probe measurement at stripline • (Done and being analyzed by Kubo-kun now.)

Horn field measurements

• Used probe holder and positioning tool from Univ. of Colorado • Data read by KEK differential-input dataloggger

Horn field measurements

• Several sources of error at the ~1% level: – Probe radial location: position read off from scale on probe holder; precision is only about 0.5 mm – Position is referenced to outer conductor which (in Horn 2 and Horn 3) may not be round, introducing more error on radius – Tsukuba horn test facility did not have precise horn current measurement – Hall probe averages over unknown (probably about 1 mm) area

Horn 1 and Horn 3

• Measurements were made in 2007 by Z. Butcher and KEK group • Analysis used peak probe voltage in pulses • Field agrees with absolute prediction from Ampere’s law to within 2% except for expected dropoff near outer conductor: Horn 1 downstream left port Horn 1 downstream left port

Horn 1 and Horn 3 • Horn 3 upstream ports have largest observed deviations from expected field; nearly 2%.

Horn 3 upstream left port Horn 3 upstream top port

Horn 2

• Measurements made in August 2008 by E. D. Zimmerman and M. M. Tzanov • Current was 250 kA, not 320 kA; analysis used sine-wave fit of central portion of pulse; ten pulses wereaveraged for each location.

• Field is within 1% of nominal at all but one port 25

concern

• Lorentz force on the wire – Field 0.2Tesla, Imax=1A – F= 0.2 N/m = 0.2gW/cm – 2.5gW per 1 turn.

– In case the force sum up coherently, 300gW.

– Elastical limit of the wire is 400gW.

• Capton tape around may not be robust for 5 years use. Photo?

• We have to see what happen in this april and May run and improve before autumn.

Supplement

1

st

trial with compensation coil

• 新幹線の中で暇だったので、ちょこちょこ電磁気の計算を。 まず、補償コイル内 の磁場が飽和していないかですが 50Ω 受けのところで 100mV 見えているとき (2mA) で、 300 巻き、半径 6cm 、比透磁率 20000 磁束密度は 約 として コア内部中心部分での 0.04

テスラ ファインメットの飽和磁束密度は 1.2 テスラですので、 60 mA (3V 普通の CT 見えるあた り は、巻き数 N ) で飽和します。 これに関しては、 FADC で読める範囲 の信号に対しては心配無さそうです。 次に、ピックアップコイルの巻き数について。 に比例して起電力が上がる一方で 自己インダクタンスが N の２乗に比例して上がり、自分で起電力を主に消費する ため、出力は 透磁率が 1 ピックアップコイル自身のインダクタンスは 現状 10μH 1/N に なります。 しかし、今回は３つ特殊な事情があります。 １．後ろに巨大な補償コイ ルがあること ２．低周波相手であること ３．ピックアップコイルが小さく、しかも比 程度しかなく、 相手にする信号も 167 Hz 相当と遅いので、電圧降下への寄与は最大で 10mΩ 相当。 受け抵抗 50Ω に比べても寄与は十分小さいです。 おまけに後ろに 2H (167Hz に対して、最大 2kΩ 相当 ) とかいう馬鹿でかいコイルが 付いているので、 ピックアップコイルのインダクタンスは無いも同然です。 よって、ピックアップコイ ルは単純に電池として働いているので 巻き数を増やして起電力を上げた方が信 号は大きくなります。 巻き数を数倍、面積を倍といった感じで現在の 10 倍程度の 信号サイズには簡単に 持っていけると思います。

2

nd

option with RC circuit

• RC 積分回路の微分方程式を解いてみました。 L が十分小さくて起電力は保存さ れるとの仮定の下で 信号線に R , 信号とグランドの間に C, 受け抵抗 50 Ω とする と 時定数は 50RC/(R+50) という計算になりました。 詳しい計算とかはまた後日。 C から見て R と 50 Ω が並列の合成抵抗のように見えてるんですかね。 信号の周 波数よりも十分長い、同じ時定数で比べた場合 信号のサイズは抵抗での電圧降 下の影響で、補償コイルの場合の ナメル線の安全電流値は 0.5A 50/(R+50) に なります。 よって、小さめの抵 抗と大きなコンデンサが理想ですが あまりやりすぎると充電期間に流れる電流 が大きくなるので、ほどほどにすべき でしょう。 ピックアップコイルに使っているエ です。 起電力は最大で 1V 程度なので、 10Ω くら いの抵抗が入っていれば全く問題ないは ずです。 ( しかもパルスですしね ) これと 4.7 mF のコンデンサで現状の補償コイルの時定数 46 ms と同等の時定数 とパ ルスサイズを得ることができるはずです。 長さ 3 センチくらいのサイズで普通に 200 4.7 mF のコンデンサは、直径 円くらいで手に入るようです。 10 数ミリ、 http://jp.rs online.com/web/search/searchBrowseAction.html?method=getProduct&R=3 654048 電解コンデンサは周波数的には低い方が得意で、静電容量は 120 Hz での測定が 基準とのことですので ホーンの信号は最も得意な領域になります。 信号サイズ自体は、どうやっても時定数とトレードオフのようなので どこかでバラ ンスを取るか増幅するかの選択になるでしょう。

• 市川さん ノイズの周波数と振幅がどうなるかにもよりますね。 FADC るので の方は 1V 10 mV を超える可能性がある以外はあまり問題無いかと思 いますが、 微分波形をピークホールドすると、ノイズがそのまま乗 のスパイクノイズでも致命的となってしまうのでは ないでしょうか。 久保 Atsuko K. Ichikawa さんは書きました : > 久保くんへ、 > > では、いまさらですが、積分なしでやる、っ ちゅうのはどうでしょうか？ > FADC は両極取れるはずなので、ソ フトウエアで積分できますね。 ルドも、ピークの高さは > インターロックの方のピークホー > 微分波形のままでも、電流値に比例する ので、良い？ >> 久保です。 > > Hajime Kubo さんは書きました : >> 市川さん >> >> >> 実際のパルスでの測定ができなかったので、 >> 代わりにちょっくら時定数について考えました。 積分をするために、 >> >> 完全な >> 信号の主成分 (167 Hz) に対して一桁低い カットオフ周波数 (10 Hz : 時定数 >> >> >> 100ms) を目標にしていますが、 そのこと自体が信号を小さくしている原因です。 パルスの形を仮定して、様々な時定数での積分波形を計算しました。 添付ファイルをご覧ください。 >> >> わりに１０倍大きくなる 」 のですから 「 信号がちょっと歪む代 >> カットオフを くらいで妥協しても良いのではないでしょうか？ > > >> >> 適当に 100 Hz