sPHENIX discussion of the superconducting solenoid

[Kin Yip <kinyip AT bnl.gov>]

Hi Joe Muratore,

As I've promised:
I've just included a part of the series of emails that I have had with Martin Berndt (who used to be the engineer at SLAC responsible for
the BaBar Magnet's power supply etc. ).   I've also included one page of the PDF file which is actually page 61 of the PDF file
"Controls & Power Supply.pdf"

on our babar disk server "\\bnl.gov\c-adnas\babar\Power Supply & Controls".

As you can read from Martin's email on Apr. 25, 2016 (last year !) and the PDF file, after slowly discharging for ~35 minutes,
at ~1338 A (in their example in the PDF),  BaBar experimenters can't wait any more and they did fast discharge (through the
dump resistor) at that point.

Just something to read and consider .........  It took me some time to find this page 61 :-)     Actually, before and after p.61,
there are other detailed discussion.

Kin

-------- Forwarded Message -------- Subject: Re: 2nd try: done many fast discharges after ~30 minutes in BaBar ? Date: Tue, 26 Apr 2016 22:17:08 -0700 From: Martin Berndt <martinberndt AT sbcglobal.net> To: Kin Yip <kinyip AT bnl.gov>

    Thanks for sending me the slides of your presentation on the low field test.  Nothing surprising in your data, but it must have felt good that everything worked out as expected, and that the coil was superconducting.  I understand that you are changing the iron yoke, and this will affect the axial forces caused by the assymmetry in the two magnet ends.  The change in the permeability of the iron as the current increases makes this a tricky problem, and both Ansaldo and SLAC made careful meassurements of the strain.

    Since it is a topic that interests me, could I find out a bit more about circuit details for the Quench Detector?  I would like that.  As you kow, the topic interests me.

    I noticed from your test circuit diagram that you chose to tie the negative terminal of the power supply to ground.  In the Babar installation we permanently connected the 66 mOhm dump resistor across the magnet (with the DCCT between the magnet and the dump resistor so it would measure only the magnet current), and then tied the midpoint of the dump resistor to ground throuh a 100 Ohm resistor.  I prefered  the midpoint connection to ground because it cuts in half the maximum terminal and coil voltage to ground, and thus reduces the stress on the insulation during a dump.  I did not want to use the coil centerpoint taps for the ground connection because I did not know how delicate they were.  The centerpoint of the dump resistor seemed like a good rugged place to ground the circuit.

    Good luck, and keep me posted.   

        Martin

----- Original Message -----

From: Kin Yip

To: Martin Berndt

Sent: Tuesday, April 26, 2016 7:26 AM

Subject: Re: 2nd try: done many fast discharges after ~30 minutes in BaBar ?

Hi,

Thanks for the information.  Sorry to bug you even after your retirement :-)    We've used a new quench detection system for our low-field test (100 A) on Mar. 22, 2016
--- which was kind of borrowed from our Superconducting Magnet Division and was originally for testing the LHC magnets (their ongoing high-luminosity upgrade) and
we're building a similar one.  I talked about the low-field test here :

https://collab.external.bnl.gov/sites/sPHENIX-Magnet/Past%20Presentations/2016-3-23%20sPHENIX%20biweekly%20Magnet%20Meeting.ppt
and also
https://indico.bnl.gov/materialDisplay.py?contribId=1&materialId=slides&confId=1998

At least, it's proved that the magnet went superconducting (as the voltage tap across coils ~ 0).

Now, we're building a temporary return yoke (boxes) for the full field test and we're building a similar quench detector system like last time.

Thanks again !

Kin

On 04/25/2016 08:56 PM, Martin Berndt wrote:

Hi Kin:
   Sorry I didn't reply to you earlier.  I got side tracked.  I am retired, but keep very busy.
   I will reply to the various points you raised
1)  The reason for doing a fast discharge after discharging the magnet slowly for about 35' was simply that the experimenters didn't want to wait more than about 45' for completely discharging the magnet when they needed access to the detector for one reason or another.  Remember that most of the remaining energy goes into the dump resistor, and only some small fraction is dissipated as eddy currents losses in the coil cylinder and in the matrix of aluminum that encloses the superconductor.  Yes, some helium would boil off, but within an hour or so the coil would again have cooled off sufficiently to allow restarting of the magnet.  Remember that this is an extremely rugged magnet.  It survived a couple crashes at full current (4600 Amps) into the dump resistor, at which time it must have generated over 300 V across the coil terminals.  The insulation was never damaged.  Such crashes were always generated by a malfunction of the Quench Detector, never on purpose.  I suspect that during such a crash quenching of the super conductor occurred fairly uniformly throughout the coil, not localized in one spot, thus not damaging the insulation.  A magnet crash at full current into the dump resistor always caused a lot of helium to boil off, and it would typically take more than 4 hours to cool down the magnet to again make it superconductive and allow a restart.
2)  I don't recall whether the magnet was charged at 2 A/s or 2.5 A/s during routine operation.  The problem we always had was with the Quench Detector malfunctioning during the voltage transient whenever a current ramp (L di/dt) is initiated, even to start a slow discharge.  Turning off the power supply for whatever reason would cause the current to decay through the freewheeling diodes, generate about 6.5 V across the teminals at 4600 Amps and an initial  di/t = 2.3 A/s.  This set the maximum di/dt we could allow.
3)  From my comments perhaps you can see why I have stated before that the best thing you can do is to completely rebuild the Quench Dector, particular the divider network that senses the voltage across themagnet.  You would then be able to run the magnet at higher ramp rates (di/dt), and not worry about crashes when nothing is wrong with the magnet.
   Keep me posted!
               Martin





----- Original Message ----- From: "Kin Yip" <kinyip AT bnl.gov>
To: <martinberndt AT sbcglobal.net>
Sent: Monday, April 25, 2016 11:44 AM
Subject: 2nd try: done many fast discharges after ~30 minutes in BaBar ?

Dear Martin,

I've just found a page in a scanned PDF file (from the BaBar documents) which seems to be
exactly what you talked about in the last email.   I've attached that page.

I've talked with engineers such as Bob Lambiase ... People seem to be a bit surprised that you guys are
"brave" enough to do "fast discharge" at 1338 A.

Even though power ~ (1338/4600)^2, so the remaining energy left is ~8.5% of the total in this example.
But 1338A/38 seconds is like >35 A/s.   Did this (fast discharge at 1338 A) not generate heat and quench
in the Magnet ??   You guys have done this (fast discharge after ~one time constant) many many times ??

In the years of mature BaBar operation, did you guys constantly ramp the magnet up at 2.5A/s ?
( I asked this because you seemed to be using 2A/s in your Word document reply to Bob Lambiase. )

Kin

-----Original Message-----
From: Martin Berndt [mailto:martinberndt AT sbcglobal.net]
Sent: Thursday, April 21, 2016 1:20 PM
To: Yip, Kin <kinyip AT bnl.gov>
Cc: Lambiase, Robert <lambiase AT bnl.gov>
Subject: Re: half an hour a full ramp-down or longer in BaBar ?

Dr Kin Yip:
    Good to hear from you.  The diode through which the magnet discharges has a relatively constant forward voltage drop of about 0.75 Vdc.  Thus the current decays towards a new value of
-577 Amperes (-0.75 V /1.3 mOhm), which of course it can't reach abecause the diode is not a battery.  It does mean however that the diode accelerates the decay, and becomes predominatly more important at reduced currents.  Our practice at SLAC was to initiate a slow discharge rate through the diodes until the current came down to about 1500 Adc, at which point we opened the DC breaker for a fast discharge through the dump resistor.
    Somewhere in the documents I sent to Robert L I have the precise numbers for the current decay with the diodes.  I don't have that here with me.
    Feel free to contact me if you have further questions.  It makes me feel still useful!
            Martin Berndt


----- Original Message -----
From: "Kin Yip" <kinyip AT bnl.gov>
To: <martinberndt AT sbcglobal.net>
Cc: "Lambiase, Robert" <lambiase AT bnl.gov>
Sent: Thursday, April 21, 2016 7:25 AM
Subject: half an hour a full ramp-down or longer in BaBar ?

Dear Martin,

My name is Kin Yip at Brookhaven National Lab. and I'm the physicist now
in charge of the sPHENIX Magnet,
formerly BaBar Magnet.   I found your email address in one of the emails
that Bob Lambiase CC-ed his email to you.

I've come across a memo which has been also shown in this webpage :

http://www.slac.stanford.edu/grp/ad/op/solnramp.html

At the end of the 3rd sentence (near the top), it said "a full rampdown
takes about 30 minutes".

I understand that this is ramp-down through the freewheeling diode with
low resistance (cable).   From the document
you wrote (2nd line on p. 4 of the attached) and other places, the
input/cable resistance is ~ 1.3 mOhm.  With the inductance
of 2.57 H at BaBar, the time constant is close to >30 minutes or so.  In
the time of one time constant, the current would only
get down to about 63% and so you need a couple of 30 minutes before you
can ramp down to 0 (from ~4600 A), right ?

So, in BaBar, when you ramped down to 0 A (from 4596 A or so), it'd take
more than an hour or really just half an hour ???

Kin Yip

Attachment: Controls & Power Supply 61.pdf
Description: Adobe PDF document