Links:

191217: monitor the modAnode for 5 hours
191211: monitor the modAnode as well as the fiber optic beam and tail clipper pulses.
191204: monitor the modAnode after replacing the dp11 switch tube.
191203: monitor the modAnode signal for 7 hours.
191125: monitor the modAnode signal for 4.5 hours using the sps timing generator.
191118: monitor the modAnode signal for 3 hours using the local timing generator
191002: clp, tpsd rf pulse response, 430power meter.
18jan18: usrp: downstairs blanking turns on late.
01oct15: tx pulse 440 and 1000 usecs rf pulses.
04apr12: can we replace the 430tx with a ring of amiser panels?
27mar12: modAnode not staying all the way off after beam off.
16jan12: waveguide measurements
19oct10: scope trace of protect turnoff
07oct10: 430 transmit pulse sequence timing
07aug10: 1 MHz chirp transmitter pulse output
03aug10: measure klystron output of 1 MHz chirped signal using scope
Hagens 430tx operations and maintenance manual

apr12: Ionizing  radiation measured in 430 transmitter room  (.pdf)

mar17: directory holding 430tx power meter files

---

17dec19: monitor modAnode signal for 5 hours (top)

    The 11dec19 430 tx test continued to show a problem with the modAnode pulse. After the 11dec19 test, carlos found a problem with one of the
transistors on the fiber optic board in the tank. He replaced the transistor. On 17dec19 the 430tx test was repeated.
 The setup was:

•  Use the sps to generate the ipp,rf,beam and phase pulses.
  
• 16 ms ipp, 640 usec beam/rf, constant phase
• The pulses sampled were:
• modAnode pulse read back from the floating dec
• commanded beam pulse output from hagen's box
• the beam pulse output from the fiber optic board in the tx room. this pulse goes via fiber to the floating dec.
• the tail clipper pulse output from the fiber optic board in the tx room. This pulse goes to the buffer dec. It is generated from the falling edge of the beam pulse from a one-shot.
• data was sampled at 12 bits, 2 usec sampling using the ri.
• data was taken from 09:18  to 14:09
  

The data was processed in blocks of 100 ipps (1.6 seconds).
An average, max hold, and min hold was computed for each time sample of the 100 ipps.

A movie was made using the 10915  1.6 sec blocks of data (.avi) (.mov)

• The time is shown in the upper right: 09:18 to 14:09 (ast)
• Only the first 1.4 milliseconds of the 16ms ipp is plotted.
• white line: avg ipp over 1.6 secs
• red line: peak hold for the 100 ipps
• green line: min hold for 100 ipps
• The 3 frames are:
• top: modAnode read back
• middle: beam pulse output from the fiber optic board to the floating dec
• This is a negative going pulse.
  
• bottom: tail clipper pulse going from the fiber optic board to the buffer dec
• This is a negative going pulse
  
• The plots were scaled to the average deflection.
  
• Events:
• the min hold showed that the modAnode pulse dropped  out once during the 5 hour run.
• at 09:34:35
  
• A 2 usec glitch was also seen in the tail clipper pulse. It occurred at the start of the beam pulse. It did not affect the modAnode pulse.
  
• at 11:01:04
  

Plotting avg,min,max value of the 3 monitored  pulses (.ps) (.pdf)

• The  avg, min hold , and max hold pulse was computed  for each 1.6 sec block of 100 ipps.
  
• Each of these is plotted in a separate color:
  
• black is  the avg ipp
• green is the min hold ipp
• The beam command exiting hagen's box was also recorded. It never had a failure.
  
• page 1: entire experiment
• top: modAnode pulse (full scale)
• The average value decreased by about 10% during the 5 hours
• You can see the dropout in the green min hold values.
  
• middle:The beam command exiting the fiber optic board on the wall in the transmitter room.
  
• The signal was active low. It remained low during every ipp.
• bottom: the tail clipper command exiting the fiber optic board on the wall in the transmitter room
• The signal was active low. It remained low during every ipp. there was a glitch (see individual ipps plots)
  
• page 2: blowup of min hold plots of the 3 signals around the  modAnode  dropout
  
• the + are individual 1.6 sec min hold averages.
  
• top: modAnode
• middle: beam command from fiber optic board in tx room (active low)
  
• bottom: tail clipper command exiting from fiber optic board in tx room (active low).
  

plotting individual ipps during the failures (.ps) (.pdf)

• The modAnode value for 100 ipps (1.6  seconds) were plotted starting at 09:34:35
• page 1 : mode Anode dropout
  
• top: avg modAnode value for each ipp
• there is 1 ipp  with a problem
• the + are individual ipps.
• bottom: 3 ipps about dropout
• red: ipp  before, green: ipp of  dropout, blue: ipp after
• Page 2: tail clipper glitch at 11:01:04
• Page 1 the mode anode pulse during glitch
• top: mod anode pulse
• bottom: tail clipper glitch.
• The tail glitch occurred on the rising edge of the modAnode pulse.
• Page 3: blowup of the tail clipper glitch
• top: mod anode
• bottom: tail clipper
• the data is sampled at 2 usec resolution
• The mode anode rise time is not affected by the tail clipper glitch.
• This must be a glitch in the monitoring. If the tail clipper did turn on when the switch tube turned on, you'd expect a much different rise time of the mode anode.
  

SUMMARY:

• The modAnode pulse and the commanded beam pulse was monitored for 5  hours.
  
• The sps was used to generate the ipp,rf,beam, and phase.
• This test was after a transistor in the fiber optic board (in the tank) was replaced.
  
• This test included monitoring of the beam and tail clipper commanded pulses that left the fiber optic board on the wall in  the hv vault.
• These pulses did not fail when the modAnode monitored pulse dropped out.
• So the problem is after the fiber optic board on the wall in the hv vault.
  
• There was only a single modAnode dropout over the 5 hours of the test (at 09:34:35)
  
• There was a single glitch in the tail clipper pulse (at 11:01:04) when the mode anode pulse rose up.
• This was probably a glitch in the monitoring, since the rise time of the modAnode pulse looked normal.

processing:x101/191217/430txtest.pro

---

11dec19: monitor modAnode signal for 4.5 hours (top)

    The 04dec19 the 430 tx test continued to show a problem with the modAnode pulse. On the morning of 11dec19, we repeated the 430tx test while adding
monitoring of the fiber optic beam, tail clipper pulse from the board  in the transmitter room on the wall. The setup was:

•  Use the sps to generate the ipp,rf,beam and phase pulses.
  
• 16 ms ipp, 640 usec beam/rf, constant phase
• The pulses sampled were:
• modAnode pulse read back from the floating dec
• commanded beam pulse output from hagen's box
• the beam pulse output from the fiber optic board in the tx room. this pulse goes via fiber to the floating dec.
• the tail clipper pulse output from the fiber optic board in the tx room. This pulse goes to the buffer dec. It is generated from the falling edge of the beam pulse from a one-shot.
• data was sampled at 12 bits, 2 usec sampling using the ri.
• data was taken from 10:10 to 11:45 and then  13:32 to 16:29
  
• There was a 1.5 hour break for an observing session.
  

The data was processed in blocks of 100 ipps (1.6 seconds).
An average, max hold, and min hold was computed for each time sample of the 100 ipps.

A movie was made using the 10237  1.6 sec blocks of data (.avi) (.mov)

• The time is shown in the upper right: 10:10:11  to 16:29:59 (ast)
• Only the first 1.4 milliseconds of the 16ms ipp is plotted.
• white line: avg ipp over 1.6 secs
• red line: peak hold for the 100 ipps
• green line: min hold for 100 ipps
• The 3 frames are:
• top: modAnode read back
• middle: beam pulse output from the fiber optic board to the floating dec
• This is a negative going pulse.
  
• bottom: tail clipper pulse going from the fiber optic board to the buffer dec
• This is a negative going pulse
  
• The plots were scaled to the average deflection.
  
• Events:
• the min hold showed that the modAnode pulse dropping out then recovering during the beam pulse.
• 11:27:48 - 11:28:14
• 11:34:19 - 11:34:28
• 11:55:13 - 12:04:44

Plotting avg,min,max value of the 3 monitored  pulses (.ps) (.pdf)

• The  avg, min hold , and max hold pulse was computed  for each 1.6 sec block of 100 ipps.
  
• Each of these is plotted in a separate color:
  
• black is  the avg ipp
• green is the min hold ipp
• The gap between 12:00 and 13:30  had the 430 test stopped for another observing session.
• The beam command exiting hagen's box was also recorded. It never had a failure.
  
• page 1: entire experiment
• top: modAnode pulse (full scale)
• The average value decreased by about 10% during the 6 hours
• You can see the dropouts in the green min hold values.
  
• middle:The beam command exiting the fiber optic board on the wall in the transmitter room.
  
• The signal was active low. It remained low during every ipp.
• bottom: the tail clipper command exiting the fiber optic board on the wall in the transmitter room
• The signal was active low. It remained low during every ipp.
• page 2: blowup of min hold plots of the 3 signals around the  modAnode  dropouts
  
• the + are individual 1.6 sec min hold averages.
  
• top: modAnode
• the data between the purple lines is the 160 seconds was used to display the individual ipp failures (see below)
  
• middle: beam command from fiber optic board in tx room (active low)
  
• bottom: tail clipper command exiting from fiber optic board in tx room (active low).
  

plotting individual ipps during the failures (.ps) (.pdf)

• The modAnode value for 10000 ipps (160  seconds) were plotted starting at 10:39:31
  
• top: avg modAnode value for each ipp
• there were 23 ipps with problems
• the + are individual ipps.
  
• plotting the bad modAnode pulses .2 to .92 milliseconds of the ipps.
  
• The 23 ipps with bad modAnode pulses during this 160 seconds were plotted: 5 ipps/frame
• Page 1-2: continue  plotting bad ipps 5  ipps/frame.
• The modAnode is dropping out then recovering.
• When it recovers it does not come back to the original value.
• The larger the dropout, the lower the final equilibrium value.
  
• Page 3: blowup in time showing rise/fall times of the modAnode dropouts.
  
• The horizontal axis covers 120 usecs.
• The beam turnon was at 200 usecs
• It is taking about 5-6  usecs for the level to fall to its minimum value
  
• The + on the plots are spaced by 2 usecond.
  

SUMMARY:

• The modAnode pulse and the commanded beam pulse was monitored for 4.5  hours. (covering 6 hours of wall time)
  
• The sps was used to generate the ipp,rf,beam, and phase.
• This test included monitoring of the beam and tail clipper commanded pulses that left the fiber optic board on the wall in  the hv vault.
• These pulses did not fail when the modAnode monitored pulse dropped out.
• So the problem is after the fiber optic board on the wall in the hv vault.
  
• The modAnode pulse had intermittent dropouts that lasted for about 8.5 minutes over the 4.5 hours of monitoring.
• If it recovered to a static value before the end of the beam:
• This value was usually less than the original value
• The lower the dip, the lower the final value was.
  
• The modAnode signal never rose up outside of the beam on command (this occurred in previous tests).

processing:x101/191211/430txtest.pro

---

04dec19: monitor modAnode signal for 5.5 hours (top)

    The 03dec19 430 tx test continued to show a problem with the modAnode pulse. On the morning of 04dec19, the dp11 switch tube was replaced and
the test was rerun. The setup was:

•  Use the sps to generate the ipp,rf,beam and phase pulses.
  
• 16 ms ipp, 640 usec beam/rf, constant phase
• The commanded beam pulse (tx console output bnc) was also sampled to make sure the timing generator was not failing.
• data was sampled at 12 bits, 1 usec sampling using the ri.
• data was taken from 10:43 to 16:12
  

The data was processed in blocks of 100 ipps (1.6 seconds).
An average, max hold, and minhold was computed for each time sample of the 100 ipps.

A movie was made using the 13425  1.6 sec blocks of data (.avi) (.mov)

• The time is shown in the upper right: 10:43:52  to 16:12:51 (ast)
• Only the first 2 milliseconds of the 16ms ipp is plotted.
• white line: avg ipp over 1.6 secs
• red line: peak hold for the 100 ipps
• green line: min hold for 100 ipps
• The plot was scaled to the average modAnode deflection.
  
• Events:
• the min hold showed that the modAnode pulse dropping out then recovering during the beam pulse.
• 11:27:48 - 11:28:14
• 11:34:19 - 11:34:28
• 11:55:13 - 12:04:48
• modAnode outside of the beam pulse
• 10:44:34
  

Plotting avg,min,max value of modAnode pulse (.ps) (.pdf)

• The  average value for the modAnode pulse was computed for each of the 1.6 sec blocks.
  
• This was done separately for the  avg, min hold, and max hold ipps.
  
• Red is the max hold ipp
• black is  the avg ipp
• green is the min hold ipp
• page 1: avg modAnode pulse
  
• top: modAnode full scale
• The average value decreased by about 10% during the 5.5 hours
• You can see the dropouts in the green min hold values.
  
• 2nd: blowup around the dropouts
  
• You can see the green min hold dropouts.
  
• 3rd:blowup of  the commanded beam pulse during the modAnode dropouts
  
• this is the front panel output of the tx timing generator
• The beam pulse did not dip during the modAnode problems.
• Page 2: more blowups of failures
• top : failures around 11.5
• bottom: failures around 12:00
• The + are the individual 1.6 second minholds.
  

plotting individual ipps during the failures (.ps) (.pdf)

• The modAnode value for 10000 ipps (160  seconds) were plotted starting at 11:56:27
  
• top: avg modAnode value for each ipp
• there were 28 ipps with problems
• the + are individual ipps.
  
• plotting the bad modAnode pulses .1 to .8 milliseconds of the ipps.
  
• The 28 ipps with bad modAnode pulses during this 160 seconds were plotted: 4 ipps/frame
• Page 2-3: continue with over plotting 4 band ipps/frame.
• The modAnode is dropping out then recovering.
• When it recovers it does not come back to the original value.
• The larger the dropout, the lower the final equilibrium value.
  
• Page 4: blowup in time showing rise/fall times of the modAnode dropouts.
  
• The horizontal axis covers 110 usecs.
• The beam turnon was at 100 usecs
• It is taking about 5-6  usecs for the level to fall to its minimum value
  
• The + on the plots are spaced by 1 usecond.
  
• Page 5: modAnode high outside of beam pulse
• At 10:44:34 the modAnode went high outside of the beam pulse. This happened for 3 ipps
• Top: plot 5 ipps around the problem. Each ipp is a different color.
• The 640 usec beam pulse starts each ipp.
• Bottom: over plot the 5 ipps (each in different color)
• The modAnode goes high outside of the beam pulse, then immediately starts to decay.
  
• It takes about 2 ms to got 1/e
  
• I don't understand why the blue ipp decay stops, and then the purple ipp picks up (16 ms later) exactly where the blue one stopped?
  

SUMMARY:

• The modAnode pulse and the commanded beam pulse was monitored for 5.5  hours.
• The sps was used to generate the ipp,rf,beam, and phase.
  
• The modAnode pulse dropped out during in 3 different instances. The longest period lasted about 9 minutes
• If it recovered to a static value before the end of the beam:
• This value was usually less than the original value
• The lower the dip, the lower the final value was.
  
• The modAnode signal rose up after the beam turned off and then immediately started to decay.
• The timing of the decay was a bit strange. On 1 ipp it rose up, decayed and then was turned off, only to reappear at the same spot in the ipp 16 ms later.
  
• The commanded beam pulse did not glitch when the modAnode dropped out.
• Previously it looked like the modAnode dropouts started 2 hours after the start of transmission.
• on 04dec19 we saw a few dropouts after 30 minutes of running (so there is no magic with the 2 hour cases).
• After the main source of dropouts around 12:00, the system ran for 2 hours without any more dropouts.
• Changing the dp11 switch tube did not solve the problem with the modAnode dropouts.
  

processing:x101/191204/430txtest.pro

---

03dec19: monitor modAnode signal for 6 hours (top)

    The 25Nov19 monitor test of the tx 430 mod anode was repeated on 03dec19 from 10:21 to 17:24 The  setup was:

•  Use the sps to generate the ipp,rf,beam and phase pulses.
  
• 16 ms ipp, 640 usec beam/rf, constant phase
• The commanded beam pulse (tx console output bnc) was also sampled to make sure the timing generator was not failing.
• data was sampled at 12 bits, 1 usec sampling using the ri.

The data was processed in blocks of 100 ipps (1.6 seconds).
An average, max hold, and minhold was computed for each time sample of the 100 ipps.

Notes on the data acquisition:

• data taking stopped at 14:48, it was restarted at 15:53
• I had requested the ri to take 999999 buffers. At 16ms/buffer this is only 4.4 hours.
• After this run i upped the numbufs to 9999999..
  
• When  processing the data there were 2 bad records in the 80GB file.
  
• I ended up skipping about 100 ipps around these bad spots.
  

A movie was made using the 13425  1.6 sec blocks of data (.avi) (.mov)

• The time is shown in the upper right: 10:21:49  to 17:24:30 (ast)
• There is a jump in time from 14:48 to 15:53
  
• Only the first 2 milliseconds of the 16ms ipp is plotted.
• white line: avg ipp over 1.6 secs
• red line: peak hold for the 100 ipps
• green line: min hold for 100 ipps
• The plot was scaled to the average modAnode deflection.
  
• Events:
• 12:33:02 to 12:42:30 the min hold showed that the modAnode pulse dropping out then recovering during the beam pulse.

Plotting avg,min,max value of modAnode pulse (.ps) (.pdf)

• The  average value for the modAnode pulse was computed for each of the 1.6 sec blocks.
  
• This was done separately for the  avg, min hold, and max hold ipps.
  
• Red is the max hold ipp
• black is  the avg ipp
• green is the min hold ipp
• page 1: avg modAnode pulse
  
• top: modAnode full scale
• The average value decreased by about 10% during the 7 hours
• the gap between 14.8 and 15.9 is when datataking stopped
• You can see the dropouts in the green min hold values.
  
• 2nd: blowup around the dropouts
  
• You can see the green min hold dropouts. They lasted for about 9 minutes.
  
• 3rd:blowup of  the commanded beam pulse during the modAnode dropouts
  
• this is the front panel output of the tx timing generator
• The beam pulse did not dip during the modAnode problems.
  

plotting individual ipps during the failures (.ps) (.pdf)

• The modAnode value for 10000 ipps (160  seconds) were plotted starting at 12:35:17
• This was during the 9  minutes of dropouts.
• top: avg modAnode value for each ipp
• there were about 35 ipps with problems
  
• plotting the bad modAnode pulses
• The 35 ipps with bad modAnode pulses during this 160 seconds were plotted: 5 ipps/frame
• Page 2-3: continue with over plotting 5 band ipps/frame.
• The modAnode is dropping out then recovering.
• When it recovers it does not come back to the original value.
• The larger the dropout, the lower the final equilibrium value.
  
• Page 4: blowup in time showing rise/fall times of the modAnode dropouts.
  
• The horizontal axis covers 50 usecs.
• The beam turnon was at 100 usecs
• It is taking about 5-6  usecs for the level to fall to its minimum value
  
• The + on the plots are spaced
  
• bottom: over plot the 6 ipps that had dropouts
• 3  ipps started off  low and then rose to full power

SUMMARY:

• The modAnode pulse and the commanded beam pulse was monitored for 7  hours.
• The sps was used to generate the ipp,rf,beam, and phase.
  
• The modAnode pulse dropped out during many ipps for about 9 minutes
• If it recovered to a static value before the end of the beam:
• This value was usually less than the original value
• The lower the dip, the lower the final value was.
  
• The modAnode signal never rose up after the beam turned off (this occurred once on 18Nov19)
• The commanded beam pulse did not glitch when the modAnode dropped out.
• On 25Nov19:
• we started at 10:15
• the modAnode had dropouts from 12:45 to 13:12 (27 minutes)
• we finished at 14:30
• we only had dropouts during these 27 minutes.
  
• On 03dec19
• we started at 10:21
• the modAnode had dropouts from 12:33 to 12:42 (about 9 minutes)
• we ended at 17:24
• we only had dropouts for these 9 minutes
• It is a bit strange that after 2 to 2.5 hours we start to get dropouts for 10-15 minutes, and then they go away for the next 5 hours.
  

processing:x101/191203/430txtest.pro

---

25Nov19: monitor modAnode signal for 4+hours (top)

    The 18Nov19 monitor test of the tx 430 mod anode was repeated on 25Nov11 from 10:15 to 14:30. The new setup was:

•  Use the sps to generate the ipp,rf,beam and phase pulses. The 18Nov19 run used the local timing generator of the tx console
• The ipps were now locked to the station clock.. no drifting.
• 16ms ipp, 640 usec rf pulse
• The commanded rf (tx console output bnc) was also sampled to make sure the timing generator was not failing.
• data was sampled at 12 bits, 1 usec sampling using the ri.

The data was processed in blocks of 100 ipps (1.6 seconds).
An average, max hold, and minhold was computed for each time sample of the 100 ipps.

A movie was made using the 9616  1.6 sec blocks of data (.avi) (.mov)

• The time is shown in the upper right: 10:14:31 to 14:30 (ast)
  
• Only the first 2 milliseconds of the 16ms ipp is plotted.
• white line: avg ipp over 1.6 secs
• red line: peak hold for the 100 ipps
• green line: min hold for 100 ipps
• The plot was scaled to the average modAnode deflection.
  
• Events:
• 12:45 to 13:12 the min hold showed that the modAnode pulse dropping out then recovering during the beam pulse.

Plotting avg,min,max value of modAnode pulse (.ps) (.pdf)

• The  average value for the modAnode pulse was computed for the 1.6 sec blocks.
  
• This was done separately for the  avg, min hold, and max hold ipps.
  
• Red is the max hold ipp
• black is  the avg ipp
• green is the min hold ipp
• page 1: avg modAnode and avg rfpulse
  
• top: modAnode full scale
• 2nd: blowup the vertical access
• You can see the green min hold dropouts
• The average value  is also decreasing over time: about 7% during the 4.5 hours
• 3rd: the commanded   rf pulse.
• this is the front panel output of the tx timing generator
• I  probably should have used the beam pulse rather than the rf (since the beam generates the modAnode pulse).
• the rf pulse did not have any dropouts when the modAnode  was failing.
• Bottom: blowup of the commanded rf pulse
  
• Page 2: minHoldValues blowup around modAnode dropouts.
  
• The dropouts lasted for about 15 minutes
• each + is a 1.6 second min Hold value
  

plotting individual ipps during the failures (.ps) (.pdf)

• The modAnode value for 600 ipps (9.6 seconds) were plotted starting at 12:53:51
• This was during the 15 minutes of dropouts.
• top: avg modAnode value for each ipp
• there were 6 dropouts in the 9.6 seconds
• middle: over plot the 600 ipps (just the first millisecond of each ipp)
• bottom: over plot the 6 ipps that had dropouts
• 3  ipps started off  low and then rose to full power
• 3  ipps rose to full power and then dropped suddenly. They then slowly rose back up to full value
• the rate of return to full power appears to be the same for these ipps.

SUMMARY:

• The modAnode pulse and the commanded rf pulse was monitored for 4.5 hours.
• The sps was used to generate the ipp,rf,beam, and phase.
  
• The modAnode pulse dropped out during many ipps for about 15 minutes
• It usually returned to the normal level before the end of the beam.
• The modAnode signal never rose up after the beam turned off (this occurred once on 18Nov19)
• The commanded rf pulse did not glitch when the modAnode dropped out.
• I probably should have monitored the beam rather than the rf..
  
• But this probably says that the problem is not with the timing generator or sps.
  

processing:x101/191125/430txtest.pro

---

18Nov19: monitor modAnode signal for 3+ hours (top)

    Problems with the 430tx transmitter were fixed during the beginning of November. On 18Nov19 the modAnode signal from the 430MHz Tx was monitored for about 3 hours. The setup was:

• the 430 tx console was set to local timing mode, 16ms ipp, 4% duty cycle.
  
• console monitor output was switched to modAnode
• a cable was run from the  430tx console ->  1-10 db variable attenuator  -> opAmp -> scope(1megaOhm inp) -> ri input
• the level was adjusted to give  a 1 volt level when the rf pulse was on (modAnode signal hi). The ri has +/- 2.5 V into 50 ohms.
• The datataking setup had 12 bit, 1 usecond continuous sampling with an ipp of 16000 usecs.

Processing the data:

    Data was processed in 100 ipp blocks (1.6 sec). The were 6806 block (10890 seconds). For each block

• for the 100 ipps of a block: compute the min,max and avg value for each usec of the 16000 usec ipp
• This should catch spikes and dropouts
• Over plotting the ipps showed that the ipp of the tx console was not 16000 usecs.
  
• Each block of data was shifted to align  it with the first block
• An initial shift of 11.87 usec/block was used to get the block close.
• A 641 usec mask (16000*4%) was then convolved with each block to get the final shift needed,.
• The actual ipp turned out to be about 16000.12 usecs.

A movie was made using the 6806 1.6 sec blocks of data (.avi) (.mov)

• The time is shown in the upper right: 12:28:30 to 15:29:58 (ast)
  
• Only the first 2 milliseconds of the 16ms ipp is plotted.
• white line: avg ipp over 1.6 secs
• red line: peak hold for the 100 ipps
• green line: min hold for 100 ipps
• The plot was scaled to the average modAnode deflection.
  
• Events:
• 14:43:16.  this block has a strange glitch

Plotting the glitch at 14:43:16 and the 430tx oscillator performance (.ps) (.pdf)

• Individual 16 ms ipps were examined around the glitch.
  
•  Page 1: ipps around the glitch
• top: 5 ipps about the glitch
• The green ipp has the glitch
• the modAnode peak after the glitch is lower than before the glitch
• bottom: over plot the 5 ipps
  
• The green line started falling during the rf pulse
• it went away when the rf pulse was turned off
• 736 usecs after the end of the rf pulse, the signal reappeared (at close to the same decay rate position during the pulse)
• 320 usecs after it reappeared, the signal jumped back to the on pulse level (this was 1359 usecs after the initial drop).
• 3490 usecs after it reached full scale it started to decay again
• This was probably the capacitor bank being drained.
  
• It did not get back to off level by the end of the ipp.
• Page 2: plot the avg pulse height around the glitch
• Top: avg height of the pulse for 200 ipps about the glitch
• The pulse height dropped by 5% after the glitch.
• It took about 30 ipps (about 48 seconds) to recover. It may have taken the capacitor bank that long to recover fully.
• Bottom:
• the dc level for the pulse during the pulse. It dropped by about 2% after the glitch.
• Page 3: 430tx timing control box  local oscillator.
• The timing control box has a 5MHz crystal oscillator that it uses  to generate the ipp and duty cycle when in local timing  mode.
• It is also used to generate the timing when checking the validity of the ipp and the pulse lengths (even when in remote timing)
• If the oscillator is failing, it could cause problems with the protection circuitry.
  
• Top: The shift (in usecs) needed to align the avg ipp of each  1.6 sec block (100 ipps)
• I used the ri 16000 ipp as the time reference (it is slaved to the station clock).
• The black line shows the data, the red line is a linear fit.
• The shift was changing by 7.4114 usec/ 1.6 seconds.
• the actual ipp was 16000.12 usecs
• The oscillator freq error was 1 part in 7.5 e-6.
  
• 2nd: oscillator drift during the 3 hours (in usecs)
• This is just the data - the linear fit.
  
• 3rd: a blowup of the oscillator drift for the first 50 seconds.
• looks like a 11.2 second oscillation
• bottom: oscillator drift rate
• the oscillator drift/ (time since start).
• the drift gets up to 3*10-8 parts/sec

Summary:

• The modAnode  signal was monitored for about 3 hours, sampling continuously at 1 usec
• A single glitch was found around 14:43:16
• the modAnode came back on.. If we were transmitting into the dome, this would probably have blown up the receiver.
• The local oscillator for the timing box is off in frequency by 1 part in 7.5e-6 (37 Hz at 5MHz).
• the oscillator drift rate got up to 3e-8 secs/sec
• We don't know if the glitch was caused by the floating dec of the tx or a glitch in the timing generator.
  
• We should have also monitored the rf pulse sent from the timing console to the 430 tx.
• Next time we could also use the remote time system to generate the pulses.
  

processing: x101/191118/430txtest.pro

---

02oct19: clp and tpsd rf pulse response, 430 power meter. (top)

    Project t3365 was using  the 430 tx with:

• 60 seconds clp with a  2 usec baud and a 495 usec rf pulse length.
• 30 seconds topside with a +/- 62.5 Khz freq hop and a 500 usec rf pulse length
• 10 seconds power profile with a 13 length barker code using a 4 usec baud.
• The usrp took complex voltage data with a 25 MHz bandwidth.

Plotting the rf pulse response

    The plots show the transmitter response to the rf pulse (.ps) (.pdf)

• The transmitter signal comes from a coupler after the klystrons
• 40 usecs of the transmitter pulse are plotted
• Page 1: clp  rf pulse
• Top complex voltage:
•  black is I, red is Q.  Y units are a/d counts
• the blue dashed lines are the baud spacing.
• Middle: the phase of the code (atan(q/i) in degrees
• Bottom: Normalized  power
  
• The power was normalized to the median value during the 495 usec pulse
  
• You can  see dips and then jumps of about 20% at the transitions.
• The variation lasts for > 1 usec.
• Page 2: topside rf pulse
• Top: voltage
• this is either the +62.5 or -62.5 khz offset.
  
• middle: phase
• Bottom: power
• It is relatively stable.
• Page 3-4 spectrum of the rf pulse
• page 3: clp.
  
• the red dashed lines are every .5 MHz (where the nulls should be)
• the right side looks correct. The 1st lobe down by 11 rather than 13 db (for (sinx/x)^2)
  
• the left side is  incorrect. The first null is down 23 db.
  
• page 4: topside
• top +/- 5 MHz
• The 30 +/- 62.5 Khz signal was put through a 256Khz butterworth filter before sending it to the tx console (to get rid of the 1 MHz tones from the d/a converter),
  
• bottom: blowup  +/- .6 MHz
• You can see the -62.5 Khz offset of the sine wave.
  

Plotting the dome 430 power meter.

    A 430 MHz power meter in the dome measures the 430 power  before going out the horn

• It is a ladybug Lb480A power meter  (more info)
• It measured the Peak power within an rf  pulse
• This is the value displayed in the large window of the gui.
  
• It also measures the average power within an rf pulse
  

    The plot shows the power meter values during the time of the rf pulses above (.ps) (.pdf)

• Top: 12 minutes of data around the display rf pulses
• Black the peak power within a pulse
• Read the average power within a pulse
• Blue dashed lines: 60 seconds of clp power
• Purple lines: 30 seconds of tpsd power
• Green lines: 10 seconds of power profile power.
• Middle: blowup around the ipps  plotted in above section.
• The clp Average power is less then the tpsd average power
• the clp  Peak power is greater then the tpsd peak power
• Bottom: Peak/Average power
• The clp Peak power is 18% higher than the average power
• the tpsd Peak  power is 7% higher than the average power.

SUMMARY:

• CLP pulse
  
• The  2usec baud transitions are taking up to 1 baud to stabilize
• There is a 20% increase in clp power at the transitions.
• Topside pulse
  
• The phase an power are relatively stable during the rf pulse.
• Power meter
• clp:
• the peak power is 20 % higher than the average power
• This is caused by the power glitch at the baud transitions.
• The signal is taking up to 1 baud (2usecs) to stabilize.
• Topside
• the peak power is about 7% higher than the average power
• We have always used the peak power when processing the 430 data. We should be using the average pulse power.
• For the same rf pulse length  the transmits 4% more average power  with topside (no baud transitions).
  

processing: x101/191002/430pwrmeter.pro, 430baud.pro

---

18jan18:Usrp, downstairs blanking turns on late.

    Experiment T3150 ran a clp experiment with a 6ms ipp and a 346 usec rf pulse. The downstairs thumb wheel blanking was set to 160 usecs.

    Data was taken with the gregorian dome into the usrp sampling system (no ch ).

• rf signal dome ->downstairs if/lo at 260 MHz -> 40 MHz filter -> dana's blanker -> power combiner (A)-> usrp input
• txSamples from waveguide probe -> atten -> 30MHz IF amp -> atten -> mixer (30MHz ->260 MHz) -> power combiner (A) -> usrp

    The usrp sampled a combination of the rf signal  (coming from the 430 receiver upstairs.. about 3.5 usecs transit time) and the tx samples (coming from the waveguide probes in the tx room .. negligible delay).

blanking:

     The receiver signal is blanked (upstairs in the dome ) and down stairs in the receiver room.

• LO and Rf blanking
• This is done upstairs in the dome.
• If the signal is blanked upstairs at T0 secs, the blanking will appear downstairs around T0 + 3.5 usecs (travel time).
• Downstairs blanking.
• Dana's blanker is used to blank the  receiver signal during the tx pulse.
• It uses the test output next to the blanking thumb wheels.
  
• I think it is triggered with the protect pulse. It stays active for the tx pulse plus the extra delay added in the thumb wheels.

Problems with the data.

    The usrp data for t3150 was analyzed for

•  17jan18 (about 4 hours then the tx broke)
• the processing went normally
• 18jan18 (15:30 to 7:30 the next day).
• The program could not synchronize (in software) with the tx pulses...
• The program uses the blanking region before and after the tx pulse to synchronize with the ipp.

Looking at the 17,18th data, the blanking region at the start of the txIpp was about 10usecs shorter than normal.

The plots show a single tx pulse for 17 and 18 jan18 (.ps) (.pdf):

• Page 1: tx pulse voltages for 17th and 18th jan18
• Black is the I voltage, red is the Q voltage
• Top plot 17jan18
• Bottom plot 18jan18.
• the blanking region before the tx pulse is much shorter on 18jan18.
• Page 2: Blowup around Tx start showing power.
• Vertical lines are drawn for events the occur (downstairs in the receiver room)
• green: rf on pulse
• blue: beam on signal. (this occurs 10 usecs before the rf on
  
• purple: tx ipp ( or protect pulse). this occurs 2 usecs before the beam on.
• moving forward in time:
• 0  usecs: txipp/protect pulse.
• 2  usecs: beam on request
• 12usecs: rf on
  
• Top: 40 ns resolution power vs time.
  
• black is 17jan18 power, red is 18jan18 power.
• The vertical scale is logarithmic.
• Bottom: power smoothed to 1 usec resolution.

What is happening:

• -12 usecs:
  
• downstairs protect pulse is generated.
• 17jan18. Dana's blanker receives this and immediately drops the rcvr power by about 25db
• 18jan18. the receiver power does not change. So dana's blanker did not blank the receiver signal
• -10usecs:
• the beam on request goes upstairs.
• -4.5 usecs
• 18jan18 rcvr signal shoots up.
• this level is much higher than the tx on power level (look at the page 1 voltages).
• The 6.5 usecs is about the time it would take a signal to go up to the platform and then come back down.
• this could be the monoplexor switching into the rf signal (or maybe some shot noise from the beam turning on?).
  
• It stays high for about 1 usec
• -3.  usecs:
• 18jan18 data gets blanked.

On 22jan18 we checked the test blanking output signal that went to dana's blanker.

• We found a 1 usec jitter in the blanking signal turning on.
• The protect signal from the 430 tx into this box, did not have this jitter (but is was less than 2Volts).
• When routing the test blanking signal to dana's blanker:
• dana's blanker worked correctly. it even followed the 1 usec jitter.
• So we don't have an explanation where the 10 usec delay came from..

processing: x101/180119/430txblnk.pro

---

01oct15: 440 and 1000 usec rf pulse:

    Data taken on 01oct15 was used to examine two different rf pulses:

• 1000 usec rf pulse with 2 usec baud
• The power was about 1 MWatt (measured upstairs) in the ch (only)
  
• 440 usec rf pulse with 1 usec baud.
• This was split upstairs into dome and ch. The combined pulse power (measured upstairs) was about 1.2 MWattts

Both of these pulses were modulated by the coded long pulse random pn code.
The plots show the different rf pulses (.ps) (.pdf)

• page 1: Voltage vs time.
• black in real, red is imaginary voltage
• top: 1000 usec pulse, 2 usec baud
• 2nd: blowup 1000Usec pulse .You can see the overshoots at the baud transitions
• 3rd: 440 usec pulse, 1 usec baud
• bottom: blowup 440 usec pulse.
• Page 2: power, phase vs time
• Top: total power in rfi pulse vs time.
• I averaged for 1 record (50, or 100 ipps) and then smoothed over time by 16 usecs
• The plot was normalized to the first 50 usecs of each pulse.
• You can see the power droop: about 4 usecs in 1000 usecs.
• The droop rate is the same for pulse rf pulse lengths.
• Bottom:Phase of code vs time:
• the arctangent of(imaginary/real) voltage samples was plotted
• the green lines are fits to the change of phase (excluding the transitions).
• There is a .01436 Degrees/usec chirp.
• Page 3: average power over entire rf pulse vs time
• the average power over each rf pulse was computed and then normalized to the median for 50 (25) seconds.
• Top: 1000 usec rf pulse. ipp = 20 millisecs (so sampling rate is 50 Hz) for 50 seconds.
  
• 2nd: blowup of 1000 usec rfpulse pwr vs time. You can see a periodic oscillation for the 1 second of data
  
• 3rd: 440 usec rf pulse, ipp=10 millisecs (so sampling rate is 100 Hz) for 25 seconds,.
• bottom: 440 usec rf pulse blowup.
• Page 4: spectrum of total pwr in rf pulse vs time
• i computed abs(fft(rftotpwr)) using the data from page 3.
• Top: spectra of 20 ms ipp. The spike at 10Hz is the alias of 60 Hz sampled with a 25Hz bandwidth
• bottom: spectra of 10ms ipp. The spike at 40 Hz is the alias of 60 Hz in a 50 Hz bw.

Summary:

• The total power droops by 4% in 1000 useconds. This rate is the same for 440 and 1000 usec pulses.
• There is a chirp across the rf pulse of .01436 degrees/usecond.
• There is a 60 Hz component in the total power in rf pulse vs time.
  

processing: x101/151001/chktxdroop.pro

---

27mar12 Mod anode not staying completely off after beam off.

    The mod anode was not staying completely off after the rf pulse turned off. The mod anode output of the transmitter console is plotted below for various HV settings. These measurements were taken with a single klystron (klyA) installed.'

• The yellow trace is the modAnode. The green trace is the beam (or rf?).
  
• hv=60KV (.png)
  
• hv=70KV (.png)
  
• hv=80KV (.png)
• The modAnode rises 1.4 divisions in about 800 usecs.
  
• The modAnode on to off is  5.8 divisions. This is about 40 Kv
• So the modAnode is moving 1.5/5.8 - 25% of the modAnode voltage during this time.. about 10Kv
• This  is a lot more than the 5kv supply so this supply is probably not the problem.
  

---

19oct10: scope trace of protect turnoff. (top)

    On 19oct10 we did a test to look at how quickly the 430 ch  receiver could be turned on after the rf pulse of the transmitter goes off. The setup was:

• High voltage off, ipa off, system on.
• 429 MHz cw tone transmitted from hilltop antenna
• rf signal ch crr1, 30 MHz IF, to 30 MHz amplifier, no filter, and then IF to scope.
• thumbwheel delays set to 1 usec.
• Scope triggered on falling edge of rf pulse.
  
• The image below shows the signals:
• Rf pulse:
  
• leaving hagens box.
• protect cmd: 
• Leaving hagens box.
• there is a 6 usec delay in hagens box from rf falling to protect cmd falling
• ACK:
  
• At hagens box.
• The protect cmd falls and is sent upstairs.
  
• The monoplexor waits 20 usecs then the diode current is turned off
• When the current thru the diode drops to a certain level, the ack level will drop
• it then comes back downstairs and shows up on the output of hagens box.
• Ack drops (downstairs) 27 usecs after protect cmd drops (downstairs).
• This is the 7 usec round trip time + the 20 usec monoplexor delay
• IF signal: 
• This is the 30 MHz IF voltage.
• There is a spike 8 usecs after protect falls.
• this changes with the thumbwheel delay.Increaing this delay moves this to the right.
• The wheel is labeled IF and LO blanking.
  
• Quest: why the 8 usec delay from protect? Is this a round trip delay??
  
• After 8 usecs the noise increases. This is coming down the cable from upstairs (but the monoplexor is still active). I wonder if the upstairs rf blanking is still active.
• When ACK falls (downstairs) we wait 12 usecs before the IF signal comes back up.
• Quest: where is this delay coming from?

Summary:

• rf Falls to IF up: 46.5 usecs.
• This corresponds to a minimum range of:
• (46.5 -7)*.15=5.8 km.
  
• The -7 is the 3.5 usecs for rf pulse to go up and the 3.5 usecs for the IF Signal to come down.
  
• 12 usecs of unknown origin ack falls to IF up.
  

ACK at hagens box, IF SIGNAL, RF pulse at hagens box, protectCmd leaving hagens box

---

07oct10: 430 transmit pulse sequence timing (top)

The picture below shows the 430 transmiter transmit sequence. a (.pdf version) is also available)