(r34) AMC13DebugLog < Main

Please log AMC13 test activity below, blog style (new entries at top)

2012-11-21, hill

Attempting an initial installation of the VadaTech Commercial MCH card for the uTCA crate.

Procedure:

Install the VadaTech module in the MCH2
Initial attempt at IPbus-based communication with the uTCA crate results in successful pinging and control of the AMC13 in MCH1 and also the AMC13 in a non-MCH slot, but unsuccessful pinging of mCTR2s
Relevant Documents are...
10/100 Ethernet has the default IP address 192.168.1.252, which conflicts with our IP assignment scheme for the AMC13s.
According to the 'Getting Started Guide', the default IP address of the GbE is 192.168.40.250. Unsuccessful ping to this address.
Run Eric's pinger.py to see what I can talk to. Within the range 192.168.1.255.......1, I can only see the AMC13s. This was to check and make sure the VadaTech MCH card didn't magically take on the same IP address as the NAT MCH card

Despite my not being able to ping the GbE, I am going to try and connect anyway, as the documentation suggests.

      [cms2] /home/chill90 > ssh root@192.168.40.250
      ssh: connect to host 192.168.40.250 port 22: Connection timed out
      [cms2] /home/chill90 > ssh 192.168.40.250
      ssh: connect to host 192.168.40.250 port 22: Connection timed out

Not surprisingly, no luck here. I get similar results if I try and access the IP address via a web browser

Next try connecting via the 10/100 Ethernet Port at 192.168.1.252. This pings successfully!
Eric has now take over.
TelNet into the card successfully.
- > telnet 192.168.1.252
- Username: root
- Password: root
The HUB card will be running Linux. The command-line interface is based on the IPMI v2.0. The procedure to reassign the IP addresses of the ethernet ports is as follows
1. Open the file /etc/rc.d/rc.conf for editing
2. net interface 0 is the 10/100 Ethernet port. Edit IPADDR0 and/or NETMASK0 and/or BROADCAST0 as desired
3. net interface 1 is the GbE port. Edit IPADDR1 and/or NETMASK1 and/or BROADCAST1 as desired
4. Only one of either GATEWAY0 or GATEWAY1 should be set!! The MCH will use the device with the set 'GATEWAY' value to send traffic to other subnets and networks
5. Power cycle the MCH for the changes to take effect

2012-10-26, hazen

NOTE: Don't design any software to use this feature yet. Some small details are likely to change, but the concept will remain.

Working on IP address setting by IPMI. Using Tom's recipe:

export IPMITOOLARGS="-H 192.168.1.11 -P \"\" -T 0x82 -B 0 -b 7"

# read Spartan IP address
>  ipmitool $IPMITOOLARGS -t 0xa4 raw 0x32 0x34 0 0 7 4
f6 01 a8 c0

# read Virtex IP address (bus addr byte determined empirically!)
>  ipmitool $IPMITOOLARGS -t 0xa4 raw 0x32 0x34 1 0 7 4
f7 01 a8 c0

Changing the address:

# set spartan address low byte to 128
> ipmitool $IPMITOOLARGS -t 0xa4 raw 0x32 0x33 0 0 7 1 0x80
# set virtex address low byte to 129
> $IPMITOOLARGS -t 0xa4 raw 0x32 0x33 1 0 7 1 0x81

[cms2] /home/hazen > ping 192.168.1.128
PING 192.168.1.128 (192.168.1.128) 56(84) bytes of data.
64 bytes from 192.168.1.128: icmp_seq=1 ttl=64 time=0.240 ms
64 bytes from 192.168.1.128: icmp_seq=2 ttl=64 time=0.060 ms

It works!

2012-09-28, eric

***This has since been fixed. There was a discrepancy between our local Makefile and the release Makefile.

Charlie has complained of ghosts in the machine. All is OK if I do the following:

Goto cms1, run ./periodic_12hz
Run DCCdiagnose.exe, ttc/trig 1 (light blinks)
Goto AMC13Tool_3 directory, run AMC13Tool
do init.amc
turn on triggers for a while

Attempting to reproduce the EvN mismatch symptom described by Jeremy so Wu can look at it by remote control. First, update mCTR2s to firmware 0.5.20.

my AMC13Tool has dependency problems. Create a directory AMC13Tool_4 and copy Charlie's 11_5_5 exe and so there. That seems OK, but the stock AMC13Tool one gets with ". ~daqowner/dist/etc/env.sh; AMC13Tool.exe" segfaults immediately frown

2012-09-14, eric and charlie

Attempting to reproduce the result below and look at uHTR spy buffer using tools here. Perl script check_xxxx.pl loops generating one software L1A and checking using dump_DTC.exe for EvN errors.

Found one! AMC13 Data is here. uHTR DAQ spy output:

> spy
  0000 2 00ff 
  0001 3 ffff Event number 16777215 (0xffffff)
  0002 3 8000 
  0003 3 03ee Orbit number 0, submodule number 1006 (0x3ee)
  0004 3 6000 Format 6, BCN 0 (0x0)
  0005 3 0017 Presamples 2, TP words 0 
  0006 3 4511 Unsuppressed 0, compact mode 1, firmware rev 0x511
  0007 3 0028 Flavor 0, Pipeline length 40 
  0008 3 2000 NS 4 WC 0
  0009 3 ffff 
  000a 3 0000 
  000b 1 ff00

This is not helpful as the EvN is for some reason all 1's.

2012-09-13, eric and charlie

Testing V0.5.11 uCTR firmware with AMC13 V=0x17 S=0xb firmware

At 1kHz fixed trigger rate for 1s, take some data. (here: warning! binary file).

See that uHTR stay in sync with AMC13 but there are some strangely-corrupted events.

Starting to read file...
First EvN is 1 (0x000001)
AMC13 EvN = 0x00000012  uHTR(10) EvN = 0x00451112 !!
AMC13 EvN = 0x0000004e  uHTR(4) EvN = 0x0045114e !!
AMC13 EvN = 0x00000218  uHTR(10) EvN = 0x0045111a !!
After reading 940 events, Last EvN is 940 (0x0003ac)

Note the 4511's. Here is a dump of EvN 12

AMC13 EvN = 0x00000012  uHTR(10) EvN = 0x00451112 !!
FED:   0 EvN: 000012  BcN: 1d5  OrN: 0008a747  TTS: 0/0 EvTyp: 1  CalTyp: 0 Size: 26
UHTR  4 [  12] EvN 000012 BcN 1d5 OrN 17
  0: 0412
  1: 0000
  2: 8000
  3: bbee
  4: 61d5
  5: 0017
  6: 4511
  7: 0028
  8: 2000
  9: ffff
 10: 0000
 11: 1208
UHTR 10 [  12] EvN 451112 BcN 1d7 OrN 04
  0: 0012
  1: 4511
  2: bbee
  3: 2000
  4: 61d7
  5: 0000
  6: 4539
  7: 2000
  8: ffff
  9: 0000
 10: 1208
 11: 0000

Note that in the 2nd uHTR payload the high byte of the 1st word is 00 (should be 0a) plus the 2nd word is 4511 (should be 0000). The 8000 word is missing, and the rest of the words are shifted up by one. The word count in the AMC13 header (see below) is 0b instead of 0c, with a zero fill word added (correctly) by the AMC13.

Below is a raw dump of the data from the AMC13 excerpted. (deadbeef and following word count added by software). Note the 4511 ffff

000770 deadbeef 0000001a 1d500008 51000012
000780 008a7470 00000000 00000010 00170410
000790 00000000 00000000 0000c00c 00000000
0007a0 00000000 0000c00b 00000412 bbee8000
0007b0 001761d5 00284511 ffff2000 12080000
0007c0 45110012 2000bbee 000061d7 20004539
0007d0 0000ffff 00001208 ad8c0000 a000000d
0007e0 deadbeef 0000001a 1d500008 51000013
0007f0 008a7530 00000000 00000010 00170410
000800 00000000 00000000 0000c00c 00000000
000810 00000000 0000c00c 00000413 1bee8000
000820 001761d5 00284511 ffff2000 13080000
000830 00000a13 1bee8000 001761d5 00284511
000840 ffff2000 13080000 766b0000 a000000d

2012-08-31, hazen

Working on AMC13 "pre-ship" certification test. Firmware up through virtex=0x10 have LSC/LDC implemented and can send DAQ data in the old "Wu" format. ] Enable by turning on "SLINK Enable" (bit 1 in reg 1). There are secret counters which monitor the LDC:

0x10 LDC_accept_cntr
0x11 LDC_abort_cntr
0x12 LDC_ACK_cntr
0x13 LDC_event_cntr
0x14 LDC_word_cntr
0x15 LDC_CRC_bad_cntr
0x16 LDC_SEQ_bad_cntr
0x17 LDC_wc_bad_cntr
0x18 LDC_frame_bad_cntr
0x19 LDC_buf_ovf_cntr
0x1a LDC_CMSCRC_err

2012-08-06, hazen

Download Jeremy's HEAD version from 8/2/12. Compile 3 test BIT/MCS files:

Default (jumper/flash set IP address)
Fixed I/P address 192.168.1.34
Fixed I/P address 192.168.115.254

Flash the 1.34 version. Responds to ping and mCTR2tool ok. Now try the fixed 115.254 version. Argh, the "reload" command once again doesn't work. Cycle crate power. It works... can talk to board at above IP.

Now try flashing the "default" firmware. "reload" works. Can still access at 115.254. Verify that SPI flash IP address is 1.32. Remove jumper J1204 and re-power. Still at 115.254 (sigh!). Maybe messed up compiling. Confusion: Slot 5 board had no J1204, but was installed in board in slot 11. Unplug slot 11 board altogether. Still get a ping response from 115.254.

Re-install J1204 jumper on slot 4 and re-power. Still get response from 115.254.

Compile new version with original code but jumper-controlled backup address changed from 115.254 to 115.240. Program this one to flash. (J1204 is installed). Do "reload". Responds at 115.240 as expected. Remove J1204 and re-power. Still at 115.240. Sigh.

Flash same firmware into other board. Same results.

Try setting IP address to 192.168.115.40 (maybe 115 is somehow special). No change.

Per Jeremy's request, program the 0.4.03 version from the web page into slot 11 board. Once again "reload" doesn't work. Power cycle. Can't access the board at all.

2012-08-02, hazen

Re-flash '2012-07-16a' version into both modules. Fix a few bugs in AMC13DaqTest.exe so it can handle nested scripts correctly. Run a few tests with script AMC13DaqTest/test.amc and see same event size in both mCTR. More tomorrow.

2012-08-01, hazen

Porting to ISE 14. Removed cores defined in ipcore_dir and substitute ones from Wu_Cores_30July.zip e-mailed by Wu. Import into ISE 14.1 and re-compile. Compiles OK and superficially works. However, seeing some discrepancies in event length between old and new firmwares.

2012-07-16, hazen

Merged changes are here: ctr2_merge_Wu.zip. Now add them to project and re-compile. Program into two mCTR2 in slots 5, 11 at IP addresses 192.168.1.32 and 192.168.1.40.

2012-07-13, hazen

Wu has made some minor changes to the code to fix the link reset problem. His changes are here: ctr2_Wu_Fix.zip but must be merged with my firmware from here: ctr2_trunk_with_DAQ_2012-07-12a_esh.zip.

I will start on this next week.

2012-07-12, hazen

Temporary solution to IP addressing: Assign a totally fixed address. In ctr2_uhtr1600.v just set ip_addr to a fixed value. Build two firmwares with addresses 192.168.1.32 and 192.168.1.40. The board now in slot 5 is 32 and slot 11 is 40.

DAQ links do not work with this firmware: ctr2_trunk_with_DAQ_2012-07-12a_esh.zip

Will ask Wu for help.

2012-07-11, hazen

Add two lines below to UCF and recompile per Wu's suggestion:

  NET "*/DAQ_Link_wu/UsrClk" TNM_NET = "DAQ_UsrClk";
  TIMESPEC "TS_DAQ_UsrClk" = PERIOD "DAQ_UsrClk" 8ns HIGH 50%;

MiniCTR2 Programming:

(instructions below from MN)

  1) Program the 4.02 firmware bitfile from JTAG.
  2) mCTR2tool.exe 192.168.115.254 -t uhtr
  3) select your device (1).  Then FLASH -> PROGRAM (you'll need an MCS file)
      If Wu's firmware is branched from ours after 3.01, then you can do this with Wu's firmware.
  To change the IP address, use the SPICFG menu of mCTR2tool.

Here is what I did:

Generate MCS file with "generic parallel flash" and zero offset
mCTR2tool.exe 192.168.1.32 -t uhtr
"FLASH" then "PROG"

Doesn't work. It's a 128M SPI flash. That works better!

IP Addressing is a problem. NAT-MCH doesn't seem to route packets to any addresses outside 192.168.1.xxx, even though the settings are now correct:

[cms2] /home/hazen/work/uHTR_Firmware > telnet 192.168.1.11
Trying 192.168.1.11...
Connected to 192.168.1.11.
Escape character is '^]'.

Welcome to NAT-MCH

nat> ifconfig
network interface nat0:
  IP address:        192.168.1.11
  broadcast address: 192.168.255.255
  netmask:           255.255.0.0
nat>

2012-07-10, hazen

Attempting to integrate Wu's DAQ link into uHTR 4.02 release under ISE 13.3 on VirtualBox ampere.bu.edu on Eric's desktop machine.

Rename Wu's MiniCTR.vhd to DAQ_Link_wu.vhd and put in .../ctr2/ctr2_uhtr1600/DAQ_Path/ in project
- Edit to remove chipscope stuff
Edit DAQ_Link.vhd to instantiate DAQ_Link_wu.vhd
Copy and add other sources:
- Add CRC16D16.vhd
- Add Hamming.vhd
- copy ipcore_dir tree
- Add dataFIFO.ngc, DataBuf.ngc, TDP16_16.ngc, SDP16_16.ngc

2012-07-09, hazen

Basic DAQ event readout now seems to work. Have added several useful commands to my new tool AMC13DaqTest which is currently only in ~hazen/src/11_4_1.

HOWTO record test data

Log on to CMS1 as user daq and start DCCdiagnose.exe to control TTC system. Log on to CMS2 and run AMC13DaqTest.exe. For now:

  $ source ~hazen/environ.sh
  $ /src/11_5_1/hcal/hcalUpgrade/amc13/bin/linux/x86_64_slc5/AMC13DaqTest.exe

cms1 (DCCdiagnose.exe)	cms2 (AMC13DaqTest.exe)	Notes
	`en 5,11`	Enable AMC inputs (will change to 0,10 at some point). Also resets AMC13
`ttc/trig 4`		Disable TTC triggers (set to VME control only)
`ttc/cmd 2`		Issue TTC Event Count Reset command
`ttc/l1a 10`		Generate 10 triggers
	`st`	Display AMC13 status
	`Ctrl 0: 03000001`
	`LSC Link Down`
	`Ctrl 1: 000e0001`
	`run mode`
	`AMC Link status: 04100410`
	`Mon buffer page: 00000000 Evts: 0000000a words: 00000712`
	`df test.dat`	Dump events to file

You can dump the file in hex as follows:

   $ od -Ax -t x8 test.dat | less
   000000 00000712deadbeef 51000001c8c00008
   000010 000000059af5ddb0 000e041000000010
   000020 0000000000000000 000000000000c701
   000030 0000c70100000000 0800800000000401
   000040 0007000600058c8c 000b000a00090008
   000050 000f000e000d000c 0013001200110010
   000060 0017001600150014 001b001a00190018
   000070 001f001e001d001c 0023002200210020
     ....
   011af0 06f306f206f106f0 06f706f606f506f4
   011b00 06fb06fa06f906f8 705f06fe06fd06fc
   011b10 0000000000000a00 a00003891aca0000

2012-07-08, hazen

New firmware version 0xe. Still see problem#2 below. Investigating.

2012-07-07, hazen

Trying to reproduce a DAQ firmware problem.

Test #1 - failure to reset

(set Xilinx board to deliver fast triggers at 20kHz)

cms1 (DCCdiagnose.exe)	cms2 (AMC13Tool.exe)	Notes
	`0xd 0`	Always zero 2nd time
`ttc/trig 4`		Disable triggers
	`wv 3 410`	Enable AMC inputs
`ttc/trig 1`		Enable triggers
	`0xd 0x712`	FAIL: Should be zero!
	`rv d`	Read word count
	`rv d`	Read word count
	`wv 0 1`	Reset all
	`wv 0 1`	Reset all
	`wv 1 1`	Set run mode

This fails intermittently.

Test 2 with AMC13DaqTest.exe

cms1 (DCCdiagnose.exe)	cms2 (AMC13DaqTest.exe)	Notes
	`en 5,11`	Enable AMC inputs 5, 11 and reset AMC13
`ttc/cmd 1`		Reset EvN to 1
`ttc/l1a`		Make one trigger
	`rd`	Read and display event
	`ne`	advance to next event
`ttc/l1a`		Make trigger
	`rv 0x4000 10`	See corrupted data

This fails consistently:

It is probably a software problem, but it seems that I am competing for use of the TTC system so stop for now.

2012-07-05, hazen

Debugged AMC13 event readout code, added file dump feature to AMC13DaqTest.cc. Discovered a firmware issue which causes data to appear incorrect after the first event, reported to Wu.

E-mailed Jeremy to ask about word order question.

2012-07-01, hazen

Install new release 11_5_0 in daqowner. Do not set as default yet. Oops, 11_5_0 is buggy. Update immediately to 11_5_1 and make it the default.

2012-06-29, hazen

Checked in to CVS several updates of work done with Charlie.

HyperDAQ

DTCManager.cc/hh

HyperDAQ

AMC13_AddressTable_S6/V6.txt

Run Control

AMC13.cc/hh

startRun()

endRun()

AMCInputEnable()

nextEventSize()

readNextEvent()

2012-06-27, rohlf

Fix AMC13Tool to support hex version names in firmware files. Fork address tables with names AMC13_AddressTable_S7/V8.txt. FIXME: This address table is now incompatible with the one used by the rest of the software!

2012-06-12, hazen

First release firmware and instructions for DAQ in AMC13:

to set up L1A:
set L1a to about 20KHz
start dccdiagnose.exe
ttc/trig 4(turn off L1A)

run amc13tool
wv 3 410 (enable amc5 and amc11)
wv 1 1 (set run bit)
wv 0 1 (reset all)

ttc/cmd 0x28(reset orbit number)
ttc/trig 1 (enable L1A)

this will take 0x800 events in the memory and throw away all when the buffer is full.

In general, the initialization  and read out remains basically the same as with DCC2.

2012-05-24, rohlf

Sucessfully updated Spartan firmware from v3 to v6 at point 5. This was accomplished remotely in Bat. 40 at CERN with AMC13Tool by first writting Spartan v6 to flash at 0x000000 (its old location) as if programming the Header, power cycling the AMC13 module with the NAT tool, reprogramming the flash Header (0x000000), Golden (0x100000), Spartan (0x200000), and Virtex (0x400000), and then issuing the software command to reconfigure from flash.

2012-05-24, rohlf

Update firmware in cDAQ lab at CERN using python software. Older software (now named amc13_python_backup-23-may2012) was used to program Spartan firmware v6 into 0x0, power cycle crate, program Header (0x000000), Golden (0x100000), Spartan (0x200000), and Virtex (0x400000) where in the process a bug was discovered in programming the header due to incorrect erase of a single page, power cycling again, and verifying the configuration.

2012-04-27, hazen

Install HCAL xDAQ 11.4.0 (as daqowner) per instructions:

To install on a teststand (as daqowner) :
      wget http://cmshcalweb01.cern.ch/hcalsw/release/installDAQ_11_4_0.perl
      perl installDAQ_11_4_0.perl --mode=teststand
      ~daqowner/common/bin/pickRelease.sh (choose 11.4.0)

You can make a code-development area on a teststand or USC using
      perl installDAQ_11_4_0.perl --mode=[teststand|usc]
--ownsource=${HOME}/src/11_4_0 --packages=hcalUpgrade --cvsuser=[your afs id]
You can list multiple packages, separated by commas.

2012-01-10, hazen

Trying to program a new AMC13 MMC. Plug in to crate, connect JTAG ICE 3 cable to JTAG connector. Fails to program. Try to force power on with "pwr_on 9" (it's in AMC slot 4) but doesn't help.

Note:

  pwr_on [fru #]
  pwr_off [fru #]

  where [fru #] is the fru number starting with 5 to16. That means if you
  want to switch on e.g. AMC2 use the command "pwr_on 6".

Remembering wisdom of Tom Gorski:

unless

This should be taken care of by the uTCA backplane, no? Meanwhile, trying to program the thing in Wu's test fixture. Doesn't work. Checking to be sure jumpers are installed. AHA ... the required ECO (soldered wire) on the T3 board was missing. Now all is OK. One can indeed program the MMC on a new board in a MicroTCA crate.

Another observation: an AMC won't power up fully without the front panel hardware installed, as the handle switch must be depressed for the negotiation with the MMC to complete.

2011-12-06, heister

cms2: SLC5 updated, installation fine tuned, PyChips and microHAL installed for the "daq" user, all tested and works

2011-12-06, hazen

Install 2nd NIC on cmssun2 (old Dell computer, Ubuntu 10.04 OS). Cable to AMC13 in uTCA crate. Install IPBus test firmware from Wu. Install PyChips. Works! (I think).

Added AMC13FlashProgramming page.

2011-12-02, hazen

Started this log. One dead NAT-MCH at Boston. 2nd NAT-MCH shipped from MN by Jeremy.

Reset I/P address to 192.168.1.11 and connect to CMS1. One can now do cms1> telnet 192.168.1.11. But, this MCH has firmware v2.0 which doesn't work with AMC13 (symptom is both red/green LEDs blink forever on AMC13 with MMC firmware V1.2).

Updating MCH firmware to V2.10 received from NAT on 9/20/11. Briefly, the procedure is:

  unzip the e-mail attachement somewhere
  copy the "bin" file to /tftpboot/mch on cms1.bu.edu
  connect to the MCH using the USB console and type "update_firmware" and follow the prompts
  type "reboot" to load the new firmware

It works! Now the AMC13 seems to be up and running with the MCH.

-- EricHazen - 02 Dec 2011

Topic revision: r34 - 26 Nov 2012 - CharlieHill

~~Edit~~
~~Attach~~