/ - Board Testing Home - // - Overview - // - Tests Without the Test Board - // - Tests with the Test Board - // - Pre-Burn Assembly - // - New Tests After Burn-In - // - Final Assembly - /
The following tests all require that the right firmware is loaded onto the mezz card and slow control. If you're starting a set of tests from the middle of the sequence, ensure that the mezz card has the right firmware on it.
The single cable test is the most-failed test, which is why it's done first. The purpose of the test is to see if there are any problems in acquiring data from the 40-pin input connectors. This is done by having the Virtex chip eject data on the second SCSI connector and looping that back over a cable into the board connector, from whence it is again acquired by the mezzanine card. Each connector is tested separately. Hence single cable test.
Common failures on this test are a short-to-ground on one or more pins or a failure in the delay chip to relay data. Unfortunately, it's difficult to tell when there's actually a problem with the board and when there's only a problem with the testing cable. The cables are nearly all flaky, but most of the time it's only one or two wires that are making intermittent contact. Try holding down one or the other edge of the cable near the plug when it's in the connector. Since each test is repeated 10x, you can tell a flaky cable is the problem when a channel will pass one round of tests fine but will fail a few tests on the next round, then pass it again on the next. You'll be able to see this clearly with experience.
It helps to understand the Single Cable testing program in ALCT_Test. Very simply, the program tells the Virtex chip what data to send over the cable. It shows this in the Send Data: field, and the return values are shown in the Read Data: field. The board takes 16 inputs so 4 values in hexadecimal are sufficient to represent all possible values that might be sent to the board. Google on "explain hexadecimal" if you're not clear on how hexadecimal works. When the cable's flaky, it will generally be the rightmost bit in the read value that's 0, or, in other words, the value in the Read Data field will be 1 less than the Sent Data field. When the problem is on the board, a bit will always read 0, usually somewhere else in the bit string.
Note that the cable doesn't have to be completely inserted into the connector for all of the pins to make contact. If you learn how to do this, it will save you some time, much effort, and the cable will last longer. The primary stress on these cables, and the reason they break, is that they have to be pulled out by pulling on the cable itself, and not the connector, and generally it's always the same side of the cable that gets yanked first, hence, that's the side that's more likely to break.
For the remainder of the tests, the mezz card needs a different firmware. Disable the clock and program the firmware for the remaining tests (don't forget to power cycle the board when it's done to load the new firmware.) Open a connection to the board and enable the clock.
This test injects test patterns into the delay chips and reads the results on the mezz card. In conjunction with the previous test it can be used to determine whether there is an error in the path between the connectors and the delay chips or further on. It also tests the function of the serial bus between the mezz card and the delay chips.
The delay chips come in groups of 6:
This test checks that the LEMO connectors at the bottom of the board work and that the board is able to set a bit mask to turn them on/off. This is done by setting a Test Pulse Strip Layer Mask (the bit mask) to values that select no channels and then each channel individually. The bit mask is entered in hex, and this is what it looks like:
| Hexadecimal (0x__) | Bit Mask | LEMO Connector Selected |
| 00 | 000000 | No channels selected |
| 01 | 000001 | Channel 1 selected |
| 02 | 000010 | Channel 2 selected |
| 04 | 000100 | Channel 3 selected |
| 08 | 001000 | Channel 4 selected |
| 10 | 010000 | Channel 5 selected |
| 20 | 100000 | Channel 6 selected |
On the board, the channels are numbered 6-1, left to right. Their numbers are also printed on the board.
A good habit to get into is to enter both digits for the mask "00", since you will have to do it for the rest of the masks anyway.
If the mask is set to "00" and all of the channels still work, you very probably don't have the slow control programmed. Instructions for this can be found on the Overview page.
I'm not really sure what this test is about. Go ask Martin.
This test checks the voltage potential between GND and the reference voltage, which is found on the lower 4th pin from the left of every chip in the row of chips found next to the LEMO connectors. It is critical that the results of this test are between 1.224-1.226V.
/ - Board Testing Home - // - Overview - // - Tests Without the Test Board - // - Tests with the Test Board - // - Pre-Burn Assembly - // - New Tests After Burn-In - // - Final Assembly - /
Matt Matolcsi (madhat@ucla.edu); Last revision: 2003/07/15