Subject: Re: Questions about Data Pump From: Denis Calvet Date: Wed, 16 Apr 2003 09:50:18 +0200 To: Hal Evans , D0 Run2b L1Cal Trigger Dear All, I probably missed parts of the discussion, but what I understand from the list of questions below is that the data pump is (or will be) a multi-channel programmable arbitrary waveform generator that could be used to drive the analog inputs of one or several ADF cards. There is no doubt that such module would be useful, though to test the trigger tower analog splitter card, I had to develop a small circuit connected to a standard arbitrary waveform generator, programmed to generate pulses that mimic the shape of the pickoff signals. So to test one ADF card, channel per channel, this is probably sufficient. Testing multiple channels at a time should also be performed and this can be an application for the data pump. We have in our lab a rather similar project of general purpose multi-channel arbitrary waveform generator that we intend to use (initially) for a couple of projects, including the test of the ADF card (this is not a top priority though, but a working generator is expected around the end of the year). Another possible use of the data pump would be to pre-load a pattern of energy deposition in several trigger towers, then play it all the way from the analog input of ADF cards until the TAB output. This of course would be an interesting test, but the ADF has already built-in the capability to pre-load 512 digital samples per channel and play them. Using the data pump for a similar task would add the analog section and ADCs in the chain - an important thing, but not a major step compared to the previous test. In addition to such kind of multi-channel waveform generator, I feel that the L1 Calo project would also need a multi-channel digital Channel Link tester. The ADF card has 3 2 Gbit/s digital output links and the TAB card has 30 such inputs. Nevis developped a single channel sender / receiver which will be certainly useful, but I see no hardware to validate in an independant way the sending part of the ADF and exercice a large number of TAB inputs. I think there is a gap here that we have no plan to fill at Saclay, except - if time allows - making a small mezzanine card with a Channel Link receiver (mainly for training/education purposes). Back to the data pump question list, a number of answers are given below. Best Regards, Denis. D0 L1 Cal Data Pump Questions: How has this been done in the past? What can we learn from that? How is this different? Form factor, is 9U required or is 6U possible. Stand alone possible, Power? The ADF crate is VME 64x chassis (+5V +3.3V) 6U, so I guess 6U is better. It could be standalone. Analog Out What type of waveforms must be generated? See the shape of the trigger pickoff signals collected by MSU http://www.pa.msu.edu/hep/d0/ftp/l1/cal_trig/pictures/trig_pickoff/ Time/frequency characteristics of waveforms? Such as duration and number of points. Number of bits of resolution - is 8 OK? The ADC of the ADF board is 10 bit 32 MHz; so the DAC of the generator should have 12 bit I guess and should play samples at 64 MHz or more. See the shape of signals I used to test the splitter card: http://www-clued0.fnal.gov/~perez/R2BTRIGGER/ADF_Description/Splitter/Tests_ of_Splitter_Proto_13dec02.pdf Output dynamic range? If variable, how is it controlled? 6 V differential for full scale of the DAC Number of channels - 1 ADF (24) or more? There are 32 channels per ADF board Can Fast RAM + D2As produce the signals? If 200 MHz works, yes. I guess so. Timescale to load wave shapes. Not critical Number of internal wave shapes I guess we want to play at least 64 Beam Crossings, assuming we generate samples at 64 MHz (i.e. BC x 8), one would need at least 512 points per channel. Frequency characteristics of Output Filter. The ADF analog input has a 2nd order Butterworth filter with cutoff frequency to select from 15 MHz to 7.57 MHz. I guess the generator should have a low pass filter with a cutoff ~30 MHz, maybe even less. Timing in Need bunch crossing clock and start output timing signals. All from SCL system or is there an alternative? No info Include a simple trigger from external input? It is always a good idea. What range of delays is needed? >From trigger x to y ns. Between channels? I guess channels should all start synchronously, and we should have the possibility to synchronously start several generators. Timing out needed to other boards Other Is there a document or can Dan Edmonds describe the BLS output characteristics. Flexibility is proportional to cost. Is this a $4k module or an $8k module? Schedule. When is this need it? How many of these boards are needed? All at the same time? This needs to be discussed, but the ADF board prototype will normally be produced and tested this summer, then connected to the analog splitter cards which are connected to the BLS signals from the detector.