v0.1 physicst shopping list

[jbqubit]

The hardware development contract funds two rounds of prototype hardware. Each prototype round will generate two copies of the prototype hardware consisting of the following parts. The list reflects the new organization of the Sayma analog front end with 4 mezzanine boards, each supporting 2 DAC channels and 2 ADC channels.

Metlino-Sayma Prototype Board Set
1 metlino_motherboard
1 metlino_clk
2 sayma_motherboard
2 sayma_clk
8 sayma_analog_mezzanine

Full Metlino-Sayma prototype hardware will be distributed as follows.

• m-labs in Hong Kong
• UMD
• NIST (A)
• Oxford (A)

(A) The additional cost of two additional development systems will be paid directly (eg credit card) to the firm responsible for printing-stuffing the contract-funded prototype hardware (eg Creotech).

[jordens]

@jbqubit It's unclear what the "issue" is and who is supposed to act and what should be done. If there is no issue, close this.

[jbqubit]

Tasks:

• @gkasprow determine cost of full Melino-Sayma prototype system
• when the time comes, Oxford @weidazh @cjbe and NIST need to arrange to pay for their copies of the est tsystem
• @gkasprow identify additional part numbers needed for a full test system
  • microTCA crate with microTCA.4 backplane
  • AMC-side power supply
  • MCH Tongue 0
  • RTM RF backplane
  • single-slot AMC adapter for stand-alone testing
  • passive RTM and AMC blanks (for power dissipation and RF shielding tests)

[dhslichter]

Also should specify the types of mezzanine cards to be included, since there will be at least 2 designs (direct input/output and up/downconverting out/input), if not more.

[jmizrahi]

@gkasprow could you provide a list of all the commercial components necessary to run the full system? I would like to go ahead and order all the necessary microTCA hardware.

[gkasprow]

• MTCA crate: 11850-026
  http://schroff.pentair.com/wcsstore/AuroraStorefrontAssetStore/User%20Downloads/LiteratureRequests/Content_schroff_ssh_39601850_microtca.4_en.pdf
  So far the RF backplne is currently available only in this crate.
• Smaller version of RF backplane will be available soon (according to
  my colleagues from WUT) for this crate
  http://schroff.pentair.com/en/schroff/MTCA/3074457345616797165.
• Supply: 600W 11098-547 or 1000W NAT-PM-AC1000
• There is version of the PSU without the power management:
  http://schroff.pentair.com/en/schroff/11098-416, but I never tested it.
• Filler module for empty AMC/RTM slots 21850-106
  http://schroff.pentair.com/en/na/21850-106
• Filler module for empty MCH/PSU slots 21850-107 http://sch but
  roff.pentair.com/en/schroff/21850-107
• MCH: NAT-MCH -Base12 -CLK-PHYS -FP0D

[jmizrahi]

@gkasprow Thanks! I assume the 1000W would be necessary for a fully populated full size crate?

[jbqubit]

Notes from discussion on 10/21 follow.

When operating Sayma ARM+RTM on benchtop use the following.

• communication with Core Device (MCH)
  • AMC backplane Port 4 uses transceivers to communicate with Core Device
  • a SATA connector on the AMC backplane is also connected to Port 4 (passive)
  • in stand-alone configuration use a to SATA to SFP adapter to connect ARTIQ Master with board
  • no change in gateware is needed
• power with +12V ATX power supply (common in PCs)
  • @gkasprow is this the 4-pin connector in the upper right-hand corner? link

Chassis:

• @gkasprow RTM RF backplane part number for 6-slot chassis
• @gkasprow RTM RF backplane part number for 12-slot chassis

[gkasprow]

When operating Sayma ARM+RTM on benchtop use the following.

• communication with Core Device (MCH)
  • AMC backplane Port 4 uses transceivers to communicate with Core
    Device
  • a SATA connector on the AMC backplane is also connected to Port 4
    (passive)
  • in stand-alone configuration use a to SATA to SFP adapter to
    connect ARTIQ Master with board
  • no change in gateware is needed
• power with +12V ATX power supply (common in PCs)
  • @gkasprow https://github.com/gkasprow is this the 4-pin
    connector in the upper right-hand corner? link
    http://www.ohwr.org/projects/afck/wiki

Yes

Chassis:

• @gkasprow https://github.com/gkasprow RTM RF backplane part number
  for 6-slot chassis

don't know the pn yet

• @gkasprow https://github.com/gkasprow RTM RF backplane part number
  for 12-slot chassis

NATIVE-LLRFR9BP
Low-Level-RF-Backplane (Add-On for NATIVE-R9): DESY design providing Zone 1
RF connections (between 12uRTMs and 3 eRTMs) as well as Zone 2 power
connections (between 12uRTMs and 2 rear PMs) and Zone 2 LVDS connections
(between 12uRTMs and NAT-MCH-RTM-FPGA).

Delivered as integrated part of NATIVE-R9.

List price is 1600 Euro.

[gkasprow]

Let me finish the power estimation.

For 12 slots this may be necessary

[jbqubit]

http://mtca.desy.de/sites/site_mtca/content/e172206/e205636/e212584/e248086/uRFB_concept_Datasheet_19.12.2014_eng.pdf

[jmizrahi]

I received this quote from NAT for the microTCA chassis, RF backplane, power supply, and MCH. Is this the same set of hardware that everyone else intends to use with these boards? Do the rest of you already have a microTCA chassis with the necessary MCH and RF backplane? They quoted a 6-8 week leadtime. @jbqubit @jordens @dhslichter @hartytp

@gkasprow they added NAT-MCH-M4 as an alternative, but this is unnecessary, correct?
Can you confirm that what they quoted me is the correct hardware set and the minimal hardware set (as the RF backplane does not appear to be available for smaller chassis's?) Given the price tag, I'd like to be 100% sure that I don't accidentally purchase incompatible hardware.

[hartytp]

@jbqubit Our process for approving new suppliers is a bit torturous, so I'd like to start adding the supplier (Creotech?) for the Sayma/Metliono hardware now. Who is the contact for this?

[gkasprow]

The production of prototype series will be done by Technosystem (http://technosystem.pl/#)
At the moment it cannot be at Creotech due to formal restriction at the WUT (conflict of interests).

[hartytp]

@gkasprow When we order hardware, will we place the order with Technosystem or WUT?

[gkasprow]

You can do it via WUT but then 30% charge will be applied. So it depends on you:) WUT will anyway subcontract HW production to Technosystem

[hartytp]

Okay, noted. Is there anyone in particular I should contact at Technosystem about this?

[hartytp]

@jbqubit @gkasprow

• What is the estimated cost of each of the pcbs we're designing? (PCB_metlino, PCB_sayma_AMC, PCB_sayma_RTM, PCB_mezzanine_analog_allaki, PCB_mezzanine_clock_template, PCB_baikal)
• Assuming that there's nothing major wrong with any of the PCBs we design (say, just a few minor noise issues) when do you estimate that a second design revision and/or hardware purchasing round will happen?

These factors will influence how much hardware we buy in this round.

[gkasprow]

@jbqubit, @hartytp

I've just got initial offers from the assembly house:
Sayma AMC for 2 pieces: 4k$/piece. Metlino price would be very similar.

Sayma AMC for 10 pieces: 3.3k$/piece

Sayma RTM for 2 pieces: 3k $/piece

Sayma RTM for 10 pieces: 2.36k$/piece

These are just components, PCB and assembly prices, there will be some man work added to run the production test.

[jbqubit]

@gkasprow here's my attempt at distilling this down to a concise list. I'm unclear on exactly which filler part to use for empty slots (mid-size/full-size). Are the fillers for the RTM and AMC side the same? What about filling unused RTM side slots?

Description	Vendor	Part Number	Quantity	Comment
crate	NAT	NATIVE-R9	1
RF backplane for R9	NAT	NATIVE-LLRFR9BP	1
MCH Tongue 1	NAT	NAT-MCH-Base12-GBE-FP0D	1
MCH Tongue 2 adapter	?	?	1	to be designed by WUT
PSU module 1 kW	NAT	NAT-PM-AC1000	1	power option 1
PSU module 1 kW	Schroff	11098-416	1	power option 2 Greg hasn't tested this
filler module for empty AMC slots	Schroff	?	10
filler module for empty RTM slots	Schroff	?	10
filler module for empty PSU slot	Schroff	?	1
filler module for empty MCH slot	Schroff	?	1

[hartytp]

Minimal parts list required to get the Metlino/Sayma system working in a uTCA.4 rack, all from NAT:

NAT PN	description	comment
NATIVE-R9	Rack	This is the only rack which currently supports a RF backplane. RF BPs are planned for more racks in future (work not funded/started yet).
NAT-PM-AC1000	1kW PSU	Cheaper, lower power-rating PSUs are also available, with similar specs. However, the price difference isn't that great.
NATIVE-LLRF	RF BP	Needed to power RTMs and optionally used to distribute clocks.
NAT-MCH-Base12-GbE	MCH	It is mechanical support for Metlino, it also delivers power. It regulates temperature inside the crate, enables power supplies and manages power distribution. Some day we will make own, simplified version.
NAT-MCH-FP0D	Double width front-panel for NAT-MCH-Base12-GbE
NAMC-Filler FS-FP1D	Double-width AMC filler + front-panel	Needed for proper air flow in the rack. Empty crate needs 6 of them (4 PM + 2 MCH) With one PM and MCH we need 4
NAMC-Filler FS-RP1D	Double-width RTM filler + front-panel (for unused MCH + PMs)	5 in our case assuming that we we only eRTM15 and no power module
NAMC-Filler MS-FP1D	Single-width AMC filler + front-panel	12 needed for an empty rack
NAMC-Filler MS-RP1D	Single-width RTM filler + front-panel	12 needed for empty rack

12 NAMC-MS-FP1D come for free with the rack.

This assumes that @gkasprow will make a simple tongue 2 board to replace the NAT-MCH-CLK-PHYS part (which we would otherwise use as a ~1k USD spacer).

I already did it.

NB on @jmizrahi quote above:

• NAMC-Splitting-Kit is mounting hardware for NAT MCH, which is not compatible with Metlino. Not required.
• NAT-MCH-M4 not needed/wanted in this system

@gkasprow please can you confirm that the above is correct?

[hartytp]

@jbqubit you beat me to it!

[hartytp]

@gkasprow Thank you for doing that!

[gkasprow]

@jmizrahi
Confirmed that:

• NAMC-Splitting-Kit is mounting hardware for NAT MCH, which is not compatible with Metlino. Not required.
• NAT-MCH-M4 not needed/wanted in this system

[jbqubit]

Some refinements of @hartytp list.

• WUT provides a front panel for the MCH slot which consists of a stack of the following boards
  1. ARTIQ_EE/PCB_Metlino (MCH Tongues 3,4)
  2. ARTIQ_ALTIUM/PCB_MCH_mordovia (MCH Tongue 2)
  3. NAT-MCH-Base12-GbE (MCH Tongue 1)
  • ->> so scratch NAT-MCH-FP0D
• PCB_MCH_mordovia is the name of the simple tongue 2 board that replaces NAT-MCH-CLK-PHYS part

[gkasprow]

There is an issue with front panels. They are manufactured (milling, printing) by Elma who has 8..12 weeks lead time. Schroff has similar lead time. Last time I waited even more because they had some material shortages. So there is very little chance that they will be delivered with first revision of the PCBs. I couldn’t design and order them before layout was finished. Just a few days ago additional connectors were added to Sayma.

[jbqubit]

Lack of front panel isn't a problem for v0.1 testing. Please ship PCB's with appropriately sized standoffs and screws to hold them together.

[sbouhabib]

@gkasprow i think there was a company that did them in 4 weeks, they are not nice ones with emblems and stuff just the metal, Ill try to check tomorrow

[dhslichter]

What about Front Panel Express/Schaeffer AG? They'll turn custom panels around in a week, with holes, legend, etc.

http://www.schaeffer-ag.de/en/

[gkasprow]

This is not standard front panel cut in a piece of aluminium sheet. It is formed bezel in which holes and slots are machined.

…

[jmizrahi]

@gkasprow if blank front panels are a stock part, and the long lead time is for machining and printing, why couldn't we buy the blank ones and send them elsewhere for the machining and printing?

[jbqubit]

@jasonamini pointed out that there are some additional parts users should buy to prepare for testing/using a Sinara system with DRTIO (SFP fiber-based). I've tested these with ARTIQ DRTIOlite (circa summer 2016).

• managed Ethernet switch with 8 RJ-45 copper ports and 2 SFP ports: Netgear ProSafe GS110TP
• 1000BASE-BX-D SFP 1490nm-TX/1310nm-RX 10km DOM Transceiver: ZyXEL SFP-BX1490-10-D; ordered from fs.com #39460, qty 6, $12/ea (*)
• 1000BASE-BX-U SFP 1310nm-TX/1490nm-RX 10km DOM Transceiver: ZyXEL ZyXEL SFP-BX1310-10-D; ordered from fs.com #39464, qty 6, $9/ea (*)
• 9/125 OS2 Duplex Single-Mode Fiber Optic Patch Cable - PVC/LSZH/OFNP; ordered from fs.com

    Fiber Type - G.652D     
    Cable Jacket - PVC(OFNR)     
    Cable Diameter - 0.9mm   
    Polish A - UPC    
    Connector A - LC    
    Polish B - UPC    
    Connector B - LC    
    2m    
    #12189 qty 6, $3.25/ea

(*) Unfortunately, fs.com no longer stocks the ZyXEL parts and it looks like they're in short supply from other distributors. DRTIO relies on use of bi-chromatic SFP transceivers with the same constraints as White Rabbit. See ohwr.org for a list of White Rabbit compatible SFP modules. @sbourdeauducq What transceivers are you using for DRTIO testing?

@jasonamini, @jmizrahi I have plenty of ZyXEL transceivers and patch cables so don't ordey any for UMD. Do order your own NetGear switch.

[sbourdeauducq]

I'm currently using a pair of generic 1000BASE-BX SFPs from fs.com (1490nm/1310nm) and they do what they are supposed to do, but they are limited to 1G. We plan to have a higher data rate for the phaser-based designs, so I have also tried similar 10G SFPs but they are acting up for yet unknown reasons. Our next step is to get the DRTIO phaser to work with a passive SFP copper cable (http://www.fs.com/products/36648.html) between two KC705s, then figure out what the heck is wrong with the 10G SFPs.

[jbqubit]

Thank you for the update @sbourdeauducq. I got an email from fs.com indicating what generic SFP is supposed to be compatible with the ZyXEL SFP.

#37930	Customized 1000BASE-BX BiDi SFP 1310nm-TX/1490nm-RX 10km Transceiver
Compatible Brands:ZyXEL	US$8.82	1294 pcs in stock

#37940	Customized 1000BASE-BX BiDi SFP 1490nm-TX/1310nm-RX 10km Transceiver
Compatible Brands:ZyXEL	US$11.76	814 pcs in stock

[sbourdeauducq]

AFAIK this "compatibility" is mostly EEPROM programming, which doesn't matter in our case since ARTIQ does not check that a SFP of an approved brand is connected and completely ignores that EEPROM.

[jbqubit]

I'm closing this Issue as this looks resolved.