Below is a list of the questions, scores, comments and my conclusions.
==============================================================================
Question 1: Should we include a module size suggestion of 4"/4000mil/101.6mm long? (that allows space for the full 40/80 pin headers)
Voting score: 15
Some of the comments:
The initial specification should not contain recommendations on the size and shape of the modules except the size of the connector. The recommended size and shape can be amended later. As long as the pin layout is specified, the modules will work in compatible backplanes, although they may not fit in certain configurations or builds.
Yes but note the RC2014 size and why
Weak yes... I'd rather keep module to 100 mm x 100 mm (or even 100 mm x 80 mm... is anyone still using Eagle?)
Conclusion:
We should include this as a suggestion module length.
==============================================================================
Question 2: Should we include a module size suggestion of 3.9"/3900mil/99mm long? (for those who want consistency with RC2014 39-pin long modules and for a wider choice of low cost PCB manufacturers)
no, link to
rc2014.co.uk on drawing for 1, with outline of spencers module shown as overlay.
Link to or appendix the RC2014 ones as commonly used profiles.
I mean - for interoperability it doesn't matter if your card is the shape of a dinosaur providing it has the connector and flat in the right place.
Yes, add a reference to Spencer's specs
Conclusions:
Include link to
rc2014.co.uk==============================================================================
Question 3: Should we include a module size suggestion similiar to the RC2014 Classic 2 low profile modules? (4"/4000mil/101.6mm long by 1"/1000mil/25.4mm high)
Voting score: 12
Some of the comments:
yes, X dimension on drawing for 1, with table of recommended X.
No, I never cared for low profile modules... Always looked like a waste of backplane space ;)
Conclusion:
We should suggest Classic 2 profile as one possible size of module.
==============================================================================
Question 4: Should we include a larger module size, perhaps 4"/4000mil/101.6mm square? (particular well suited to prototyping but also allows greater functionality or nicer component spacing and labelling)
Voting score: 16
Some of the comments:
Given there is no standard established shape what is wrong with modules may be taller or longer than the suggested profiles but this may cause problems with cases.
Yes, although, as specified before, 100 mm x 80 mm or 100 mm x 100 mm would be my preference.
Conclusion:
We should suggest a large module size.
==============================================================================
Question 5: Should we recommend the PCB extends below the header mounting holes by 0.15"/150mil/3.81mm as proposed by Tadeusz yesterday?
Voting score: 10
Some of the comments:
maybe, but no extension past plastic support of dual row header
Conclusion:
Yes, we should recommend the PCB extends to the bottom edge of the plastic of the header.
==============================================================================
Question 6: Should we recommend the same size 45 degree bevel and the same curve radius as Spencer's RC2014 modules, or should we recommend different sizes or shapes to differentiate new RCBus modules from RC2014 modules?
Some of the comments:
Slightly different dimensions
Not a fan of the curve and 45 degree bevel at all along with the bullet hole. Hinders part placement. Too much like RC2014. I'm going to come out with a FPGA board it will grow vertically some what.
Don't care but having something is helpful
Weak no. 45 bevel is cute, but it eats up PCB space, and doesn't add to functionality. If decided on a rectangular shape, perhaps a couple of M3 holes on each one of the corners would be nice (so standoffs can be used to build a more rigid construction, say 5 mm (or 5.08 mm if we want to be imperial system compliant) from the top and left/right edges.
Voting score: -10
Conclusion:
A slightly different module shape should be recommended.
==============================================================================
Question 7: Should we recommend header pins be centred along the length of the board, rather than be offset like Spencer's RC2014 standard size modules?
Voting score: 5
Some of the comments:
pin 1 to left edge should match spencers modules, centred on wider module.
Don't see that it matters either way
No. Instead, pin 1 center should be placed 1.27 mm (50 mil for fellow US people) to the right from the left edge of the board. In this case, the connector will be "centered" for 101.6 mm / 4 in board.
Conclusion:
The header should be centred for modules that are 4"/101.6mm long.
==============================================================================
Question 8: Anything else we need to decide on this issue?
Voting score: -12
Conclusion:
Looks like we should now be able to pin down the recommendations for module sizes.
==============================================================================
Question 11: Should the backplane include a dedicated 3V3 pin?
The alternative is to generate 3V3 on any module that needs it. Having a pin on the bus for this will not mean the supply is always present, so will that mean a module needing 3V3 will require a suitable power module to be present, or does it mean those modules needing 3V3 will have a regulator on them anyway plus a jumper to determine if its 3V3 supply is connected to the bus?
I've faced the same issue with my designs that have a voltage supervisor. I initially included the supervisor on processor modules but later introduced power supply modules that include a supervisor. So now some combination of boards mean I have two supervisors!
I also think 3.3V is nice, but the Paralax Propeller and WiFi modules did just fine without it. The BAI/BAO cascade is also a bit of a leap into the future, especially as we still don't have established request signals to invoke even one rather than several DMA chips. However, if we want to use such a cascade, we need to provide for it on the backplane, otherwise we will have unnecessary wire connections.
Voting score: 2
Some of the comments:
No. Current backplanes are not prepared for this, the number of modules requiring this voltage in complete kits is small, additional voltages should be provided by the module which requires them (an implicit example is RS232 with MAX232 which generates +10V and -10V), the only safe place to provide sufficiently wide tracks is 40-80.
Would be nice but where is a good question.
Weak yes, I like the proposal of using pins 40/80 for it
Conclusion:
No clear consensus. Best leave it unspecified for now.
==============================================================================
Question 12: Should BAI and BAO be on pins 42 and 43?
Mark T suggested it would be better to put them near IEI and IEO (pins 38 and 39, previously known as USER 6 and 7). Mark's reasons were to keep the new pins free for other processors and proximity to the back of the PCB is more convenient for link wires etc. As the likely implementation will involve jumpers on the backplane to select daisy chain or direct connections there might be a space issue near IEI and IEO. Also, USER 6 and 7 are more likely to be used by existing designs than the new pins.
Voting score: 1
Some of the comments:
Yes. The proposed circuit allows the cascade to be connected using jumpers, no wire connections are needed to implement it.
Placing the BAI/BAO cascade on pins 77 and 78 does not seem to be a good idea. If you don't want to use vias, then put the two 2x3 pin headers side by side. Maintaining 0.6" spacing between the slots will be impossible. I consider the return to wiring the module to the backplane to be too much of a step backwards. The drawing also included the 3.3V line - the only sensible place for the backplane.
I think not impossible, just a little bit awkward. Difficult to change links without removing modules. The headers for the links could be put on the bottom side of the backplane but then soldering the backplane connectors and header pins for the links is a little tricky.
I was considering a similar arrangement as links on the middle pair of connectors can allow a daisy chain using any pair of user pins, but doesn’t work with the proposed standard for IEI/IEO on n38 and N39.
I think n40 and n80 as a daisy chain would also work for BAI/BAO. Then only extended width cards could use BAI/BAO, but that doesn’t seem to me to be a disadvantage.
Weak no. Without looking much at the layout, I'd put them near IEI and IEO pins.
Conclusion:
No clear consensus.
We need an answer as the backplane will need a mechanism to support this, or we will be left with flying leads.
A little debate needed I think.
==============================================================================
Question 13: Should we call the fixed function pins "common pins/signals"?
These are the pins that have a single clearly defined function, so mainly the standard RC2014 bus signals.
Other options are "Base pins/signals" or "Fixed function pins/signals" or "Core pins/signals" or "Standard pins/signals" or ???.
Voting score: 5
Some of the comments:
I think to discuss signals collectively would be Standard, Enhanced and Extended. With Extended qualified by processor group. Also with Enhanced and Extended not being inclusive of the subset.
Conclusion:
Not entirely clear.
I think perhaps collective names for these groups should be "Standard" (for RCBus 40-pin) and "Enhanced" (for RCBus 80-pin)
==============================================================================
Question 14: Should we call the multi-function (ie. processor specific or use case dependent) pins/signals "Extended"?
These are the pins currently named n## and include the RC2014 USER pins. I 've found that they need a simple name when trying to document them.
Other options are "Multi-function" or ???
Voting score: 6
Some of the comments:
I think extended is a good term
Yes, call them CPU-dependent? Extended can mean many things
Conclusion:
Not entirely clear.
I think perhaps a collective name for this group should be "Extended"
==============================================================================
Question 15: Should the multi-function pins be identified by labels of the form n## (eg. n41)?
If not what should they be labelled?
Voting score: 5
Some of the comments:
Maybe the solution would be to call them by the names of the RCZ*80 bus with a # or * sign?
I think usefull to have a common method of describing the physical connection, not sure about n41 etc as n prefix is common for active low. Is pin41 or p41 not easy to use?
Convention seems to be to give them a base name and where needed refer to them with a / list of their forms - at least that's what half the microcontrollers seem to do.
Weak yes... I don't have a better suggestion
Conclusion:
Not entirely clear.
I think they should be referred to as "pin ##" (where there is room) or "p##" where space is tight.
==============================================================================
Question 16: Should we have a logo of some sort for RCBus?
The alternative is just to use "RCBus".
Voting score: 6
Some of the comments:
yes, with the text as a fallback
I have a dilemma here, there should be a recognisable designation, clearly indicating the supported standard, but there may be people who see this as an attempt to build a new brand. I will certainly leave the 1980s style lettering/logo in my backplane, will I use it in the modules? I do not know.
do you intend to copyright or control how the logo can be used?
Does it matter - if someone makes one then fine if not then fine 8)
Yes. Do we have graphic designers? :)
Conclusion:
Not unanimous but it looks like we want a logo.
I for one have no interest in copyright or control of the logo.
==============================================================================
Question 17: Should we recommend a minimum spacing between sockets of 0.6"/600mil/15.24mm?
This is the spacing I use for SC112, SC113, SC116 and on my motherboard designs.
The RC2014 backplanes are 0.65"/650mil/16.51mm.
Voting score: 9
Some of the comments:
Yes. A minimum spacing of 0.6" is sufficient.
I prefer 16.51. 15.24 is too tight with socketed can oscillators.
Weak no for 0.6". It seems to tight in some cases - socketed oscillators, vertical connectors, D-sub, Mini-DIN, RJ45 modular connectors
Conclusion:
On balance it looks like 0.6" is favoured.
Perhaps we should just suggest 0.6" and 0.8" are likely to be commonly used spacings.
==============================================================================
Question 18: Should we suggest a wide spacing of 0.8"/800mil/20.32mm as an alternative?
This give more wiggle room but it is not enough for really large components. The alternative is to simply say there will be odd cases where you need to fit the module in the last slot or have it cover a second (unusable) slot.
Voting score: -1
Some of the comments:
No. I think that wide modules can occupy two slots, something like in 19" racks - 2U
no, 17 is just a recommendation
Does it matter. Specify the minimum spacing and "wider spacings may be used and may be more convenient"
Yes for 0.8" alternative... Maybe someone would design a backplane with both 0.6" and 0.8", e.g., 3 slots with 0.6" spacing and two with 0.8" spacing? 0.6" is really narrow for almost any kind of vertical connectors
Conclusion:
No consensus.
See conclusion to question 17.
==============================================================================
Question 19: Should the labels for the bus signals include a symbol/character to indicate active low?
Voting score: 14
Some of the comments:
Yes. This is a clear signal indication
yes, pick something that works well on silkscreen, I think lower case n rather than /, with upper case for the rest of the label.
Conclusion:
Polarity should be included in the signal name.
I think the suggestion of lower case n as a prefix for active low signals followed by all upper case signal name, should be suggested as the naming convention.
==============================================================================
Question 20: Is there anything fundamentally wrong with the attached render of a backplane?
Voting score: -7
Some of the comments:
Kind of... I'd use top side for the ground fill and the bottom for the signals (or the opposite way)... No real need for that wide traces, especially given that we operate with CMOS logic. At the same time, having a ground plane will reduce EMI
Conclusion:
Close to unanimous as many gave no opinion. So no major objections to the backplane render.
==============================================================================
Question 21: Anything else we need to decide on this issue?
Voting score: -7
Some of the comments:
All this discussion makes me think why wouldn't we do ECB instead :) Good quality, although a tad pricey, DIN 41612 connectors. Better pinout than RC2014... And if you fancy not buffering your bus - just don't do that :). Andrew Lynch (N8VEM), one of the people that IMHO restarted the Z80 interest among hobbyists 15 or so years ago, is now working on 100 mm x 100 mm version of ECB. A bit more modular than the original approach. Still buffered bus. Still ECB pinout / DIN 41612 connectors.
Conclusion:
No major issues raised.