Chuck,
Forgive this long email...
SInce everything is interrelated the only way to deal with it is from a system engineering perspective (in my opinion). You can't design the various subsystems with each group working in a vacuum. If you do, this is what happens.
For example: I've researched the maximum amount of bending that should be allowed on a SMT circuit board. AVX, the capacitor manufacturer, suggests a maximum bend radius spec for SMT circuit boards of 60 inches. That works out to 0.0084" maximum deflection over any 1" segment. That's about the thickness of three sheets of paper. The tolerances of the existing sheet metal enclosure with separate heat sinks and multiple swaged-on stand-offs is way too loose, by at least an order of magnitude. The enclosure I have is also warped and flexes. To me that means we need a milled enclosure...
If we're going to do that we might as well do it right and make it into two separate cells with noisy digital circuits in the front and analog in the back...
If we do that then you probably don't have to worry too much about radiated emissions or changing the PCB form factor or connector...
Filtering conducted EMI would rise to the top of the list of concerns. Moving the switching frequecies of all the supplies up as high as possible would ease the filtering burden on everything on the satellite and tend to push any spurs out of the passband. And so forth and so on...
Once we had some hard data on the amount and characteristics of the conducted EMI from the power distribution point, then John could start designing in the necessary EMI filtering and CPB layout to fit into the box and meet the yet to be determined EMI susceptability requirement for the receiver. If his design didn't look like it would be able to meet the requiement then there would be some push-back to the power distrubution subsystem to clean up their radiated EMI, etc.
I guess the bottom line would be that since you can't know the radiated and conducted EMI susceptability of everything that may end up connected to the CAN-Do module, all you can do is try to make it as clean as you can and get the switching frequency as high as possible. By the way, I just looked up the Maxim converter you're using and it looks like it's very lightly loaded which would explain wny its running at 5 kHz instead of the 200 kHz they spec. I spent some time looking for a better choice and couldn't find anything, but I don't know what you requirements are.
73,
Juan
On 7/17/07, Chuck Green greencl@mindspring.com wrote:
Hi Juan,
I agree that specifications prior to design would be helpful (required?). But as far as the EMI issues are concerned, we do seem to have a chicken/egg problem. And I'm not sure it is practical to design a widget that meets the requirements of receiver modules. Most modules simply don't need anything nearly this good. But we should do everything practical to accommodate receiver modules, and maybe meet their requirements completely.
Thanks, Chuck
Juan Rivera wrote:
Chuck,
My comments might have been buried in the flurry of email a while back. Things seem a bit quieter now so here they are:
- Don't get too far into a redesign until a top-level EMI requirement
is
created. This can't be done properly until prototype solar panel power converters are fabricated and tested. I would work to create a new
power
supply with a switching frequency of at least 500 kHz while you wait however. This will make filtering much easier, the filter components
will
be smaller, and any spurs that make it into the receiver will be outside
the
passband of the IF.
- The EMI requirements for radiated and conducted emissions and
susceptibility should flow out of that test data and not be guesses.
- Once we have an EMI requirement then tradeoffs need to be considered
between the CAN-Do module and the enclosure - one or two
compartment? Sheet
metal or milled construction? The results of that tradeoff study will determine how much room you have to work with, how much front panel
space,
and how much shielding and filtering are required.
If I had my way the enclosure would be a two-cell milled enclosure with
all
the RF and IF exiting out the side of the rear cell. The CAN-Do module
and
the Receiver switching power supply would both be located in the first
cell
with feed-thru filtering in the common bulkhead between cells (all
digital
power in the front and all analog in the rear). That would mean that
the
CAN-Do connector would be the only connector on the front of the
case. If
that were true then the only reason to change connectors would be to
save
weight or increase reliability. It would also mean that the existing
CAN-Do
PCB footprint would be fine as it is.
In my presentation I will suggest that the next revision of the 70 cm receiver should be postponed until all of these issues are resolved.
73,
Juan
-----Original Message----- From: eagle-bounces@amsat.org [mailto:eagle-bounces@amsat.org] On Behalf
Of
Chuck Green Sent: Monday, July 16, 2007 7:47 PM To: Louis McFadin Cc: Dave Black ((Work)); Dave Black ((Home)); David Smith; AMSAT Eagle; Samsonoff@Mac. Com; Juan.Rivera ((Work)) Subject: [eagle] Re: CAN-Do-Too! ??????????
Thanks, Lou.
I don't know of any reason not to use them either. Obviously, it needs to meet our mounting configuration requirement so the new mechanical design can meet the objectives I stated earlier. This means it must mount on the edge of a PCB. I think the HD15-D has three rows of pins so I'm not sure how this can work, but I haven't looked at the various parts available so maybe this problem has been solved.
If we are going to seriously consider using HD connectors I think we need the blessing of AMSAT's VP of Engineering and the EAGLE project coordinator. This would be true for any change that would be pervasive in the satellite.
I am a little disappointed that there have not been any comments regarding the changes I saw as being made with a new design. Additions/changes/questions/etc. I don't think we should do a new design without this discussion. Maybe people feel these issues have been well covered in the past. If so, a simple "looks good to me" would be helpful.
And no one has stepped up to say they are well qualified and will design a new power supply. Without this, I don't see a new design happening, but maybe.
And finally, I see that no one has dared touch the subject of parts procurement I raised.
Obviously, most of these comments are really meant for the Cc list.
Thanks, Chuck
Louis McFadin wrote:
Chuck, Mouser has a very large selection of D-sub connectors including the high density versions. Most are in stock. I see no inherent reason for not using them.
Lou McFadin W5DID w5did@mac.com mailto:w5did@mac.com
On Jul 16, 2007, at 7:19 PM, Chuck Green wrote:
I have had one experience with the high density D connectors. They were much larger pin count than 9 or 15! After someone absolutely insisted that we use them I did the board lay out. Turned out that they were *totally* unavailable!!! I did the board layout again@#$%&^* using standard Sub-D's. That was a number of years ago so I would hope things have changed. If someone is absolutely confident they can obtain the parts we need then I'm not at all opposed to using them (remember, I'm not volunteering to do parts procurement for this project; this is a good time to use someone that's good at parts procurement).
While at Goddard for P3D vib test I noticed NASA satellites using standard Sub-D's. That was also a few years ago. Anyone know of High Density Sub-D's being used on other satellites?
Chuck
Bdale Garbee wrote:
On Tue, 2007-07-10 at 09:02 -0700, Chuck Green wrote:
The sub-miniature D connector series has served us well. If anyone has *experience* with something they think might be a better choice, we'd love to hear about it.
At the AMSAT annual meeting that was held near Washington, D.C., a couple of years ago (three?), someone approached me after the CAN-Do! talk that Stephen and I gave to ask why we weren't using the higher-density connectors that put 15 pins in the same shell size as
the
9-pin version of the series we have been using... and followed up by sending me what looked like mil/aero-spec samples of such a part that
I
probably still have in my basement somewhere. I'm sorry that I can't recall at all who that person was, but it was someone who claimed to
be
using such connectors professionally with good results.
At the time, we weren't likely to be redesigning the units any time soon, so I didn't take any action on this suggestion. If we're going
to
revisit the design and think we need more than 9 pins, it might be
worth
investigating higher density connectors like that?
Bdale
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle