Thin FR4 v. Thick FR6 Backplanes ...

[Walt Perko]

Hi,  

Okay, these discussions on the S100Group are focused on backplanes for both Altair 8800 and much more modern 10MHz + systems, but still there is some good information coming out of the discussions.  

There is no current conclusion ... but the thickness of the boards doesn't seem to be a real concern ... yet.  Hopefully, this discussion does bring up some more concrete answers.  

Quote from John Monahan:  

FWIW the S100 motherboards I made were 2.6 MM thick. My "regular boards are ~1.68MM.

The third motherboard worked "most of the time" but was unreliable with the FPGA Z80 SBC which uses 74VLC245 bus drivers.

What I never understood is the Morrow board (same size) worked fine in that setup.  I came across a second used Morrow board and took the easy way out!

 May circle back some day. But getting these sockets is a pain and expensive.

John

Quote from Jeffery Wilson:  

Hi Walt (et al),

 

Seems there should be some way to test the designs to see if they're working as antennas or something else ... maybe a simple test board set to connect to an oscilloscope or other display testers.

It's not that simple to do as every conductor is an antenna of some sort...it just depends on frequencies, current flow, ground planes/traces, and line impedance.  Thus, an S-100 bus WILL (and MUST) radiate given the right frequencies, current loads (i.e. - Bus Capacitance), and bus design.  

The question is how much?  And that is an equation that changes with every board change (adding or removing line capacitance), every new program (i.e. bus access cycles/frequencies change with the loaded program), and slot changes (i.e. you're changing the feed-point on an antenna strip-line).

Simply put:

• A wire that is only connected at one end to a frequency source is a monopole antenna.
• A wire above a ground plane, but only connected at one end to a frequency source is a dipole antenna.
  (Even wires that are connected at both ends above a ground plane, or where one end is connected to ground are antennas...just in much higher frequencies if the necessary harmonics are present.  WiFi Loop and Folded F antennas work on this principle).
• A wire that is completely surrounded by a ground plane is a coax, and will radiate from the disconnected end.

So, the final answer is THEY ARE ALL ANTENNAS.  The difference in the radiated output power of these antennas is down to the AC current flow in the line (which is a function of the frequency of the instantaneous current change, NOT the DC current), its separation from a ground-plane or ground-trace (this in addition to any line capacitance and inductance sets the line impedance), and the native impedance of the output driver.

If you want a really quiet S-100 bus, then you need to cap the signals on the top and bottom of the PCB with ground planes (which turns it into something similar to a coax cable...which is what I do on the S-100 ATX Motherboard).  You still have to bring the signals up to a connector to connect to an S-100 board, so the coax analogy is not perfect.  And in addition to the top and bottom ground planes you will need to control the current injection of the output drivers when they change state (which I do on the S-100 ATX Motherboard by inserting a series termination resistor from each connector down the "coax bus").  This limits the current injection by the output driver during signal change transitions which serves multiple purposes:

• It reduces the signal transition ringing due to the capacitive current back-flow of the input gates on the various S-100 cards.
  
• It limits EMI/RF radiation from the signal lines as the AC current is reduced.
• It limits signal transition harmonics (since the ringing component of the signal is reduced/eliminated).

Of course, you can only do so much, and if an add-in S-100 card has noisy transitions internal to the card, it is going to radiate as well.  This is how an IMSAI can play Daisy for you!

So, going back to your original question ("Seems there should be some way to test the designs to see if they're working as antennas or something else."):

If there is a frequency, any trace connected to it is an antenna.  Whether the signal it produces is detectable is dependent on the factors I listed above.

And I haven't even broached the subject of multiple higher frequencies and side-bands produced when you mix frequencies...so I'll stop here.  ;-)

Let me know if I can help on any of the bus related discussions, but if it's not S-100 related, perhaps we can make it a direct email discussion?  (Unless the group at large likes delving into the minutiae of signaling theory...)

Thanks and Best Regards,

Jeff

Quote from Greg  B:  

as for the method of testing... making something and proving a design satisfies everyone gets to be tricky to difficult...

>> what S-100 line drivers and receivers are used 74xx, 74LSxx, 74HCTxx, 74Fxx, etc. down below in this chain one person mentions they are using 74VC series

>> what are gate "fan outs" or the ability to drive other cards in a system at a particular speed?   each gate type family has their own abilities. 

>> I have a 10 card back plane, some have 3, 6, 9, 12, 18, 22...  lots of variation, lots of different uA to mA loading and signal capacitance

>> there are a 100s or more different S-100 cards with different layout qualities (with aging solder joints signal qualities vary too)

>> Ground planes and Ground guard lines parallel to signals help prevent cross talk but do add pF level additional capacitance, hence more fan out to over come those characteristics

>> over shoot and ringing can be problems (that little ripple when going to 1 or 0 until the signal stabilizes at a flat level)

>> Thus when someone tests their own board out, record what their set up or system configuration as a  snap shot in time with a, b, c, .... xxx, yyy, zzz parts.

>> if you want to know if energy is radiating out of signals as an antenna, get a cheap old AM band radio, sweep across the AM band... the buzzing you hear is the noise of

      everything around you... the loudest will be the closest to the cheap radio.  Looping software tends to almost make a musical pattern.

>> one can use an oscilloscope to look at rising and falling edges... faster speeds need sharper edges or fast slew rates, faster slew rates contain higher frequency content in their edges

     The FCC regulates what frequency zones one can and can not radiate radio noise.   Hence there is EMC compliance certification for mass production electronic products.

>> I noticed on some S-100 cards bus signal receivers are Schmitt trigger type (ei. 74xx16) to square up and clean signals locally into each S-100 card but that adds 25nS delays

     not critical for 1 to 4 MHz systems, but could be critical for maybe 7 to 20MHz CPUs

Regards  Greg  B

Quote from Peter Higgins:  

...and therein lies the problem. As I see it there are two very different requirements:

- systems with vintage S100 board designs, for which practically any S100 backplane will work. These systems do not need a "new" backplane design, and all currently available options easily meet the requirements.

- systems built from John Monahan's new S100 board designs, with CPUs running at 10MHz+ bus speeds and/or fast switching TTL LVC bus drivers. It can be very difficult to get these boards to work reliably with many vintage S100 backplane designs, including the popular 9 slot N8VEM design. As John discovered with his recent prototypes, designing a backplane that works reliably with these systems is far more challenging.