Bnc installation

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Oh, BNCs are just lots more cooler.

When I was learning electronics and building stuff in the late 1950s post-war surplus era, I had tons of BNC cables and connectors and used them for everything. Clicking them on and off just seemed so darned industrial. NO electrical advantage, just style points.
 
RCA's are terrible (c**p) and add an impedance match, not what you want for a digital signal. The return and signal should also be kept together, from transmitter to reciever, for digital signals, do not seperate the return by staring it to the PSU or some other scheme.
 
Half an inch should not do much harm, but the shorter the better.

Let's do a rough back-of-envelope calculation. Assume 16-bit data at 48kHz sampling rate. So each stereo sample has 32 bits of data, plus some overhead. I can't be bothered to look up the details, so let's just assume 40 bits per sample in total. This gives a basic data rate of 1920kbit/s. I think it uses Manchester encoding, but I can't even be bothered to look that up either. Let's assume a basic signal rate of 2MHz. You need a reasonable square wave, so let's assume at least the first 50 harmonics. This means the link has to be good to about 100MHz, which is not too difficult.

A disturbance in a transmission line will not do much harm if it is less than about wavelength/50 in size. 100MHz is 3m in free space, so we can have up to 6cm disturbance. Let's not use more than half of this allowance, so the other end can do it too. So we get 3cm.

This is a back-of-envelope calculation, so the result could easily be out by a factor of 2 either way. However, it gives a rough idea of what is needed. If you might want to use 24-bit audio at 192kHz then you may need to be more careful, and treat the whole thing as a UHF circuit.
 
Driver's rise/fall times are these which matter in transmission lines.
If the transmission line is longer than 1/10 of transition time's wave length, then short discontinuities will cause reflections too - that's why they use these bnc/tnc/sma/etcs despite they are short relatively to the overal length of the transmission line.

For HC family it's 2.48m wavelength, so the longest transmission line without reflections would be... 24cm.
AC family gets worse with 10cm limit.

Oh, and the jitter... The faster the transition time and the cleaner it is - the less jitter we'll have.

But yup, SPDIF worked for me with nokia's charger wire too, bit perfect et cetera, no problems. It should work with mains cables too.
The question is - why do the BNC mods - and ignore other transmission line caveats at the same time?

By the way, are these the real 75-ohm BNCs you have? /making sure they aren't 50s - 50s have plastic insulation inside the connector coupling space - that's how you tell them apart/
 
Last edited:
Yep mismatch.
Any pigtail should be as short as possible, they have the same effect as a slot in the ground plane, creating a discontinouity in the return path. Whether it would have much of an effect, probably not at lower data rates, but could have on noise.
I was reffering to one scheme were the SPDIF return went via a long wire back to the main PSU star point, this greatly increased the loop area for the signal and created a torterous path for the return current path.
Always best to employ best practice if possible. I have seen problems in real world, on a DDR2 memory board, the bigest problems we had were with a 12MHz JTAG clock. With 100MHz clocks etc every where this measly 12MHz was not thought critical, but going to numerous devices, it turned out the driver chips output was not up to the job, and we had to use a clock buffer.
SPDIF is a manchester encodiing based transmitted signal.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.