0
\$\begingroup\$

My circuit has to emit an S/PDIF digital audio stream over an RCA connector. According to the specs (from Wikipedia) the signal should have an output impedance of 75Ω and a peak-peak voltage of 0.5V.

I've found a schematic for a TTL to COAX converter: TTL to COAX converter schematic

but it's meant to work with 5V digital output, while my device (MAX II CPLD) works at 3.3V. So I'm trying to adapt this schematic for 3.3V by changing the resistor values.

As far as I understand the base circuit, including the input device, should look like this:

schematic

simulate this circuit – Schematic created using CircuitLab

I have to compute Rx and Ry so that I can properly set up a voltage divider from 3.3 to 0.5V, and make a current of (0.5V / 75Ω =) 6.67mA flow from my device to the load. But I'm struggling in setting up the system of equations.

So far I came up with this one for the voltage dividing: $${1 \over {{1 \over R_y} + {1 \over 75}}} = 0.18R_x$$

And this one for the currents: $${3.3 \over {R_x + {75R_y \over R_y + 75}}} = {{0.5 \over R_y} + 6.67mA}$$

But when "developing" them they become really long. So I'm not sure if the approach is actually right or if I just did some steps wrong.

I'd also like to know the total current, to check if it's below the maximum source current of the CPLD.

\$\endgroup\$
4
  • \$\begingroup\$ How strong drive does the CPLD output pin have? It might require buffering to be able to actually drive a piece of coaxial cable, unless it is extremely strong. \$\endgroup\$
    – Justme
    Commented Aug 18, 2020 at 15:53
  • \$\begingroup\$ It could be 16 or 60mA (will eventually make another question to clarify), in any case the cable will be very short, 10cm tops. \$\endgroup\$ Commented Aug 18, 2020 at 15:59
  • \$\begingroup\$ You should be aware that your circuit is not guaranteed to work reliably. The minimum output voltage for TTL is 2.4 volts, but the minimum required voltage for 74HC is something greater than 2.5 volts. If you want to use the circuit, either put a 1k pullup resistor to +5 on the input, or replace the 74HC with 74HCT. \$\endgroup\$ Commented Aug 18, 2020 at 20:15
  • \$\begingroup\$ I won't use those 74HC logic chips, I'm more interested in what's happening after. \$\endgroup\$ Commented Aug 19, 2020 at 16:28

3 Answers 3

3
\$\begingroup\$

The 390Ω || 91Ω will give you 73.8Ω looking into the TTL gate output from the load end which should be terminated in 75Ω. So this part had been set-up for matching the source to the tx-line and the 75Ω load. All you need to do is simultaneously ensure that the values of Rx || Ry = 75Ω, and that the voltage division generated by your Rx-Ry voltage divider is 0.5V into a 75Ω load to give you the whatever current you're after. The final Rx and Ry values should satisfy both conditions.

One equation will resolve to (Ry∙Rl)/(Ry∙Rl+Rx(Ry+Rl))∙V=0.5 the other will resolve to (Rx∙Ry)/(Rx+Ry)=Rl

Rl is your load impedance and V is your source voltage. You'll need to solve the simultaneous equations for Rx and Ry.

\$\endgroup\$
6
  • \$\begingroup\$ By "390 || 91" you mean parallel resistance? In the schematic I see those two resistors in series. Same for Rx and Ry. Could you clarify? \$\endgroup\$ Commented Aug 18, 2020 at 17:07
  • \$\begingroup\$ To find the output impedance, consider a voltage source on the outside, and short circuit the source on the inside, so the resistors are in parallel. Or if you like use Z=Voc/Isc and derive from there. \$\endgroup\$
    – tomnexus
    Commented Aug 18, 2020 at 18:32
  • \$\begingroup\$ Yes 390 || 91 Ω are in parallel. This is because looking from the load into the source (from right to left), the voltage source has 0 Ω (in paractice this will be some low value). This makes one side of Rx actually like "shorted" to ground. It is a common way to analyze such circuits where a voltage source is considered (ideally) to have 0 Ω and a current source (ideally) an infinite impedance. Now if you short the Rx side to ground from the load point of view it looks like a parallel resistance 390 || 91 Ω. This is why the value turns out to be almost 75Ω because they worked it out this way. \$\endgroup\$
    – citizen
    Commented Aug 19, 2020 at 8:27
  • \$\begingroup\$ So now if you were to drive the left side of the circuit with a signal from the load side (the inverse way), the circuit on the left side looks like a 75Ω load, just as you want it on the right side (the load side). \$\endgroup\$
    – citizen
    Commented Aug 19, 2020 at 8:30
  • \$\begingroup\$ Thanks, I wasn't aware of this way of analysis. Are there textbooks/websites/videos to introduce such concepts to beginners? \$\endgroup\$ Commented Aug 19, 2020 at 16:29
1
\$\begingroup\$

There is a very simple way to calculate the resistors.

The starting point is what you know already, unloaded signal is 1V, loaded signal is 0.5V, output and input impedances are 75 ohms.

If you short circuit the output, you get I=U/R=1V/75R=13.3mA short circuit current, so you need to get same short circuit current with 3.3V supply, so recalculate the resistance.

Then when you have that resistance, you need the unloaded voltage. Since you have 3.3V-1.0V=2.3V over the resistance you just calculated, you know the unloaded current.

Then you use that unloaded current and 1V to calculate the other resistor.

And then you can check if you end up with similar values used on other 3.3V SPDIF transmitters.

\$\endgroup\$
0
\$\begingroup\$

If it’s 5.5 V logic @ 3.3V consider the output already as 75 ohms +/50% then scale voltage with a 75 Ohm pad design. But 50 Ohms +/-50% at 5V.

verify yourself using Vol/Iol=Zol then load it to expect 50%.

103R Shunt and 220R Series divider ratio R/(R+Rs)=30% but source is ≈ 50R so series R reduces by 50 Solution Rs= 170, Rsh=103 Voc=1V, @ 75 Ohm Open cct.

I though it was 1Vpp no load and 0.5Vpp matched load. (?)

Thus 1V=30% of 3.3 with source = ——— Also If it was 3.6V logic (74ALCxx) @ 3.3V consider it as 25 Ohms and do the same.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.