View Full Version : RCA Digital Output Stage Impedance

2005-09-12, 12:36
I am sorry if this has been asked before, but I can not seem to find it.

I simply want to know what the output impedance of the RCA digital output is. Is the total impedance of the driver output stage plus any series resistance = 75 ohms? If not, what is it?



Deaf Cat
2005-09-12, 13:57
I'm not too sure, but I think my digi coax cable is rated at 100ohms, I know it is higher than normal phono to phone leads........

2005-09-12, 14:01

I may not have been too clear. I want to know the output impedance of the Squeezebox 2 digital output. Is there a schematic available for that part of the circuit?



Deaf Cat
2005-09-12, 14:27
No, you were clear don't worry about that, I just thought the little info I do know might have just helped a little bit..

Intrested to hear what it may be.

Mike Hanson
2005-09-13, 07:39
S/PDIF is supposed to be on coaxial 75-ohm cable, connectors, etc. Therefore, I would start with the assumption that the output stage on the SB2 is also 75-ohm.

BTW, RCA connectors can never truly be 75-ohm, by their very nature. The proper connector is BNC, but market pressures have caused RCAs to appear more often (even though they are wrong).

-=> Mike Hanson <=-

2005-09-13, 08:30
the product info page (http://www.slimdevices.com/pi_specs.html)
says the following about the coax out:
Coax connector: RCA, 500mVpp into 75 ohms


2005-09-13, 08:47
Yes, I saw the product info describing the output drive into a 75 ohm load. But the real question is...is the SB2 properly backmatched with a 75 ohm impedance to absorb the reflected wave from the load? If the SB2 output impedance is small, say 25 ohms like a typical CMOS driver, then the actual destination load impedance must be 75 ohms to absorb the forward wave. My DAC (Wadia 861) does not have a 75ohm load and thus the transmission line must be backmatched to prevent reflections from the source.

It seems that I will simply add a BNC "Tee" with a 75 ohm load at the Wadia to guarantee a reflection free transimission line.

Oh, and you are right, RCA connectors were never intended for high frequency transmission and it really should be a BNC connector.



2005-09-13, 11:35
The driver impedance is 75 ohms, as should be your cable. If the load is not terminated (?!?!?!) then yes you'd need to add a 75 ohm resistor. I have never heard of a DAC not having a termination resistor. Some have a jumper inside so you can make it terminated or not (in order to drive multiple devices on one line). Very surprised about the Wadia.

2005-09-13, 13:42
Thank you Sean,

That was the information I was looking for.

Actually, it is acceptable to back match a 75 ohm transmission line with only a source match and an infinite impedance at the load. This is appropriately known as "back-matching". In this type of transmission system, the voltage at the load reflects in phase and doubles. It is then absorbed back at the source side.

However, I agree with you, I have always felt more comfortable with source and load matched to the cable characteristic impedance.



2005-09-13, 14:32
The PCI bus (the only example I can think of) does this in order to reduce EMI and switching current at the expense of speed. It's definitely not the s/pdif spec though, and seems like kind of a bad idea for a self-clocking transmission line where cable length is not controlled.

2005-09-15, 13:40
This is a bit off topic, but I wanted to correct an erroneous statement I made previously. The Wadia 861 CDP that I use for an external DAC with the SB2 does indeed have a 75 ohm termination. However, it is only terminated for AC signals (i.e. they use a cap in series with 75ohms) which is perfectly acceptable. I used an ohm meter to measure it, and of course, I measured an open circuit. Wadia actually provided me with the input circuit and it is implemented properly and includes high speed clamp diodes as well.

The Wadia 861 is a magnificent DAC that simply sounds musical and never harsh. Voices and instruments are palpable in 3 dimensional space and it is a joy to listen to. It is expensive, but can be had on Audiogon for less than 1/2 MSRP.

It is truely fantastic that the SB2 can be purchased for a mere $200. It has completely changed the way I listen to music. Well done.


2005-09-15, 14:13
Very interesting - sounds fine but I wonder why they did it that way. Presumably there's a transformer already knocking out the DC?

2005-09-15, 15:04

It doesn't look like a transformer coupled interface although that could be used as well.

I can think of two reasons to use a DC block for tranmission line matching.

1. To reduce quiescent power dissipation.

2. To eliminate ground loops when using single ended interfaces.

I doubt they did it to reduce power and I suspect they did it to eliminate 60Hz ground loops as the shield of the BNC is AC coupled to chassis ground as well. The interface schematic does not actually show the receiver, but it appears that it is differential as both the center conductor and the shield are carried out to what ever receiving device they use.

2005-09-15, 21:48
Hmmm.. if Wadia is good with capacitor coupling at their price point, it makes you wonder why anyone is spending for transformers. AFAICT the waveforms are identical at these frequencies... can anyone shed some light?

Andrew L. Weekes
2005-09-16, 05:45
Hmmm.. if Wadia is good with capacitor coupling at their price point, it makes you wonder why anyone is spending for transformers. AFAICT the waveforms are identical at these frequencies... can anyone shed some light?

Without seeing the full Wadia circuit, it's hard to be specific, but very few (even expensive) products have a well-designed SPDIF circuit - one reason for using a (good) transformer is is makes it easier to acheive a wide-band, purely resistive, 75R match, which is critical for good SPDIF performance.

The SB2 for example can *never* be a true 75R resistive impedance, since the phono socket, for one, prevents it from ever being so, it causes an impedance mismatch, which will cause reflections on the SPDIF transmission line.

That said, most devices with a transformer don't acheive a good match either, for a number of reasons, like transformer choice, lack of proper testing and engineering etc.

This seems to be critical for optimal SPIDF performance and is in the realms of RF engineering rather than audio. One needs to use Time Domain Reflectometry (TDR) to see the reflections in the line (you can't see it on a 'scope), that come from the mismatches inherent in most setups - eliminating these (by proper, wideband, impedance matching) works wonders.

The SPDIF interface should be considered an RF circuit and should receive the same attentions as any RF transmission line interface.


P.S. Another reason for the cap could be to prevent saturation of the following transformer (a common problem anyway, even without DC) in the presence of any DC, but in reality this probably creates more mismatch problems and it's probably better to assume no DC at the sending end (which should be the case, in a properly engineered transmission circuit).

Andrew L. Weekes
2005-09-16, 06:13
Another reason for using transformers, is they are just about the only way to acheive any significant high frequency common mode rejection.

Trying to do this actively, is virtually impossible.


2005-09-16, 07:39
FYI - here's an interesting article that discusses digital coax cable length with specific regards to SPDIF impedences et al. It's written by Steve Nugent of Empirical Audio...I have a modified Sony S7700 done by EA, and the work is fantastic.