RF Attenuators (Again)!

[jkeny]

Can I draw people's attention to a thread & series of measurements (done by an expert in RF) in this thread Reflections and attenuators

In the thread he explains how reflections will be attenuated by twice the amount of the dB value of the attenuator (I.e 10dB attenuator will decrease the signal by 10dB & the reflection by 20dB). He explains what it will NOT attenuate - overshoot, ringing, etc.

He also shows a number of scope shots to illustrate what happens & explains in detail each shot.

I urge everybody (who believes in measurements) to read this thread & educate themselves on the operation of these RF attenuators.

[godfrey]

Hi John

Nice to see some intelligent conversation on the subject.
...and relevant tests and measurements too!
Wassup with that - don't they have trolls there too?

I seem to remember that around here the subject got buried in idiotic arguments like "Properly designed equipment is properly terminated, therefore there are no reflections ..."

btw, something I've been wondering about:
IIRC, the Hiface output signal is higher than the standard level, the waveform doesn't look great (according to some scope shots I've seen), and the output impedance isn't a great match for the cable.

Have you considered tweaking your modified Hifaces to get a cleaner output signal and a better impedance match? I'm starting to get the idea that lousy impedance matching is almost normal with digital audio. If you get it right, you'd be ahead of the pack. Makes more sense to me to fix the root cause of the problem instead of treating the symptoms with RF attenuators.

On a vaguely related note, I suspect variable output level would be tweaker's paradise, and justifiably so as different DACs are probably happier with different signal levels.

Cheers - Godfrey

[jkeny]

Hi Godfrey,
Good comments!
But a couple of points - the overshoot on the SPDIF waveform is a result of the output trafo used. This isn't something that is fixed by the RF attenuators - it is somewhat ameliorated.

The output impedance is measured on computeraudiophile as 73ohms - a perfectly acceptable match to the cable & within the margin of error.

The reason why you see a bad SPDIF waveform is because of the rise time speed - the higher speed causes more reflections. This is why when measurements were done in the previous thread here of a cheap Chinese USB-SPIF converter it showed few reflections.

The seeming advantage of a faster rise time is to reduce inter-symbol interference. So what? This seems to be a method of reducing deterministic jitter - the worst jitter as far as sonics are concerned, it seems. So if the reflections can be tamed we end up with a sonic advantage. RF attenuators help in this taming.

Yes you are correct in your assumption about impedance matching in digital it would seem.

The Rf attenuators are so cheap ($15) that buying a number of different values is the equivalent of a variable one!

[godfrey]

Originally Posted by jkeny

The output impedance is measured on computeraudiophile as 73ohms - a perfectly acceptable match to the cable & within the margin of error.

Hmm, I wonder...
Given that there's an output trafo involved, I'd expect that to be strongly frequency-dependent. What would be interesting is a reverse input test, i.e. send a pulse up the cable to the Hiface's output and see what bounces back.

This is of academic interest only for me anyway, since I'm not using any SPDIF equip. It's just my curiosity getting the better of me again.

[jkeny]

Have a look at the way they measured it using a scope, the peak voltage of the SPDIF waveform for both unterminated & a 75 ohm termination. It is a correct way of measuring the output impedance but has a certain margin of error.

I agree - I use I2S & find it far superior

[wakibaki]

Hmm, curious that none of these guys come over here to argue their case. Or perhaps they've already been banned for their antisocial behaviour?

w

[jkeny]

Originally Posted by wakibaki

Hmm, curious that none of these guys come over here to argue their case. Or perhaps they've already been banned for their antisocial behaviour?

w

That's the totality of your comment - nothing about the attenuators or results - oh well, it reveals a lot!