Non-Oversampling dac & tweeters

[lucpes]

I have found mixed opinions on the net regarding the possible negative influence (eg. kaboom) that the HF garbage (images) that non-os dacs produce might have on extended response tweeters that do not have a protective self-inductance.

I have a pair of Infinity RSII's (EMIT planar-magnetic tweeter - flat to 32kHz) which came from the factory with 1.5A fuses on the tweeters (paralelled, 2 ohms impedance).

edit: amplifier bandwith is DC-150kHz, -3dB

Should I worry about this?

some spectrum data:

http://www.diyparadiso.com/teksten/i.../solnodac1.htm

(not I'm going after but the non-os output should be similar)

Thanks!

[pburke]

Magnepan 3.5R here, tweeter fuses removed, non-OS DAC, happy camper.

[lucpes]

So I guess everything would be ok. I'm looking for more opinions though...

[Werner]

This is generally unproblematic, for two reasons.

First, the zero-order-hold nature of most NONOS DACs (exception: the TNT Convertus Decima, which was just an experiment) implies a treble rolloff following Sin(f)/f, which attenuates the HF output strongly above 35kHz, and even nulls it completely at 44.1kHz.

Secondly, the spectrum occupied by most music falls off drastically above 4kHz or so.

These two things combined mean that in the band 20-35kHz there is not a lot of energy (due to the nature of music), and in the band 35-50kHz there is not a lot again, due to the Sin(f)/f droop.

Remember that it is power that destructs tweeters, and even with a signal voltage attenuation of 10 dB (not that much), power is down to 10% of the original level.

[lucpes]

Thanks werner!

PS. I just put the tweeter fuses back in (sounds a bit with my current setup) and if everything goes ok I'll remove them after I finish building & testing the thingie.

[lucpes]

The DAC is up & running and no problem in two days with fuses on, I'll just remove them

[Bricolo]

Originally posted by Werner
This is generally unproblematic, for two reasons.

First, the zero-order-hold nature of most NONOS DACs (exception: the TNT Convertus Decima, which was just an experiment) implies a treble rolloff following Sin(f)/f, which attenuates the HF output strongly above 35kHz, and even nulls it completely at 44.1kHz.

Secondly, the spectrum occupied by most music falls off drastically above 4kHz or so.

These two things combined mean that in the band 20-35kHz there is not a lot of energy (due to the nature of music), and in the band 35-50kHz there is not a lot again, due to the Sin(f)/f droop.

Remember that it is power that destructs tweeters, and even with a signal voltage attenuation of 10 dB (not that much), power is down to 10% of the original level.

AFAIK, this is untrue.

The sinc(f) rolloff exists in a non-os dac, agreed. But this is for the music reproduction. In other words, information stored in your cd, will have a slight treble rolloff when player through such a dac.
And, even more obvious: there's absolutely no music content on a CD, above 22kHz (Fs/s) so 35kHz and 44kHz sounds aren't attenuated, there aren't any! It's impossible to have them.

The 44.1kHz signal you have on the output of non-os dacs is due to the fact that your signal is sampled at 44.1kHz, so the dac sets a value on its output, maintain it for 1/44100 second, outputs a new value and keep it also for the same period, etc...

If you look closely at this output signal, it will be made of small steps
And those steps cause the 44.1kHz "garbage" on the output.
Decompose the signal in fourrier series, you'll see that the regularly spaced steps (at every 1/44100th of a second) will show an important signal at this frequency.

[rfbrw]

Bricolo, you seem to be forgetting something. Without any filtering the entire 0Hz to 22k05Hz spectrum repeats every 44k1Hz.

[Bricolo]

could you explain?

[Werner]

Originally posted by Bricolo


AFAIK, this is untrue.

AFAIK, there is a lot you don't know.