For the TDA1545A its a current output, they say 200nS, to 1LSB. That's a DAC I've been using for the past few years.
Do explain why its 'kinda misleading'. Its not my statement (you added 'can') but don't let that stop you.
It doesn't, but higher speed DACs don't output waveforms that resemble the input in the ways that matter to our ears. Superficially of course with infinite amounts of oversampling the 'steps' vanish but the steps are there by design they're not an artifact of the DAC's non-ideal implementation.
I'm not myself a fan of the roll-off and I correct it in my designs.
The zero order hold is there by design, as already mentioned. The error introduced by this is mathematically described and amounts only to a linear distortion - i.e. a frequency response droop which is fairly easily corrected. When I spoke of 'at the wrong value' I was speaking of values of DAC output which aren't linear distortions of the input waveform and hence cannot be undone by later processing.
You and me both - hopefully you've now grasped that I was talking about distortions introduced by the DAC which can't be cleaned up later?
I think you need to read some electronics book .
500LSB is a simple matlab test on normal samples of a music signal...
On Nos 16khz is possible by a certain compensation network ... 😱
Images on output of all dac' s have infinite replays! Don't tel me others stupid lies 😡
Only the distance between them is changed 🙂
I thought it was clear, but let me try an analogous blanket statement:
"High fuel consumption is bad and faster cars consume more fuel."
Now I hope it is also clear why I added 'can'.
Oh no no no, it's NOS DACs that don't output waveforms that resemble the input ... and that's what matters, to our ears and in general. 😛
Except that it produces a lot of images too which is not linear distortion and which you can't properly get rid of even with a fancy analog filter, without other deleterious effects anyway.
And this is again just a blanket statement to act as FUD. In reality the TDA1541A performs worse also in terms of nonlinear distortion, dynamic range ... than more modern, oversampling products.
So another number you pull out of thin air and again completely evade my simple questions about it.
And you obviously didn't understand any of the previous posts on that 16 kHz nonsense.
And of course there will still be images (I didn't say otherwise yet you say I "lie"), but the digital part attenuates them in a range where a simpler, better performing analog low pass hasn't reached its desired attenuation yet.
Okay.. I give up. I'll end this pointless exercise here.
I give you the benefit of the doubt that you don't deliberately misrepresent what I said and just assume that you don't understand English well enough to have a discussion.
Nope, still lost.
So is that sentence above just a restatement of your earlier misunderstanding (regarding zero order hold), or another (different) mistaken opinion?
Oversampling DACs produce lots of images too so I can't see what your point is. What are the 'deleterious effects' you're referring to? Phase anomalies perchance? Are you arguing for linear phase filtering in conjunction with oversampling?
I haven't mentioned the TDA1541A so far but you're welcome to take pot-shots at it as you wish.
Then I can't help you.
Neither, and there was no misunderstanding.
But it seems that you agree that your statement is a mistaken opinion.
Sigh...
Read my sentence and you will see what my point is and why your response doesn't even make any sense.
On the deleterious effects, let me also just make blanket statements: higher order analog filtering means worse performance.
A typo. TDA1545A it is.
Could you share the parameters of the low pass you use with your NOS DAC (Q, order, cutoff frequency) and for the droop correction so I can do a quick simulation on how the output probably looks like?
Last edited:
Nope, your claim didn't work. I don't see your point and your claim that my statement doesn't make sense also doesn't fly. I can quite see how it won't make sense to you but that's another matter entirely.
There are a selection of filters on my blog - the most recent one uses two inductors (10mH). Not sure what the Q is, given two inductors its 5th order notionally. The filter I'm using for droop correction is 2nd order, single inductor. If you look at this thread you'll see matt's putting together a complete schematic of the modified parts of the DAC - http://www.diyaudio.com/forums/digi...ign-mod-not-play-music-not-6.html#post4426584
Xonr you have misunderstand dac world.....😱
Digital filter is on digital domain and not after the dac.... ohi ohi
you must learn the bases of the conversion
Digital filter is on digital domain and not after the dac.... ohi ohi
you must learn the bases of the conversion
Of course you don't, you're still skipping the middle part of my sentence about filtering these images which with a NOS DAC will already start at Fs/2.
Thanks, I'll take a look at it.
No I haven't, but you really should learn English.
@abraxalito
I've done a quick sim but the results are not very good. Maybe I've messed up but I get a few dB probably audible ripple ...
Last edited:
If there's a few dB then there's an error somewhere, either from me or you. Do share so I can help track down where the problem is. When I was designing the filter I aimed for around half a dB ripple. What DCR did you put in for the inductors? That will affect the Q - the value (from memory) should be 7.4ohm.
the your way
I have evaluate your solution ..
I think it can decrease the settling time but you need more then one cycle and you need
an appropriate phase on additional sine wave .
if step goes up you need a normal sine otherwise a 180 degrees phase sine wave
I have evaluate your solution ..
I think it can decrease the settling time but you need more then one cycle and you need
an appropriate phase on additional sine wave .
if step goes up you need a normal sine otherwise a 180 degrees phase sine wave
@abraxalito: The response certainly changed with some series resistance added, but it's still like +/- 2 dB and the roll-off starts even earlier now.
Anyway, I just wanted to get a rough impression of the filter, which I got, and it's not positive.
@gumo73: So if you don't understand something please don't say "lies!" or "read book X". That's just offensive.
Regarding your latest response:
Forget settling time.
Also, you can't counter those stairsteps with an inverted sine. The "height" of each stairstep can be wildly different, so the sine wave itself would have sharp amplitude/phase transitions.
This just gets worse with a (much) higher frequency sine wave.
Anyway, I just wanted to get a rough impression of the filter, which I got, and it's not positive.
@gumo73: So if you don't understand something please don't say "lies!" or "read book X". That's just offensive.
Regarding your latest response:
Forget settling time.
Also, you can't counter those stairsteps with an inverted sine. The "height" of each stairstep can be wildly different, so the sine wave itself would have sharp amplitude/phase transitions.
This just gets worse with a (much) higher frequency sine wave.
Then something got changed vis-a-vis the original design. I'll look over the schematics that matt posted and compare them with my original LTSpice file. Perhaps somewhere along the way there was an error in transcribing the component values.
Clearly, for whatever reason, you didn't simulate the filter as originally designed. If you're content with that, then so am I 🙂
Ok, so I've taken another look and added a missing R to ground on the input side of the filter. It is flatter now, but it is still essentially the same filter.
If you send a bandlimited impulse through it you will get a delayed and distorted output with this, so my first impression didn't change.
But I could see how this might sound different to you.
If you send a bandlimited impulse through it you will get a delayed and distorted output with this, so my first impression didn't change.
But I could see how this might sound different to you.
The distortion will be phase distortion for sure because the filter's fairly steep. I'm not sure I can hear the phase distortion. Delay bothers me even less than phase distortion.
I don't know what you have studied in your life but I'm a electronic engineering
This is the end whit you to toking about stupid things....
A realistic implementation is by a square wave or a triangle because you need a fast start and stop and a fine phase control on signal added on current dac output this is possible by a current sum to signal the output of I/V can setup to right value too fast
than a normal implementation
- Status
- Not open for further replies.