I have a focusrite saffire that I'm finally getting around to examine properly, and there is an issue with clipping when the volume knob is set to full output.
The purpose of this thread is to try to figure out why it distorts, if this means that the performance is reduced significantly, and if there is anything I can do to correct it.
Here are some of the details:
The dac chip is a CS42428 codec:
http://www.cirrus.com/en/pubs/proDatasheet/CS42428_F1.pdf
I'm providing these numbers from memory and they are likely off a little bit, but this should give an idea of when the problem happens.
The volume knob potentiometer is fed (I believe it was) ~3.2V and it attenuates it. After about 2.8V it begins to clip heavily.
The output after the pot goes to the power board and I couldn't follow it, and I tried to measure the change outside the dac but couldn't find it.
If anyone can help me try to figure out what it all means and if there is anything I could do about it I would greatly appreciate it. Thanks.
The purpose of this thread is to try to figure out why it distorts, if this means that the performance is reduced significantly, and if there is anything I can do to correct it.
Here are some of the details:
The dac chip is a CS42428 codec:
http://www.cirrus.com/en/pubs/proDatasheet/CS42428_F1.pdf
I'm providing these numbers from memory and they are likely off a little bit, but this should give an idea of when the problem happens.
The volume knob potentiometer is fed (I believe it was) ~3.2V and it attenuates it. After about 2.8V it begins to clip heavily.
The output after the pot goes to the power board and I couldn't follow it, and I tried to measure the change outside the dac but couldn't find it.
If anyone can help me try to figure out what it all means and if there is anything I could do about it I would greatly appreciate it. Thanks.
In the software control, there is a way to override the hardware knob and force full volume. This too has the same issue, it must be at like 75-80% of the way up before it clips.
Does this mean that the dac doesn't have enough voltage being supplied to it, and it is simply clipping trying to put out the volume? The analog voltage measures around 4.88. The digital voltage measures around 3.2. The requirements for the chip are:
VA: min 4.75, typ 5, max 5.25.
VD: min 3.13, typ 3.3, max 5.25.
Is any of that indicative of the clipping issues?
Does this mean that the dac doesn't have enough voltage being supplied to it, and it is simply clipping trying to put out the volume? The analog voltage measures around 4.88. The digital voltage measures around 3.2. The requirements for the chip are:
VA: min 4.75, typ 5, max 5.25.
VD: min 3.13, typ 3.3, max 5.25.
Is any of that indicative of the clipping issues?
Well am I missing anything by having to run the volume down so far? Globulator its not the material, I am outputting a sinewave at -1 and just raising the output knob.
No, not unless noise is becoming intrusive or actual volume is inadequate for listening. If the actual volume is inadequate then you can put a preamp in the system, again, as long as noise (hiss) does not become intrusive, or use a more powerful power amp.
w
That isn't the issue, there is perfectly adequate output level at any of the levels. The concern is if the attenuation is either being done digitally and I'm losing steps in the material or if the dac is being fed all the steps but because of the level the dac is no longer operating within its optimal conditions and I'm losing some SNR performance.
If the answer is yes to option 2 above then i can live with it. I would rather not have a device which has to truncate data to operate properly.
Neither of these are true. Losing steps in the material would require software modification of the numbers going into the DAC i.e divide by 2, divide by 4 which would result in low numbers dropping out, not clipping. It would result in loss of resolution and SNR, but this is not what is happening, what is happening is clipping. It's not a question of the DAC no longer operating within its optimal conditions and a loss of SNR performance. If there is a loss of SNR it would mean increased audibility of hiss, that is what is meant by Signal to Noise Ratio. It's not something abstract, if you can't hear it, you can't hear it.
The attenuation is being done digitally insofar as it is under software control, but this will be a digital pot (which may be alternately controlled by a wheel) or other form of digital gain control after the DAC. There must be some gain device before the output if you can drive it into clipping.
w
OK I had a look at the DAC datasheet, the volume control is integral to the DAC, it will not be responsible for the effects you describe, who in their right mind would design a chip that acted that way?
Last edited:
wakibaki,
Thanks for taking a look at the pdf. Can you briefly explain what you think is going on? I'm sorry if you have already done so with the edit but I can't figure out if you are saying its really is a strange design or if my concern is ill-founded because nobody would implement it such that my concerns would even be warranted.
Thanks for taking a look at the pdf. Can you briefly explain what you think is going on? I'm sorry if you have already done so with the edit but I can't figure out if you are saying its really is a strange design or if my concern is ill-founded because nobody would implement it such that my concerns would even be warranted.
OK, it's 3 in the morning here, this is taxing my brain and it's going to take a fair bit of writing. I was a bit over-hasty with the remark about the digital pot or gain control. I need to get some sleep but I will write a comprehensive explanation tomorrow after I have had a chance to do some calculations to verify that my understanding is correct. The datasheet is by no means forthcoming about exactly how the volume control is implemented, but on reflection it looks as though there is a loss of resolution and SNR as the samples being presented to the DAC are no longer full-scale. Let me sleep on it and I will get back to you...
w
w
panomaniac, it is something to do with the output. It only clips at the input when I use full balanced output to balanced input.
sangram, there is no setting for that. The pot is not in the signal path, I believe it gets converted to a digital value somewhere. I think we can forget about the pot though, as I can bypass the hardware knob and use the software knob. At full software volume (which is full hardware volume) it behaves the same.
Check this out:
TS unbalanced out (using the output jack) to TS unbalanced in, output clipping (-20 room at input).
TS unbalanced out (manually connecting to one of the differential out legs, meaning nothing is in the jack) and jumping the ground to the same ground pin as normal NO CLIPPING (-20 room at input).
From that information it seems that something is going on at the output stage that doesn't like being run in unbalanced mode.
sangram, there is no setting for that. The pot is not in the signal path, I believe it gets converted to a digital value somewhere. I think we can forget about the pot though, as I can bypass the hardware knob and use the software knob. At full software volume (which is full hardware volume) it behaves the same.
Check this out:
TS unbalanced out (using the output jack) to TS unbalanced in, output clipping (-20 room at input).
TS unbalanced out (manually connecting to one of the differential out legs, meaning nothing is in the jack) and jumping the ground to the same ground pin as normal NO CLIPPING (-20 room at input).
From that information it seems that something is going on at the output stage that doesn't like being run in unbalanced mode.
Hi
Is it clipping when sending a 0dBFS file to the mixer from a software input? In both TS and TRS mode? Or only in loopback mode?
If you are getting clipping at the input it is possible that the input gain is turned up. Most cards of this kind (though admittedly not all) have a software switch for 'consumer' and 'pro' level. Some have it only for outputs with the inputs assumed to be 'consumer' level, so a pro level TS output will clip it pretty easily, even when not turned to max.
My Delta 66 used to have a level toggle in older driver revisions, but that has been replaced by a mixer fader for each input and output. When left at maximum level for all, it will clip in loopback mode even at -10dB output because input sensitivity is way too high. Thankfully, my e-Mu still has a control for the output (don't remember input, never recorded with this card).
Is it clipping when sending a 0dBFS file to the mixer from a software input? In both TS and TRS mode? Or only in loopback mode?
If you are getting clipping at the input it is possible that the input gain is turned up. Most cards of this kind (though admittedly not all) have a software switch for 'consumer' and 'pro' level. Some have it only for outputs with the inputs assumed to be 'consumer' level, so a pro level TS output will clip it pretty easily, even when not turned to max.
My Delta 66 used to have a level toggle in older driver revisions, but that has been replaced by a mixer fader for each input and output. When left at maximum level for all, it will clip in loopback mode even at -10dB output because input sensitivity is way too high. Thankfully, my e-Mu still has a control for the output (don't remember input, never recorded with this card).
sangram, I use a -1 dBFS tone generator from software. No clipping going on in the software side.
It clips in TS mode, no clipping in TRS. The input clip happens in loopback because the output is so strong. It says 0dBFS is 16dBu which I gather is pretty damn high output for most non-pro devices.
It clips in TS mode, no clipping in TRS. The input clip happens in loopback because the output is so strong. It says 0dBFS is 16dBu which I gather is pretty damn high output for most non-pro devices.
That may be true but the input stuff doesn't matter. The output clips when using an unbalanced connector, its not a function of the input clipping, so what matters is figuring out why it does it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.