I have recently removed my TDA 1543 from the original DAC board and fed it with a new 8.3V regulated supply. I have changed the output resistors to 2K7 and bias resistor to 1K5 as per;
http://www.diyaudio.com/forums/digi...nting-cd-pro-14.html?postid=173483#post173483
and
http://www.diyaudio.com/forums/audio-sector/187748-pushing-limits-tda1543-nos-dac.html
I can not figure out whats up, I'm getting really bad distortion if I increase the volume over 5% from my digital source. The output is clear and listenable below this level, although at a relatively low level and not using anywhere near the range it should.
I have measured the dc voltage across the output resistors and it's at an expected 4.3V. I have also measured the ac output voltage after the output caps, with an audio source I get roughly 0.2V.
I originally thought that I may have blown the TDA, but I have replaced it with two others and am getting exactly the same results.
I'm stuck, I don't have a clue where to go from here?
http://www.diyaudio.com/forums/digi...nting-cd-pro-14.html?postid=173483#post173483
and
http://www.diyaudio.com/forums/audio-sector/187748-pushing-limits-tda1543-nos-dac.html
I can not figure out whats up, I'm getting really bad distortion if I increase the volume over 5% from my digital source. The output is clear and listenable below this level, although at a relatively low level and not using anywhere near the range it should.
I have measured the dc voltage across the output resistors and it's at an expected 4.3V. I have also measured the ac output voltage after the output caps, with an audio source I get roughly 0.2V.
I originally thought that I may have blown the TDA, but I have replaced it with two others and am getting exactly the same results.
I'm stuck, I don't have a clue where to go from here?
I'd not go beyond 8V which is the recommended maximum supply (although not the absolute max, which is 9V). Your volume control is a digital one? Try sending the DAC a 400Hz sinewave (generate the test waveform in Audacity) at full scale and look at the output on a scope (or at least measure with a multimeter if you have no scope).
I use a computer as a digital source via coax cable to my DAC, so i can reduce the volume/output range using windows volume control. I do usually though, leave this as full scale and control the volume with my amplifier.
I downloaded a sine wave generator app, and set to it 400Hz. I measured the output at different output levels (from windows) with my rather inaccurate (at small ac voltages) multimeter;
2% - 0.0V
3% - 0.2V
4% - 0.5V
5% - 0.8V
6% - 1.3V
7% - 1.4V Distortion starts.
Looking at this, I figured that the output is not as low as I first thought, I checked my amp and realised that it also wasn't turned up as much as I first thought. Comparing with the 5.3V peak to peak, 1.85V rms available (at the 8.3V supply), its not surprising that distortion is setting in when it is.
So the question is, are the threads I was previously using as a guide on the output and bias resistors correct and if not, how do I calculate the values for my resistors?
I downloaded a sine wave generator app, and set to it 400Hz. I measured the output at different output levels (from windows) with my rather inaccurate (at small ac voltages) multimeter;
2% - 0.0V
3% - 0.2V
4% - 0.5V
5% - 0.8V
6% - 1.3V
7% - 1.4V Distortion starts.
Looking at this, I figured that the output is not as low as I first thought, I checked my amp and realised that it also wasn't turned up as much as I first thought. Comparing with the 5.3V peak to peak, 1.85V rms available (at the 8.3V supply), its not surprising that distortion is setting in when it is.
So the question is, are the threads I was previously using as a guide on the output and bias resistors correct and if not, how do I calculate the values for my resistors?
Since TDA1543's full scale current is 2.3mA I reckon your I/V resistor is a little on the high side, giving 6.2V peak to peak. But the real problem looks to be that you have too much digital gain in your system - as 5% corresponds to 0.8V then 100% would work out to be 16V. Perhaps there's a setting somewhere in Windows or on your S/PDIF output card's driver which is putting in excess digital gain.
I believe to have traked down my problem to the DIR9001 reciver, spec here;
http://www.ti.com/lit/ds/symlink/dir9001.pdf
The output must have switched from 16 to 24 bit, which correlates with 2^8 / 2^12 = 6.25%
But what I can't figure out is that if you look on page 17 of the DIR9001 spec sheet, it only outputs I2S format in 24 bit anyway? This DAC used to work perfectly well, with full scale output before I lifted the TDA from the main board. I must be missing something, I just can't understand whats happening.
http://www.ti.com/lit/ds/symlink/dir9001.pdf
The output must have switched from 16 to 24 bit, which correlates with 2^8 / 2^12 = 6.25%
But what I can't figure out is that if you look on page 17 of the DIR9001 spec sheet, it only outputs I2S format in 24 bit anyway? This DAC used to work perfectly well, with full scale output before I lifted the TDA from the main board. I must be missing something, I just can't understand whats happening.
I thought it only output I2S in 32 bit, 64X fs for the bit clock. I've never seen an S/PDIF receiver output 32X fs or 48X fs in hardware mode. It seems the WM8805 is the only one that can do 32X fs and that's only in software mode. In any case the rate of the bitclock is ignored - the TDA1543 accepts just the 16 relevant bits.
The only way the I2S interface can result in digital gain is if the phase of the WS gets shifted.
The only way the I2S interface can result in digital gain is if the phase of the WS gets shifted.
I have read up on the I2S bus format, and now understand the truncation of the lower bits is not a problem. So this suggests the WS has somehow shifted four clock cycles?
I'm going to check the wiring I installed for the data bus to the new TDA board. Are there any faux pas that should be aware of, at the moment I'm just using three 1mm solid wires about 20mm long from the previous TDA socket straight up to the new board? Also, I wondered if the power supply to the new board may need a small bypass smoothing cap to help out with digital switching?
I'm going to check the wiring I installed for the data bus to the new TDA board. Are there any faux pas that should be aware of, at the moment I'm just using three 1mm solid wires about 20mm long from the previous TDA socket straight up to the new board? Also, I wondered if the power supply to the new board may need a small bypass smoothing cap to help out with digital switching?
Its definitely not the computer, I have plugged in another DAC and it plays fine!
I added a 100nF across the TDA supply that made no difference.
I checked the new connections, all below 0.2ohms and over 2M insulation (to ground and each other)
Now I'm out of ideas, I'll take all and any suggestions.
I added a 100nF across the TDA supply that made no difference.
I checked the new connections, all below 0.2ohms and over 2M insulation (to ground and each other)
Now I'm out of ideas, I'll take all and any suggestions.
See my website for my understanding of TDA1543 resistor values. You need much smaller values, even for an 8.3V supply. Try Rout 1.5K and Rref 1.1K, or Rout 1.2k and Rref 1K.
Reducing the supply voltage makes resistor values more critical, and generally smaller.
Reducing the supply voltage makes resistor values more critical, and generally smaller.
Hi Dave, I did see your site before I started building this project, but I couldn't tell how you arrived at your values! I used these two links to justify my choices as in the second, I can clearly follow the math;
http://www.diyaudio.com/forums/audio-sector/187748-pushing-limits-tda1543-nos-dac.html
and
http://www.diyaudio.com/forums/digi...nting-cd-pro-14.html?postid=173483#post173483
I'm really convinced though that my problem is a four cycle shift on the word select line causing a x16 digital gain. Again, I'll restate that I get quite clear and respectable output when the digital output is set at 6%. I have connected a second DAC and this has worked perfectly at 100% digital output.
I have also recently reduced the voltage on the TDA to 6.3V and replaced the output and bias resistors with 1K2 ones.
The distortion I get is not musical either, like if it was clipping. It’s a harsh noise attached to the louder parts of the music when it starts, and quickly descends into complete white noise.
http://www.diyaudio.com/forums/audio-sector/187748-pushing-limits-tda1543-nos-dac.html
and
http://www.diyaudio.com/forums/digi...nting-cd-pro-14.html?postid=173483#post173483
I'm really convinced though that my problem is a four cycle shift on the word select line causing a x16 digital gain. Again, I'll restate that I get quite clear and respectable output when the digital output is set at 6%. I have connected a second DAC and this has worked perfectly at 100% digital output.
I have also recently reduced the voltage on the TDA to 6.3V and replaced the output and bias resistors with 1K2 ones.
The distortion I get is not musical either, like if it was clipping. It’s a harsh noise attached to the louder parts of the music when it starts, and quickly descends into complete white noise.
In your second link Thorsten uses the same logic as I did, and so arrives at similar values. You have to balance up the signal current and the offset current so the voltage developed at the output does not get too close to either supply rail and cause clipping. This sets quite stringent limits on resistor values for a passive I/V with a grounded output resistor. Many people seem to get the values wrong, and don't seem to notice the peak clipping they get as a result. A well engineered CD will not stray near the limits too often anyway.
At 6V supply you must use 900-1K for Rref, and about 820 for Rout, if my sums are correct. However, the exact values may depend on manufacturing tolerances of the DAC chip so try up to 10% either way. If you are unlucky, the narrow window I show on my plot could disappear altogether so then you just have to choose between +ve peak clipping or -ve peak clipping or a bit of each.
If, as well, you have accidentally multiplied the data by 16 then this will make any clipping problem far worse. It may turn out that you have a different problem. Once you have solved it, you then may find clipping is the next problem.
At 6V supply you must use 900-1K for Rref, and about 820 for Rout, if my sums are correct. However, the exact values may depend on manufacturing tolerances of the DAC chip so try up to 10% either way. If you are unlucky, the narrow window I show on my plot could disappear altogether so then you just have to choose between +ve peak clipping or -ve peak clipping or a bit of each.
If, as well, you have accidentally multiplied the data by 16 then this will make any clipping problem far worse. It may turn out that you have a different problem. Once you have solved it, you then may find clipping is the next problem.
Well, I have given up on this. I can't figure out for the life of me whats going on!
I recently replaced the I2S link wires with screened audio interconnect, hoping that it was a ground path issue between the two boards, but to no avail.
My only and last thoughts are either I have damaged the DIR9001 or the different voltages (DIR runs on 3V) are effecting the digital signal. I don't really have the equipment to test the I2S signals so it looks like I'm going to have to put it back to the way it was to see if it still works as standard.
I recently replaced the I2S link wires with screened audio interconnect, hoping that it was a ground path issue between the two boards, but to no avail.
My only and last thoughts are either I have damaged the DIR9001 or the different voltages (DIR runs on 3V) are effecting the digital signal. I don't really have the equipment to test the I2S signals so it looks like I'm going to have to put it back to the way it was to see if it still works as standard.
I've bought a batch of 50 new tda1543 chips from hk. After trying four different (single chip dac) of them getting only "white noise" I googled around and found this post. Tried to lower the software volume feeding the usb to spdif and it seems like i got a similar clipping problem. Sound quality (after short listening test no burning in etc.) is OK but for some reason _much_ more sensitive. Going to experiment on breadboard to figure out if it's just crappy fake chips or if I/V resistor tweaking can help.
I've had success with the TDA1543 DAC having it feed into a voltage divider network of I/V resistors. As the DAC needs to source as well as sink current thru its Iout pin, this provides both abilities. The divider resistors effectively create this:
As most voltage regulator chips can only source current and not sink it (A fancy way to say that they can only suck electrons but not spit out electrons), a voltage divider resistor circuit does both and is really cheap. Select the ratio of resistors to provide about a little less than half the supply voltage (like 2.2V for 5V supply), and select the resistor ohmmage to create the Thevenon equivalent of the desired I/V load resistance.
An externally hosted image should be here but it was not working when we last tested it.
As most voltage regulator chips can only source current and not sink it (A fancy way to say that they can only suck electrons but not spit out electrons), a voltage divider resistor circuit does both and is really cheap. Select the ratio of resistors to provide about a little less than half the supply voltage (like 2.2V for 5V supply), and select the resistor ohmmage to create the Thevenon equivalent of the desired I/V load resistance.
I never got to the bottom of this, I put the TDA1543 back onto the original board with the original regs and it worked again as intended.
I then built another TDA1543 project from scratch, using stripboard and a CS8412 and it did exactly the same thing! I bought a 40MHz scope to try and track down the problem which confirmed a 4 bit shift on the ws line. For the life of me I couldn't fix it, I tried all sorts of modifications to the ground rails and decoupling caps figuring it was a ground plane problem but nothing worked.
Eventually I sidestepped the issue using a couple of 4-bit shift registers on the data line, I think I ended up using a 7 bit shift in the end as the registers seemed to add more delay to the ws line!
End of story, I got it to work with a half assed work around and closed up the project box. Sounds quite good and very open, but has issues with sibilance in the higher end that is a little fatiguing. I think that's just down to the TDA1543 in NOS (and no filtering), but it may be because I've introduced a shed load of jitter with the registers (or fixed a load with half-done dem reclocking!). I'm playing around and still learning with other projects now, I may return to the TDA1543 DACs in the future once I have a little more experience. I've been thinking about putting 8 together in a linear interpolation (using shift registers) topology with a filtered valve output stage.
Sorry I'm not much more help, do come back and let us know if you fix the problem.
Adam.
I then built another TDA1543 project from scratch, using stripboard and a CS8412 and it did exactly the same thing! I bought a 40MHz scope to try and track down the problem which confirmed a 4 bit shift on the ws line. For the life of me I couldn't fix it, I tried all sorts of modifications to the ground rails and decoupling caps figuring it was a ground plane problem but nothing worked.
Eventually I sidestepped the issue using a couple of 4-bit shift registers on the data line, I think I ended up using a 7 bit shift in the end as the registers seemed to add more delay to the ws line!
End of story, I got it to work with a half assed work around and closed up the project box. Sounds quite good and very open, but has issues with sibilance in the higher end that is a little fatiguing. I think that's just down to the TDA1543 in NOS (and no filtering), but it may be because I've introduced a shed load of jitter with the registers (or fixed a load with half-done dem reclocking!). I'm playing around and still learning with other projects now, I may return to the TDA1543 DACs in the future once I have a little more experience. I've been thinking about putting 8 together in a linear interpolation (using shift registers) topology with a filtered valve output stage.
Sorry I'm not much more help, do come back and let us know if you fix the problem.
Adam.
Thank you for responding.
First of all I'm relieved that I'm not imagining things and that this is a real and - as of yet - unexplained phenomenon regarding the TDA 1543. But as my experience with digital is very limited I'm afraid that we both need some help from experts in the field.
So if anyone read this - please help.
Vidar
First of all I'm relieved that I'm not imagining things and that this is a real and - as of yet - unexplained phenomenon regarding the TDA 1543. But as my experience with digital is very limited I'm afraid that we both need some help from experts in the field.
So if anyone read this - please help.
Vidar
I'm as puzzled as everyone else is here.
Let me summarize what it seems to me are the symptoms.
Adam has a known working TDA1543 chip which when transplanted into another board with a DIR9001 receiver gives 24dB of digital gain. The same chip in yet another implementation with a CS8412 receiver also gives the same digital gain. So Adam - what's the digital receiver in your first DAC?
The scope confirms that the CS8412 receiver puts out a WS which has been shifted by 4 clocks. But even more puzzling, Adam needs an additional 3 BCK delays to get this DAC working.
We have two further posters with what look to be similar problems, Henry and Vidar. Henry doesn't say what his receiver is but Vidar's using neither DIR9001 nor CS8412.
On the surface this looks to be a problem with the receiver as Adam noticed the WS shifted on his scope, yet with at least 3 different I2S sources giving the same problem, a receiver problem looks improbable.
Let me summarize what it seems to me are the symptoms.
Adam has a known working TDA1543 chip which when transplanted into another board with a DIR9001 receiver gives 24dB of digital gain. The same chip in yet another implementation with a CS8412 receiver also gives the same digital gain. So Adam - what's the digital receiver in your first DAC?
The scope confirms that the CS8412 receiver puts out a WS which has been shifted by 4 clocks. But even more puzzling, Adam needs an additional 3 BCK delays to get this DAC working.
We have two further posters with what look to be similar problems, Henry and Vidar. Henry doesn't say what his receiver is but Vidar's using neither DIR9001 nor CS8412.
On the surface this looks to be a problem with the receiver as Adam noticed the WS shifted on his scope, yet with at least 3 different I2S sources giving the same problem, a receiver problem looks improbable.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.