Hi,
If I do this it will not be a public design, as almost all of the design tricks that will be used are not ones I wish to release into the public domain. And it would be "by invitation only" as I have no desire to deal with all the troubles of the usual group buys.
Ciao T
If I do this it will not be a public design, as almost all of the design tricks that will be used are not ones I wish to release into the public domain. And it would be "by invitation only" as I have no desire to deal with all the troubles of the usual group buys.
Ciao T
Agree, that is not a special skill/secret. There are also tutorials on line:
http://www.analog.com/static/imported-files/tutorials/MT-101.pdf
www.analog.com/static/imported-files/tutorials/MT-043.pdf
Last edited:
Hi,
True.
It is quite simple and mostly even amenable to fairly reliable spice simulation, if you take care to account for all the parasitics (traces, groundplane).
BTW, the two "tutorial's" you posted are incomplete and they do not show the best options, so clearly there room for improvement for even those who apply what they learn from these tutorials.
One thing I do observe however is that for both projects posted here, chinese kits sold on e-pay and much of serious high end gear as well as low end gear designers get it badly wrong.
About the only place I see correct decoupling these days are computer motherboards and there mainly because intel designs the critical part of the decoupling in into their processors and you have to get the rest rest right or the board will not work properly...
Meanwhile I have no problem with people insisting that what I do is common, simple and widely implemented in Audio, as at least we have a contention we can (trivially easily) falsify.
Ciao T
True.
It is quite simple and mostly even amenable to fairly reliable spice simulation, if you take care to account for all the parasitics (traces, groundplane).
BTW, the two "tutorial's" you posted are incomplete and they do not show the best options, so clearly there room for improvement for even those who apply what they learn from these tutorials.
One thing I do observe however is that for both projects posted here, chinese kits sold on e-pay and much of serious high end gear as well as low end gear designers get it badly wrong.
About the only place I see correct decoupling these days are computer motherboards and there mainly because intel designs the critical part of the decoupling in into their processors and you have to get the rest rest right or the board will not work properly...
Meanwhile I have no problem with people insisting that what I do is common, simple and widely implemented in Audio, as at least we have a contention we can (trivially easily) falsify.
Ciao T
I didn't post those tutorials, AD engineers did, a long time ago. Sure, they are just "decoupling 101" (altough nothing wrong in what they show) but I see here lots of folks that don't know even that.
I was under impression (like Pano) that you stated that the arrangement of decoupling caps is some kind of a secret, sorry if I got it worng.
Also, I know that you are vehement against OpAmps because they are "slow".
Explain how is that working for TDA1541 that has settling time of 1 us (one micro second), compared with:
LM4562 = 1.2us
OPA2134 = 1us
OP627 = 0.55 us
AD797 = 0.8 us
AD8651 = 0.2 us
I was under impression (like Pano) that you stated that the arrangement of decoupling caps is some kind of a secret, sorry if I got it worng.
Also, I know that you are vehement against OpAmps because they are "slow".
Explain how is that working for TDA1541 that has settling time of 1 us (one micro second), compared with:
LM4562 = 1.2us
OPA2134 = 1us
OP627 = 0.55 us
AD797 = 0.8 us
AD8651 = 0.2 us
Last edited:
Hi,
First, not all DAC's specify settling to 1LSB, not for a 1LSB step and these times are maximum (usually for a step from 0 to 65536) which would be on a 2V RMS standard output a 5.6V Step.
Second, Op-Amp specifications for settling time tend to vary widely, some specify 0.1% settling, some 0.01%, some both and some specify unity gain and a 10V step, others specify different gains and voltage steps. Sometimes a comparison of large signal and small signal step response is also illustrated...
So not only are the numbers you posted for Op-Amp's not directly comparable with each other, they also cannot be compared directly to converter settling time.
Ciao T
First, not all DAC's specify settling to 1LSB, not for a 1LSB step and these times are maximum (usually for a step from 0 to 65536) which would be on a 2V RMS standard output a 5.6V Step.
Second, Op-Amp specifications for settling time tend to vary widely, some specify 0.1% settling, some 0.01%, some both and some specify unity gain and a 10V step, others specify different gains and voltage steps. Sometimes a comparison of large signal and small signal step response is also illustrated...
So not only are the numbers you posted for Op-Amp's not directly comparable with each other, they also cannot be compared directly to converter settling time.
Ciao T
I choose the times at 0.01% for OpAmps and the time for the TDA is at 1LSB - like I said.
For full swing I am sure is worse (albeit not published by Philips, I gave the PCM comparation, because BB cared to show both).
What matters in the end is Slew Rate, and again, albeit Philips didn't care to publish their SR, comparation with BB shows a SR of 12V/us for that generation of DAC's. Below 20V/us (or higher) of modern OpAmps.
Sure, I see where the legend started - with OpMAps used in '90s having SR of 3-9V/us (LM833, NE5532)...
For full swing I am sure is worse (albeit not published by Philips, I gave the PCM comparation, because BB cared to show both).
What matters in the end is Slew Rate, and again, albeit Philips didn't care to publish their SR, comparation with BB shows a SR of 12V/us for that generation of DAC's. Below 20V/us (or higher) of modern OpAmps.
Sure, I see where the legend started - with OpMAps used in '90s having SR of 3-9V/us (LM833, NE5532)...
Last edited:
Hi,
That is patently untrue.
I have for example recommended the OPA637 (settling to 0.01% in 0.45uS for a 10V step) as I/V converter for TDA1541A DAC's and used them also as such.
Note, the OPA637 requires some circuit changes to work in normal TDA1541, without these it will oscillate, so no point running out and getting them. Plus, I feel that other options sound better.
But many Op-Amp's are already sonically challenged by normal signals, what they face with IV converter service, especially with oversampled signals is by far more challenging, especially given that the apparent good linearity they have at low frequencies rapidly deteriorates at higher frequencies.
The first image for a 1KHz full scale signal with a 4 * Oversampling system (standard TDA1541 use) will fall at 175.4KHz and have nearly the same level as the actual signal and the next one up will be at around 700KHz and still quite high in level. There are reasons why they do not use Audio Op-Amp's for RF service.
Ciao T
That is patently untrue.
I have for example recommended the OPA637 (settling to 0.01% in 0.45uS for a 10V step) as I/V converter for TDA1541A DAC's and used them also as such.
Note, the OPA637 requires some circuit changes to work in normal TDA1541, without these it will oscillate, so no point running out and getting them. Plus, I feel that other options sound better.
But many Op-Amp's are already sonically challenged by normal signals, what they face with IV converter service, especially with oversampled signals is by far more challenging, especially given that the apparent good linearity they have at low frequencies rapidly deteriorates at higher frequencies.
The first image for a 1KHz full scale signal with a 4 * Oversampling system (standard TDA1541 use) will fall at 175.4KHz and have nearly the same level as the actual signal and the next one up will be at around 700KHz and still quite high in level. There are reasons why they do not use Audio Op-Amp's for RF service.
Ciao T
Hi,
Actually the datasheet very clearly says "Current settling time TO +/-1LSB - 0.5uS <emphasis by me>" in the TDA1541A one I am looking at (Feb 1991). While the "from" is not specified it is normally a full traverse of the output from silence, or in the case of the TDA1541 a 2mA step.
When BB are talking "Voltage Output" they are talking about the behaviour of the build in Op-Amp plus DAC, not that of the DAC alone (see block diagram), for the converter alone a 1mA (= 1MSB) Step has 0.25uS settling time to to ±0.006% of FSR according to the datasheet. So you can see the build in Op-Amp of the DAC, if used degrades settling time sixfold...
You do not know, because no slew rate is given for the output current.
I'll do the measurement for the current rise time (30% - 70%) for a real TDA1541A when I get time.
I can tell you that using a 150MHz 'scope and running the DAC at quad speed (~5uS samples) you cannot see any rounding off of the steps at sensible zoom levels with a purely resistive I/V. And even that, by producing more outvoltage than the nominal 50mV compliance range will slow things down...
In the 90's we used OPA(2)604 which was 25V/uS and OPA627/637 (55/135V/uS) plus stuff like AD811 (2500V/uS) and LM6181 (1400V/uS). Well, i don't know the "we", certainly I DID.
The pedestrian stuff you mention above more properly belongs to the 80's.
As for the Legend, what Legend? I don't see any.
Ciao T
Actually the datasheet very clearly says "Current settling time TO +/-1LSB - 0.5uS <emphasis by me>" in the TDA1541A one I am looking at (Feb 1991). While the "from" is not specified it is normally a full traverse of the output from silence, or in the case of the TDA1541 a 2mA step.
When BB are talking "Voltage Output" they are talking about the behaviour of the build in Op-Amp plus DAC, not that of the DAC alone (see block diagram), for the converter alone a 1mA (= 1MSB) Step has 0.25uS settling time to to ±0.006% of FSR according to the datasheet. So you can see the build in Op-Amp of the DAC, if used degrades settling time sixfold...
You do not know, because no slew rate is given for the output current.
I'll do the measurement for the current rise time (30% - 70%) for a real TDA1541A when I get time.
I can tell you that using a 150MHz 'scope and running the DAC at quad speed (~5uS samples) you cannot see any rounding off of the steps at sensible zoom levels with a purely resistive I/V. And even that, by producing more outvoltage than the nominal 50mV compliance range will slow things down...
In the 90's we used OPA(2)604 which was 25V/uS and OPA627/637 (55/135V/uS) plus stuff like AD811 (2500V/uS) and LM6181 (1400V/uS). Well, i don't know the "we", certainly I DID.
The pedestrian stuff you mention above more properly belongs to the 80's.
As for the Legend, what Legend? I don't see any.
Ciao T
Back then, I was working in a metrology lab (time/frequency), so I had "toys" around me to measure. My hearing was better than now also.
And yes, now I am eyballing the datasheets too...
I still have to hear a TDA1541 to sound above "well". SAA7220 doesn't do good in that direction, NOS is absolutelly horrid, tubes are noisy and distort (in a pleasant way, but still)...
PS: Looks like you put a lot of trust in "hearing". I didn't see your reply when it was that post here about listening two files posted here and saying which is the "uncompressed one". I did say public my oppinion then, and I was right. Did upset the OP because I did reveal the good answer (instead of PM like he wanted). And, even if I still hear distortions well, I trust measurements more.
And yes, now I am eyballing the datasheets too...
I still have to hear a TDA1541 to sound above "well". SAA7220 doesn't do good in that direction, NOS is absolutelly horrid, tubes are noisy and distort (in a pleasant way, but still)...
PS: Looks like you put a lot of trust in "hearing". I didn't see your reply when it was that post here about listening two files posted here and saying which is the "uncompressed one". I did say public my oppinion then, and I was right. Did upset the OP because I did reveal the good answer (instead of PM like he wanted). And, even if I still hear distortions well, I trust measurements more.
Last edited:
Hi,
So, you know any audio Op-Amp's that do better at 2V peak-Peak into 100 Ohm at 10MHz? The AD811 for example at 2V P-P into 100 Ohm and at 30KHz is -110dB (0.0003%) 2nd HD and 3rd HD is at around -120dB (0.0001%) and remember, this is with 100 ohm load.
Again, I see few if any Audio Op-Amp's that compare well.
The LM4562 is shown as -105dB THD at 20KHz with 3V into 600 Ohm. For this the AD811 will be AT LEAST 6-10dB better.
The OPA134 offers 0.002% 2nd HD at 1V RMS into 600 Ohm, by 30KHz and if we reduce the load for the AD811 to 600 Ohm as well We will be looking at least at 30dB worse HD for OPA134.
The OPA627 does offer a 20KHz THD into 600 Ohm that is broadly comparable with that of the AD811 but the funny part is that it is actually emphatically NOT an Audio Op-Amp (though I like to use it for that job) and the AD797 driving 600 Ohm too offers a broadly comparable distortion performance to the AD811 driving 100 Ohm to 0.7V.
The AD8651 you mention is actually listed as video line driver, but with 0.0006% THD at 0.7 RMS into 600 Ohm it really compares to the AD811.
So, I am not sure what your purpose was with that little quip, however the facts do not bear out your contention in the least.
Ciao T
So, you know any audio Op-Amp's that do better at 2V peak-Peak into 100 Ohm at 10MHz? The AD811 for example at 2V P-P into 100 Ohm and at 30KHz is -110dB (0.0003%) 2nd HD and 3rd HD is at around -120dB (0.0001%) and remember, this is with 100 ohm load.
Again, I see few if any Audio Op-Amp's that compare well.
The LM4562 is shown as -105dB THD at 20KHz with 3V into 600 Ohm. For this the AD811 will be AT LEAST 6-10dB better.
The OPA134 offers 0.002% 2nd HD at 1V RMS into 600 Ohm, by 30KHz and if we reduce the load for the AD811 to 600 Ohm as well We will be looking at least at 30dB worse HD for OPA134.
The OPA627 does offer a 20KHz THD into 600 Ohm that is broadly comparable with that of the AD811 but the funny part is that it is actually emphatically NOT an Audio Op-Amp (though I like to use it for that job) and the AD797 driving 600 Ohm too offers a broadly comparable distortion performance to the AD811 driving 100 Ohm to 0.7V.
The AD8651 you mention is actually listed as video line driver, but with 0.0006% THD at 0.7 RMS into 600 Ohm it really compares to the AD811.
So, I am not sure what your purpose was with that little quip, however the facts do not bear out your contention in the least.
Ciao T
I think that you feel that way because you have only heard circuits that don't do a good enough job of removing the ground noise which is generated by the current draw of the TDA1541, from the output stage power supply.
I'm not talking about aliasing, or the low pass filter, that is another issue.
All TDA1541 kits are plagued by this issue which causes an audible effect, which some would assume is aliasing, but actually is not. It is ground noise.
Glad I found out how to solve it.
Last edited:
No, I trust my hearing but its not infallible. No senses are. So I cross check.
That would be because I was not interested in replying
Which measurements do you trust? Relative humidity?
I don't deny that I didn't count that in... but I doubt that not-filtered aliasing products had nothing to do with the disturbing noises that I heard.
Too bad we cannot test your NOS against some of my OS DAC's
I was not interested in the test, but I wasn't scared to say what I hear. Of course if I was wrong, you would remaind me thatThat would be because I was not interested in replying
@thorsten:
PCM61 is rated for 8x OS and, to be able to do that, has CURRENT settling time listed as 350ns for 1mA step (has a max +/- 1mA range).
PCM63 is rated for 16x OS and, to be able to do that, has CURRENT settling times listed as 200ns for a 2mA step (max swing +/- 2mA).
You want to convince me that TDA1541 with max 4x OS capabilities will swing faster the outputs than the BB DAC's (rated at 8 and 16x) and the Phillips datasheet is wrong when it says the settling time is 1 micro second?
Well, to give you credit for something, I become curious to test the new, fast, low noise ADA4897-1 in my PCM61 DAC. I need an adaptor single-dual now...
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.