Re: Re: Re: Re: Re: the digital circuit
Konnichiwa,
I am quite familiar with Lavry, since the days of dbtech....
Yes and most of this contributes only to the long term stability of absolute levels, not oif relative levels. It is a rather complex way of doing things, no doubt apropriate in a high end pro-audio unit but strictly speaking unneccesary for good sound, IMHO.
But I don't think this discussion focuses on making a DAC that "measures good" in the first instance. I believe we all would be happy with "good enough".
Sayonara
Konnichiwa,
Terry_Demol said:
I am quite familiar with Lavry, since the days of dbtech....
Terry_Demol said:
Yes and most of this contributes only to the long term stability of absolute levels, not oif relative levels. It is a rather complex way of doing things, no doubt apropriate in a high end pro-audio unit but strictly speaking unneccesary for good sound, IMHO.
Terry_Demol said:
But I don't think this discussion focuses on making a DAC that "measures good" in the first instance. I believe we all would be happy with "good enough".
Sayonara
I think Caddock sells some resistors with better than 5 ppm TC. However, their 50 ppm ones are expensive enough (up to $17 for MK132 from one retailer). I think that Roederstein MK3 are also 50 ppm, and they are only a quarter each (Percyaudio), but getting hard to find all values as they are no longer manufactured.
Taking the "minimum number of gain stages" fad to an extreme, some time ago I was daydreaming of a 4 kV discrete R2R power DAC that would drive ESLs directly... I'm crazy enough to try it, but not rich enough... 😀Kuei Yang Wang said:
Konnichiwa,
Meggit makes SMD resistors with +/-10ppm tempco & 0.1% tolerance for around a buck each and Welwyn makes leaded resistors with +/-15ppm tempco & 0.1% tolerance for the same, endcustomer retail from an expensive source and < 10pcs qty.
I suspect if one where to manufacture 100pcs stereo balanced R2R Dac's with the Meggit SMD's (needs in total around 10,000pcs R and 10,000pcs 2R) one could achieve MUCH BETTER pricing.
Sayonara
Prune said:
Meggit makes SMD resistors with +/-10ppm tempco & 0.1% tolerance for around a buck each and Welwyn makes leaded resistors with +/-15ppm tempco & 0.1% tolerance for the same, endcustomer retail from an expensive source and < 10pcs qty.
I suspect if one where to manufacture 100pcs stereo balanced R2R Dac's with the Meggit SMD's (needs in total around 10,000pcs R and 10,000pcs 2R) one could achieve MUCH BETTER pricing.
Sayonara
Re: Re: Re: Re: Re: Re: the digital circuit
Why would absolute levels be important past say 0.05dB
in a pro audio application, which is way above DAC linearity
levels?
The temperature control is for *relative* levels, or do you
assume all resistors drift the same for changes in temp?
My aim WRT using Lavry as example was to put some perspective
in what to expect WRT acheivable DAC linearity.
I must admit that I am somewhat confused about your
design criterion here. What are the design objectives of a discrete
DAC? If you are not concerned about DAC linearity what
parameters are you chasing?
Or is it merely an excercise of "lets just see if we can do it for the challenge" which is also OK.
Cheers,
Terry
Kuei Yang Wang said:
Why would absolute levels be important past say 0.05dB
in a pro audio application, which is way above DAC linearity
levels?
The temperature control is for *relative* levels, or do you
assume all resistors drift the same for changes in temp?
My aim WRT using Lavry as example was to put some perspective
in what to expect WRT acheivable DAC linearity.
I must admit that I am somewhat confused about your
design criterion here. What are the design objectives of a discrete
DAC? If you are not concerned about DAC linearity what
parameters are you chasing?
Or is it merely an excercise of "lets just see if we can do it for the challenge" which is also OK.
Cheers,
Terry
temp controle
If the dac is made temperature controlled, resistors with higher drift can be chosen, which would have a good influence on the price 🙂
The drift drift in opamp etc. is also less important, so I think a temperature controled dac would be the cheapest.
But if the room temperature is constant, the temperature drift would be minor. If the adjustment is done after the warmup period, the perfomance could be fine with medium cost components. Kuei Yang Wang also mentiod this earlier. 🙂
If the dac is made temperature controlled, resistors with higher drift can be chosen, which would have a good influence on the price 🙂
The drift drift in opamp etc. is also less important, so I think a temperature controled dac would be the cheapest.
But if the room temperature is constant, the temperature drift would be minor. If the adjustment is done after the warmup period, the perfomance could be fine with medium cost components. Kuei Yang Wang also mentiod this earlier. 🙂
Bernhard said:
Is this the schematic ? http://www.diyaudio.com/forums/attachment.php?s=&postid=508263&stamp=1099947368
There's no I/V resistors in the schematic I made. The resistors is part of the constant current sources. The value of the resistor dertermines the current flow.
Best regards Lennart
Bernhard said:
Yes the 2R will work a an I/V resistor.
And Yes it has a unipolar zero, if you want to make it bipolar, the switch should should toggle between a positive and negative reference..
BTW in this schematic no reference is used, this will lead to a bad performance
Lgrau said:
BTW in this schematic no reference is used, this will lead to a bad performance
What best use as reference, REF02 or TL431 or ?
Bernhard said:
A better choise would be a max6250:
http://pdfserv.maxim-ic.com/en/ds/MAX6225-MAX6250.pdf
or even better a jung regulator.
Jitter problems
I like to hear about the filter causing, jitter. Do you have a link or maybe like to write a few words about it ?
Is it the filter in the receiver, you're talking about ?
I meant that the jitter added by the dac itself could be less.
The dac should be made as less sensitive to incomming jitter as possible.
I like to hear about the filter causing, jitter. Do you have a link or maybe like to write a few words about it ?
Is it the filter in the receiver, you're talking about ?
I meant that the jitter added by the dac itself could be less.
The dac should be made as less sensitive to incomming jitter as possible.
Any recommendation for a - logic programmable gain (1 & -1 ) - op amp / buffer ?
Something with a control pin that changes from non-inverting to inverting.
Something with a control pin that changes from non-inverting to inverting.
Bernhard said:
If you are thinking of a solution for a R2R dac, you can just do it as shown in the first figure on this page :
http://www.maxim-ic.com/appnotes.cfm/appnote_number/583
to Carlosfm : looking forward to you answer, but just take you the time you need 😉
Lgrau said:
This is about deglitching. ?
What I look for is to avoid analog cmos switches, when switching between Vref+ and Vref-.
Bernhard said:
Ok, but i don't think it's possible to make a logic, which can switch as fast and precise, as a switch solution can do.
But it is always interesting to try to go other ways around the problem.
I still mean that the switch is the most critical part of the design, so a better alternative solution would be very nice.
If I decide to use the current switch DAC there is no need for a programmable gain solution. 😉
Re: Jitter problems
Jocko was talking about a post I made somewhere, discussing non-oversampling.
I said that everytime I bypass a digital filter I hear improvements, and most of the times I get tighter bass.
This means lower jitter.
Of course this may be more noticeable with some chips.
There are dacs that are more sensitive to jitter than others, but they usually have one thing in common: jitter rejection starts at high frequencies, so much so that talking about jitter rejection is pure marketing.
Your project is ambitious, good luck.
Forget digital filter/oversampling for now, it will also simplify the design.😀
Lgrau said:
Jocko was talking about a post I made somewhere, discussing non-oversampling.
I said that everytime I bypass a digital filter I hear improvements, and most of the times I get tighter bass.
This means lower jitter.
Of course this may be more noticeable with some chips.
There are dacs that are more sensitive to jitter than others, but they usually have one thing in common: jitter rejection starts at high frequencies, so much so that talking about jitter rejection is pure marketing.
Your project is ambitious, good luck.
Forget digital filter/oversampling for now, it will also simplify the design.😀
Re: Re: Jitter problems
Thank you for your for sharing your expiriences.
I certianly will start with a DAC without oversampling, there is no need to make it more complicated than it is necessary 🙂
BTW I'm thinking about splitting the MSB bits up into more less valued bits.
e.g. Making 2 *½ MSB instead of 1MSB, this would make it easier to make a DAC with a high resulution, working monolotich the entire range. But the logic Will be a bit more complicated.
It could be a help if it is hard to get the wanted resolution, but I don't think it's nescesary if it "just" should deliver a true 16bit output.
carlosfm said:
Thank you for your for sharing your expiriences.
I certianly will start with a DAC without oversampling, there is no need to make it more complicated than it is necessary 🙂
BTW I'm thinking about splitting the MSB bits up into more less valued bits.
e.g. Making 2 *½ MSB instead of 1MSB, this would make it easier to make a DAC with a high resulution, working monolotich the entire range. But the logic Will be a bit more complicated.
It could be a help if it is hard to get the wanted resolution, but I don't think it's nescesary if it "just" should deliver a true 16bit output.
- Status
- Not open for further replies.