It's a neat idea, and something I've been wanting to make for a long time, but I've never got around to it. On this site someone has made another discrete DAC and posted some results.
A couple of thought of mine on this topic:
- Matching the resistors in the resistor ladder is critical. I would get an accurate multimeter and a whole or part reel of 1206 size 1% resistors of eBay. Use one value of resistor for the whole ladder (one resistor for the 2R part, two in parallel for the R part).
- I would use a discrete, very low Ron CMOS buffer between the logic and the resistor ladder. I would look at using something like IR's IRF7307. This reduces the impedance (and more importantly, the variance in that impedance) of the feed to the resistor ladder by two orders of magnitude.
A couple of thought of mine on this topic:
- Matching the resistors in the resistor ladder is critical. I would get an accurate multimeter and a whole or part reel of 1206 size 1% resistors of eBay. Use one value of resistor for the whole ladder (one resistor for the 2R part, two in parallel for the R part).
- I would use a discrete, very low Ron CMOS buffer between the logic and the resistor ladder. I would look at using something like IR's IRF7307. This reduces the impedance (and more importantly, the variance in that impedance) of the feed to the resistor ladder by two orders of magnitude.
I experimented with a discrete DAC some time ago. Here is some desription (sorry, it is Hungarian only, but there are some pictures and a circuit diagram at the end):
DDAC project
DDAC project
IMO, you are much better off implementing a solution based on one of the many DAC chips from T.I., Analog Devices. or Linear Technology which are intended for high-performance instrument grade waveform-synthesis. While not directly intended for the audio market, a number of these DACs feature excellent A.C., datasheet documented, specs. Digital audio actually is a class of digital waveform-synthesis, if you think about it. MOst of these instrument market DACs use either R2R Ladder or Resistor- String conversion technology.
Their application to audio is not as straight forward as it is with audio purposed DACs. Some of them have serial data inputs while others have parallel. Some accept a 2s complement input data format while others require unsigned straight binary and would need 2s complement formatting logic. In short, you will have to implement you own input data/frame formatting interface logic. In addition, most are what's called multiplying DACs, or M-DACs for short. Which means that they require a reference voltage that can be either D.C. (what our familiar HiFi audio converters use internally), or an A.C. signal which is multiplied with the current data sample input to the DAC so to produce an output which is the product of the two. However, if any of that were a serious concern we wouldn't be considering building our own discrete diy DAC in the first place. would we!?
In any case, while not as simple to use as are actual audio DAC chips, M-DACs are far easier to implement than a discrete diy converter, will have better A.C. specs (THD, SFDR, SNR, FS), DNL, INL, and lower glitch-energy. Best of all they all utilize some form of resistor and FET-switch based converter technology for up through 18-bits and are in active production. I'm not aware of any audio market active production DACs that are not Sigma-Delta, with the exception of the PCM1704. The PCM1704 had been discontinued but was placed back in to active production reportedly because of demand from the instrument waveform-synthesis market. How ironic.
Their application to audio is not as straight forward as it is with audio purposed DACs. Some of them have serial data inputs while others have parallel. Some accept a 2s complement input data format while others require unsigned straight binary and would need 2s complement formatting logic. In short, you will have to implement you own input data/frame formatting interface logic. In addition, most are what's called multiplying DACs, or M-DACs for short. Which means that they require a reference voltage that can be either D.C. (what our familiar HiFi audio converters use internally), or an A.C. signal which is multiplied with the current data sample input to the DAC so to produce an output which is the product of the two. However, if any of that were a serious concern we wouldn't be considering building our own discrete diy DAC in the first place. would we!?
In any case, while not as simple to use as are actual audio DAC chips, M-DACs are far easier to implement than a discrete diy converter, will have better A.C. specs (THD, SFDR, SNR, FS), DNL, INL, and lower glitch-energy. Best of all they all utilize some form of resistor and FET-switch based converter technology for up through 18-bits and are in active production. I'm not aware of any audio market active production DACs that are not Sigma-Delta, with the exception of the PCM1704. The PCM1704 had been discontinued but was placed back in to active production reportedly because of demand from the instrument waveform-synthesis market. How ironic.
I am a discrete R2R DAC lover, there are really musical and it is possible to experiment them for very low cost. Even when 1% resistors are used the sound is already good!
Take a CS8412 ot CS8416, some cheap logic chips and 100 1% resistors and that makes a good DAC, better than a conventional midle cost DAC.
Take a CS8412 ot CS8416, some cheap logic chips and 100 1% resistors and that makes a good DAC, better than a conventional midle cost DAC.
I have been thinking of making a discrete ladder DAC myself, instead of off the shelf. This would be my first time to build a DAC and I hope to learn from the process and then maybe in time progress onto a discrete I.C. 'off-the-shelf' type. I was wondering how complex it would be beginning to end to make this ladder DAC as the schematics I have seen in the past have not told the full story.
Can anyone give me any usefull information or even where to go to do this. I have a rake of 1% resistors ready for the purpose but wanted to learn more about the full process first. How the waveform feeds into the ladder then feeds out etc etc etc
Thanks
Can anyone give me any usefull information or even where to go to do this. I have a rake of 1% resistors ready for the purpose but wanted to learn more about the full process first. How the waveform feeds into the ladder then feeds out etc etc etc
Thanks
You can use most part of my discrete DAC, see above. Best method is to convert from I2S to offset binary. Old 14-bit Philips CD players have the SAA7000 chip that can do this conversion. Then you need a serial-in/parallel-out shift register and precision switches + resistors. 1% is good only for proof of concept testing. 2R-R ladder is voltage output, you need a high input impedance buffer and that's all. But don't expect more than 12 bit resolution :-( It's for fun after all...
O.K. thanks. If I want true 16 or 24 bit then it will have to be purpose built I.C.'s then ? O.K., cool.
I spent a long time developping a non-compromise R2R DAC, but sorry it is a commercial link now:
TOTALDAC DAC and digital crossover
TOTALDAC DAC and digital crossover
Thanks guys,
Well, Marce, I don't know about where I could get laser trimming done, and vincent_brient....WOW, you built that ? It looks trés fantastique but above my capabilities at the moment which is why I would like to try something fairly simple, if a DAC could be simple, so I can learn the basics and then work on. I haven't built too many of my pieces of Hi-Fi yet but am more than willing to learn.
I feel a bit stuck now and unsure of what to do....suggestions ???
Thanks a lot
Gareth
Well, Marce, I don't know about where I could get laser trimming done, and vincent_brient....WOW, you built that ? It looks trés fantastique but above my capabilities at the moment which is why I would like to try something fairly simple, if a DAC could be simple, so I can learn the basics and then work on. I haven't built too many of my pieces of Hi-Fi yet but am more than willing to learn.
I feel a bit stuck now and unsure of what to do....suggestions ???
Thanks a lot
Gareth
O.K. I am dtermined to build a ladder-DAC now. After seeing the schematic in an electronics book I have I have always wanted to go down that route. The book is by Ray Marsdon and I am sure that he says you can build up to 16 bit, I will have to check that though.
vincent-brient's pictures have provided the extra inspiration needed !!! I have SMD resistors, quite a lot of spares, cap's trim pot's etc etc, also I have a DAC board that uses Pioneer DIP's on it (14 or 16 pin - not too sure off the top of my head). This is from the very late eighties or early nineties, used in an AKAI DAC. I am wondering if these can be brought back to use with my project - need to find out the part/model numbers though. Then take it from there. Would it be tto much to ask for help as I go along with this ? This is my first attempt and although I have an electrical background it is not in electronics so you may have to bear with me a bit. Is this O.K. ? Any thoughts ?
Thanks
Gareth
vincent-brient's pictures have provided the extra inspiration needed !!! I have SMD resistors, quite a lot of spares, cap's trim pot's etc etc, also I have a DAC board that uses Pioneer DIP's on it (14 or 16 pin - not too sure off the top of my head). This is from the very late eighties or early nineties, used in an AKAI DAC. I am wondering if these can be brought back to use with my project - need to find out the part/model numbers though. Then take it from there. Would it be tto much to ask for help as I go along with this ? This is my first attempt and although I have an electrical background it is not in electronics so you may have to bear with me a bit. Is this O.K. ? Any thoughts ?
Thanks
Gareth
Last edited:
I'm guessing its in voltage out mode with the logic as switch element coupled between two reference voltages?
I guess there is a compromise to be found between accuracy loss from the switches and the noise floor.
Are you using some deglitching techniques ?
I understand I'm just curious what hurdles there are to doing such a project
If I'm ever going to build an R2R dac I would do it with a small cpld per channel to split up the data from an i2s stream and drive the switching elements, and if necessary generate the control pulses for a S/H amplifier.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.