I'm giving up my woodworking endeavors due to not having a stable workspace. People don't like sawdust.
My background is machinist and welder and long ago graphic designer. I enjoy taking on projects outside my realm of understanding.
I want to build some electronics from the ground up. My first real project will be either a preamp or a spl meter. The first question I have is about intellectual property.
Are some circuit designs so common that I could use them without infringing on someones intellectual property? If not, would understanding how digital information is converted to a signal be a good starting point, or are there chips that convert data and its a waste of time to learn how they do it? I suppose converting data to a signal must be extremely complicated.
If there is a book that would help get me started on this path please share. The closest I've got to doing electronics work is building some subwoofers. If you think what I want to do is not feasible without formal training I would appreciate you saying so. I would enjoy this type of work because I like problem solving.
My background is machinist and welder and long ago graphic designer. I enjoy taking on projects outside my realm of understanding.
I want to build some electronics from the ground up. My first real project will be either a preamp or a spl meter. The first question I have is about intellectual property.
Are some circuit designs so common that I could use them without infringing on someones intellectual property? If not, would understanding how digital information is converted to a signal be a good starting point, or are there chips that convert data and its a waste of time to learn how they do it? I suppose converting data to a signal must be extremely complicated.
If there is a book that would help get me started on this path please share. The closest I've got to doing electronics work is building some subwoofers. If you think what I want to do is not feasible without formal training I would appreciate you saying so. I would enjoy this type of work because I like problem solving.
From what I’ve gathered here, using others designs for personal use is ok, just not for profit.
Lots of research with this hobby, read, lots of time spent not moving around much versus traditional shop activities, so there are some health considerations if you tend to dive in head first perhaps.
A preamp would be a fun first project I would imagine.
Lots of research with this hobby, read, lots of time spent not moving around much versus traditional shop activities, so there are some health considerations if you tend to dive in head first perhaps.
A preamp would be a fun first project I would imagine.
Hi Bret, I see you just joined diyaudio last month, so welcome!
Do you just want to build electronics or to design and build electronics? Designing requires more knowledge than building someone else's design, although in both cases it helps to understand what you are doing.
If your first project will be a preamp or an SPL meter, why do you want to learn about digital to analog converters?
Converting a digital signal to an analog signal can be very simple when the accuracy requirements are not too high, see for example Resistor ladder - Wikipedia and scroll to the part about R-2R ladders. The problem with audio DACs is that the accuracy requirements usually lead to far more complicated circuits. That is, either the DAC itself gets complicated or you need a large amount of digital preprocessing to get away with a simple DAC.
An exception occurs when you only play so-called direct stream digital files. That format allows the use of quite simple DAC circuits without any digital preprocessing if you are not after extreme signal-to-noise ratios. See the "The best DAC is no DAC" thread.
Most DIY audio DAC builders just make an application circuit for an audio DAC chip, so most of the complexity is hidden inside the chip. There are some exceptions on this forum, like the Soekris circuits, the "no DAC", the Signalyst designs, xx3stksm's designs (he uses DAC chips, but in a very non-standard way) and my valve DAC.
About intellectual property: in most countries patents either protect the intellectual property of a company until 20 years after the patent is granted or 23 years after it is applied for. Anything that was published or patented longer ago than that is therefore free to use, even commercially. Copyright applies much longer than that, but only protects the publication and not the ideas described in the publication.
Do you just want to build electronics or to design and build electronics? Designing requires more knowledge than building someone else's design, although in both cases it helps to understand what you are doing.
If your first project will be a preamp or an SPL meter, why do you want to learn about digital to analog converters?
Converting a digital signal to an analog signal can be very simple when the accuracy requirements are not too high, see for example Resistor ladder - Wikipedia and scroll to the part about R-2R ladders. The problem with audio DACs is that the accuracy requirements usually lead to far more complicated circuits. That is, either the DAC itself gets complicated or you need a large amount of digital preprocessing to get away with a simple DAC.
An exception occurs when you only play so-called direct stream digital files. That format allows the use of quite simple DAC circuits without any digital preprocessing if you are not after extreme signal-to-noise ratios. See the "The best DAC is no DAC" thread.
Most DIY audio DAC builders just make an application circuit for an audio DAC chip, so most of the complexity is hidden inside the chip. There are some exceptions on this forum, like the Soekris circuits, the "no DAC", the Signalyst designs, xx3stksm's designs (he uses DAC chips, but in a very non-standard way) and my valve DAC.
About intellectual property: in most countries patents either protect the intellectual property of a company until 20 years after the patent is granted or 23 years after it is applied for. Anything that was published or patented longer ago than that is therefore free to use, even commercially. Copyright applies much longer than that, but only protects the publication and not the ideas described in the publication.
Last edited:
Hi Bret,
Probably not the best area of the forum to pose such a question, but we do try to be helpful in this area anyway.
To be successful and building electronics from the ground up, rather from a kit, you need to have at least some book learning in electronics. How would you describe your background with electricity and electronics so far?
Probably not the best area of the forum to pose such a question, but we do try to be helpful in this area anyway.
To be successful and building electronics from the ground up, rather from a kit, you need to have at least some book learning in electronics. How would you describe your background with electricity and electronics so far?
Phase - Thank you for bringing up the health concern. Maybe the spl meter I would try and market, so I am trying to understand if there is such a thing as a 'standard' design or one that is so common it is considered 'public domain.' I'm not looking to take someone elses design and put my name on it, rather looking for a starting point to build off of and make something unique.
Marcelvdg - Thank you. If you read my reply to phase, I sort of answered your first question. I'm looking to learn to design, but I don't know if it is necessary to 'reinvent the wheel' with every feature I want to add. Maybe preamp is not the correct term. I want to take a digital file from a thumb drive and pass it as an analog signal to my DSP. My DSP does not have a digital input and I have a couple of features to add to this, but I want to make it as simple as possible. No balance, no tone controls. What I don't understand is how converting it to a signal might color the sound, or interpret it in some way. Maybe I will understand if I read up on some of the concepts you mentioned.
Markw4 - I wasn't sure where to post this, but since I'm interested in line level stuff, not amplifiers, I thought I would ask here. I'm looking for recommendations on books. My experience with electronics is just basic stuff. I have had a multimeter my whole life and have used it to test things over the years. I don't have a good understanding of the relationship between current, voltage and resistance.
Marcelvdg - Thank you. If you read my reply to phase, I sort of answered your first question. I'm looking to learn to design, but I don't know if it is necessary to 'reinvent the wheel' with every feature I want to add. Maybe preamp is not the correct term. I want to take a digital file from a thumb drive and pass it as an analog signal to my DSP. My DSP does not have a digital input and I have a couple of features to add to this, but I want to make it as simple as possible. No balance, no tone controls. What I don't understand is how converting it to a signal might color the sound, or interpret it in some way. Maybe I will understand if I read up on some of the concepts you mentioned.
Markw4 - I wasn't sure where to post this, but since I'm interested in line level stuff, not amplifiers, I thought I would ask here. I'm looking for recommendations on books. My experience with electronics is just basic stuff. I have had a multimeter my whole life and have used it to test things over the years. I don't have a good understanding of the relationship between current, voltage and resistance.
There is a very good and well-known book entitled, The Art of Electronics. Think it may be in its third edition by now. Not cheap, but well worth the cost. You might take a look and see how much is understandable. If most of it, then great. You have a lot of reading to do. If not, then we could recommend something more introductory.
If you want a project to work on, I have been helping people modify cheap ebay dacs into quite good ones. It is a rather involved project though. The thread is here: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-320.html#post5609058 One part of the project involves constructing an output stage and instructions with pictures can be found here: Dropbox - Output Stage Instructions.zip You could take a look at that and see if it looks too hard, or within your present abilities. If interested, it is something we could provide a lot of support for and answer any questions. It might be good in that respect at least.
If you want a project to work on, I have been helping people modify cheap ebay dacs into quite good ones. It is a rather involved project though. The thread is here: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-320.html#post5609058 One part of the project involves constructing an output stage and instructions with pictures can be found here: Dropbox - Output Stage Instructions.zip You could take a look at that and see if it looks too hard, or within your present abilities. If interested, it is something we could provide a lot of support for and answer any questions. It might be good in that respect at least.
Last edited:
That sounds like you want to build a DAC, so then you are on the right forum. Technically it would be more elegant to modify your DSP to add a digital input, though, assuming that is possible somehow.
Coming back to why DACs change the sound to some extent (though not necessarily to an audible extent - you can find enormously long threads here about the question whether DAC imperfections are audible): there are all kinds of imperfections that you run into when you convert a digital signal into analog form. I've listed a few below, I hope it's not too detailed for you.
One issue is filtering. A digital signal has only a finite number of values per second (finite sample rate). When you keep the signal constant between the samples, you get a staircase-shaped signal that doesn't resemble a normal audio signal. It can be shown mathematically that you actually lose some treble and add ultrasonic rubbish. Filtering can correct for this, but each type of filter has its own imperfections.
Another problem with audio DACs is the accuracy you need. For example, imagine you make a 16-bit DAC by just making binary-scaled current sources and switching them on and off. When the code goes from binary 0111111111111111 to 1000000000000000 , 15 current sources switch off and the 16th switches on. The 15 that turn off ideally have a total weight of 32767 unit currents and the one that turns on 32768 unit currents, but when the one that turns on is 0.01 % too low, the signal actually goes in the wrong direction: from 32767 unit currents to less than 32765 unit currents. That can result in gross distortion of very soft sounds.
One approach the chip manufacturers can use to solve this issue is to laser trim the current sources (or resistors or capacitors, depending on the exact kind of DAC) until they meet the tough accuracy requirements. It's not an ideal solution because it is expensive and because mismatch between the long-term drifts of the current sources could cause the DAC to become less accurate again over time.
Another approach is thermometer coding: using 65535 unit current sources and switching on the number of sources indicated by the digital input code. This quickly becomes expensive in terms of chip area and difficult in terms of wiring, especially when you go to 20 or 24 bits. Compromises are possible: thermometer coding the upper few bits and doing the rest binary scaled.
An approach that has become very popular is sigma-delta or delta-sigma modulation. For a DAC, this boils down to using digital preprocessing to change the digital signal into a digital signal with much fewer bits (possibly just one) and a much higher sample rate.
Basically a sigma-delta modulator is a negative feedback loop around a very course rounding stage (so-called quantizer) that reduces the number of bits. The quantizer runs at a much higher speed than the input sample rate. The quantizer produces a lot of quantization noise (euphemism for quantization distortion, that is, round-off error). The feedback loop suppresses the quantization noise in the audio band and actually makes it somewhat worse outside the audio band.
When the quantizer and the DAC have only one bit, there is nothing that needs to match. Modern sigma-deltas often use a few bits rather than only one, but have some additional tricks (data weighted averaging, dynamic element matching) included to ensure that slight mismatch in the DAC doesn't impact the performance too much.
It is again not an ideal solution for various reasons. The ultrasonic noise can easily get converted to audible noise when it ends up somewhere where it shouldn't, for example when some of it couples to the reference voltage of the DAC. On top of that there are potential problems with idle tones for very-low-level signals, especially in single-bit sigma-delta converters.
One issue is filtering. A digital signal has only a finite number of values per second (finite sample rate). When you keep the signal constant between the samples, you get a staircase-shaped signal that doesn't resemble a normal audio signal. It can be shown mathematically that you actually lose some treble and add ultrasonic rubbish. Filtering can correct for this, but each type of filter has its own imperfections.
Another problem with audio DACs is the accuracy you need. For example, imagine you make a 16-bit DAC by just making binary-scaled current sources and switching them on and off. When the code goes from binary 0111111111111111 to 1000000000000000 , 15 current sources switch off and the 16th switches on. The 15 that turn off ideally have a total weight of 32767 unit currents and the one that turns on 32768 unit currents, but when the one that turns on is 0.01 % too low, the signal actually goes in the wrong direction: from 32767 unit currents to less than 32765 unit currents. That can result in gross distortion of very soft sounds.
One approach the chip manufacturers can use to solve this issue is to laser trim the current sources (or resistors or capacitors, depending on the exact kind of DAC) until they meet the tough accuracy requirements. It's not an ideal solution because it is expensive and because mismatch between the long-term drifts of the current sources could cause the DAC to become less accurate again over time.
Another approach is thermometer coding: using 65535 unit current sources and switching on the number of sources indicated by the digital input code. This quickly becomes expensive in terms of chip area and difficult in terms of wiring, especially when you go to 20 or 24 bits. Compromises are possible: thermometer coding the upper few bits and doing the rest binary scaled.
An approach that has become very popular is sigma-delta or delta-sigma modulation. For a DAC, this boils down to using digital preprocessing to change the digital signal into a digital signal with much fewer bits (possibly just one) and a much higher sample rate.
Basically a sigma-delta modulator is a negative feedback loop around a very course rounding stage (so-called quantizer) that reduces the number of bits. The quantizer runs at a much higher speed than the input sample rate. The quantizer produces a lot of quantization noise (euphemism for quantization distortion, that is, round-off error). The feedback loop suppresses the quantization noise in the audio band and actually makes it somewhat worse outside the audio band.
When the quantizer and the DAC have only one bit, there is nothing that needs to match. Modern sigma-deltas often use a few bits rather than only one, but have some additional tricks (data weighted averaging, dynamic element matching) included to ensure that slight mismatch in the DAC doesn't impact the performance too much.
It is again not an ideal solution for various reasons. The ultrasonic noise can easily get converted to audible noise when it ends up somewhere where it shouldn't, for example when some of it couples to the reference voltage of the DAC. On top of that there are potential problems with idle tones for very-low-level signals, especially in single-bit sigma-delta converters.
So this staircase shaped signal could show up on a oscilloscope if it wasn't filtered? Is the shape like that because between samples there is no information and it somehow gets interpolated? I follow the first two sampling methods, but you lost me at "Basically a sigma-delta modulator is a negative feedback loop around a very course rounding stage." I don't know these terms and I'll look them up. Thank you for taking the time to explain.
I think that was supposed to be coarse, not course. Just mention that for clarity, since this sounds its like its a new subject to you.
The stair-step is not interpolated or it might look more like a ramp or a curve connecting the two endpoints. Rather, it more like something called a sample-and-hold where a momentary signal is sampled and then held at a constant output level until the next momentary signal comes along and sets a new voltage to hold. That type of process produces flat stair-step waveforms.
With some dacs, signals at that stage might be observable with an oscilloscope. With other types of dacs those signals may be inaccessible inside an integrated circuit.
The stair-step is not interpolated or it might look more like a ramp or a curve connecting the two endpoints. Rather, it more like something called a sample-and-hold where a momentary signal is sampled and then held at a constant output level until the next momentary signal comes along and sets a new voltage to hold. That type of process produces flat stair-step waveforms.
With some dacs, signals at that stage might be observable with an oscilloscope. With other types of dacs those signals may be inaccessible inside an integrated circuit.
Last edited:
Negative feedback is a concept that's used all over electronics (and outside electronics), so it is also worthwhile to look that up if you should decide to make something else than a DAC. Unfortunately you will probably also find a lot of misinformation about it.
Might I suggest using your woodworking skills to build some loudspeakers and then build active crossovers and amplifiers to drive them.
You will then be able to actually hear the fruits of your labours and how changing the filter parameters affects the sound.
Years ago I built some open dipole speakers with full equalisation and five power amps and enjoyed the build and the end results.
The Phoenix from:
Linkwitz Lab - Loudspeaker Design
During the build your SPL meter would come in useful during the testing and adjusting stage.
Batteryman - I have two projects I'm trying to finish now. They are both stave constructions, which is like a wood barrel, hot tub and canoe construction method, where several planks are joined to make round or curved shapes. I have one enclosure built but can't afford the drivers at the moment. The other is a vintage speaker I'm rebuilding. The parts are on order for the speaker and I'm trying to repair a plate amp I was given to power it. I haven't even started on that enclosure yet. I want to start building electronics because the things I want for my personal use are not commercially available. Everything has way too many features I don't use and I don't like paying extra for clutter and ugliness.
Scottjoplin - I started reading some 'negative feedback' articles last night but it didn't make sense. are negative feedback and balanced lines related in any way? Is it error correction?
Scottjoplin - I started reading some 'negative feedback' articles last night but it didn't make sense. are negative feedback and balanced lines related in any way? Is it error correction?
I agree, commercial product manufacturers seem to always be in a battle of features and specs, without asking customers whether they actually want them and of course you can usually make a better product for less.
I envy your woodworking skills! (& too old to learn new tricks)
Woodworking is a hobby and all my skills are whatever I can carry over from my metalworking/manufacturing experience. I feel like I'm too old to try and do this electronics thing. I'll give it my 100% focus for a while and will know if it seems possible or not soon enough. I recently tried to jump into database and user interface design and gave up after a month. I wrote a program that worked with a database then learned that everything I created was vulnerable to hacking and that I would have to start over.
There is not much of a relation between feedback and balanced line technology. Many balanced circuits use feedback, but the same holds for unbalanced circuits.
The silly thing about it is that you have been applying negative feedback since the moment you learned to walk and probably even earlier. When you walk on a long and narrow sidewalk, the reason you don't eventually step on the street or hit the walls of the buildings, is the fact that you are continuously looking how far you are from the edge and from the walls and correcting your direction accordingly. That's negative feedback: compare the output of an inaccurate process (walking) with the desired output (following the sidewalk) and correcting it where needed.
Negative feedback is an error reduction technique; in the example, you can follow the sidewalk far more accurately with your eyes open than with your eyes closed. There are other error reduction techniques, such as error feedforward, but negative feedback is by far the most common one.
You can apply negative feedback to an amplifier (also to a sigma-delta modulator, but an amplifier is more convenient as an example). Suppose you want to make an amplifier that amplifies a signal voltage 20 times. Without any error reduction techniques, it is very well possible to make an amplifier with a gain of about 20, but the actual gain will be dependent on all sorts of things, such as temperature and the batch of transistors. Worst of all for audio, it will depend on the momentary voltage of the signal itself. When the gain varies depending on where you are in the signal waveform, you get distortion.
Alternatively, you can make an inaccurate amplifier with much more gain than 20 times and with a differential input. Using two accurate resistors, you can accurately divide the output voltage by 20. The differential input of the inaccurate high-gain amplifier then gets connected between the input voltage and the voltage from the voltage divider.
Connected like this, the inaccurate high-gain amplifier is driven by the difference between the input voltage and 1/20th of the output voltage. That is, it basically compares its input voltage to 1/20th of its output voltage and corrects its output as needed.
The silly thing about it is that you have been applying negative feedback since the moment you learned to walk and probably even earlier. When you walk on a long and narrow sidewalk, the reason you don't eventually step on the street or hit the walls of the buildings, is the fact that you are continuously looking how far you are from the edge and from the walls and correcting your direction accordingly. That's negative feedback: compare the output of an inaccurate process (walking) with the desired output (following the sidewalk) and correcting it where needed.
Negative feedback is an error reduction technique; in the example, you can follow the sidewalk far more accurately with your eyes open than with your eyes closed. There are other error reduction techniques, such as error feedforward, but negative feedback is by far the most common one.
You can apply negative feedback to an amplifier (also to a sigma-delta modulator, but an amplifier is more convenient as an example). Suppose you want to make an amplifier that amplifies a signal voltage 20 times. Without any error reduction techniques, it is very well possible to make an amplifier with a gain of about 20, but the actual gain will be dependent on all sorts of things, such as temperature and the batch of transistors. Worst of all for audio, it will depend on the momentary voltage of the signal itself. When the gain varies depending on where you are in the signal waveform, you get distortion.
Alternatively, you can make an inaccurate amplifier with much more gain than 20 times and with a differential input. Using two accurate resistors, you can accurately divide the output voltage by 20. The differential input of the inaccurate high-gain amplifier then gets connected between the input voltage and the voltage from the voltage divider.
Connected like this, the inaccurate high-gain amplifier is driven by the difference between the input voltage and 1/20th of the output voltage. That is, it basically compares its input voltage to 1/20th of its output voltage and corrects its output as needed.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.