mikewu99 said:
I got that book... was my text in my engineering graphics course.
I didn't mean to say that it was invented on the Mac, just that the Mac set the standards for graphical interfaces and that was just pure brute force numbers of units in the field. Most of the work for the Mac GUI predated the Mac at Xerox PARC (Palo Alto Research Centre)
dave
I've been out of contact for a few days so this is pretty late, but I must say that it is just about impossible to meaningfully discuss an AC signal in anything other than a purely resistive circuit without understanding complex numbers. Near as I can tell that covers just about the entire audio field.
Now this is not terribly difficult to understand. The process of handling complex number is basic trig (even the cheapest scientific calculator does rectangular to polar conversions). I remember doing this stuff in high school. My initial entry into electronics was via Ham radio and there was no way you were going to be successfull dealing with RF without having a least a cursory understanding of the frequency related nature of impendance and parasitic coupling. There is no simple way to use this knowledge without the math.
Besides being a analysis tool, math is also a very concise language that is extremely well suited to describing some very complex concepts that are important to understanding how things really work. If you don't understand the words then there is no way for anyone to really explain the concepts to you.
I guess the point I'm stumbling around is why do DIY Audio? Nice esthetics are great to look at (Peter your enclosures make me crazy with envy), but are not really the heart of the matter. The point at least for me to DIY audio is that I know that it is possible to get better results than most of the commercially available gear, for at least two reasons. First my time is free and second I'm not that worried about trimming another $1.27 from the bill of materials cost. This means that I can pursue some extreme concept that are not commercially feasible to manufacture, but can result in superior performance. But extreme audio concepts are also not normally available predigested or in a kit form.
I design high volume electronics products for a living and trimming $1 from the parts cost of a board is considered a real achievement. It is very relaxing to me to be freed from the tyranny of "value engineering" and DIY audio gives me that freedom. But what is the point if all you are doing is cookbook designs or building kits without any real understanding what it is that makes one circuit better than another.
The following comments are possibly going to offend some people and please forgive me if they do, because that is not my intent. This entire thread speaks to what I see as a great divide in the DIY community.
In my opinion, a lot of the tweeker activity and "platinum plated" mentality I have seen in DIY audio is really just due to enthusiasts who want to make a difference but lack the mental tools to understand what is actually happening inside their gear. After all it doesn't take much thoughtful analysis to plug in a new interconnect cable or replace all of the caps in a piece of high end gear with order of magnitude more expensive parts. Never mind if the change is actually doing anything helpful. After all if something has been "tweeked and moded" it must be better, especially if the new parts cost a lot.
The reality is that despite the lack of "pedigree" for the parts replaced, at least those parts were chosen based on the requirements of the circuit. I very rarely see any real analysis done of the impact these changes make on the circuit so at least in my mind it's a toss up if the changes made things better or worse. The standard refrain is "try it and let your ears judge". Never mind if for example the change in a caps ESR or frequency response totally screws up the design criteria and reduces reliability or function.
I am not saying that caps and cables don't matter and I am also not saying that only an electrical engineer can make meaningful improvements to audio gear. What I am saying is that if you don't work to develop analytical skills, then one line of reasoning will often sound just as good as another and then you are the natural prey for the con men and hucksters in audio.
Any good design is a balancing act that maximises desirable goals while minimising negatives. You can just pile stuff up and for a while it might not collapse, but if you want to really design a robust, high quality piece of gear then it has to be a "design" not just a pile of parts thrown together by the winds of fashion or luck.
Phil
Now this is not terribly difficult to understand. The process of handling complex number is basic trig (even the cheapest scientific calculator does rectangular to polar conversions). I remember doing this stuff in high school. My initial entry into electronics was via Ham radio and there was no way you were going to be successfull dealing with RF without having a least a cursory understanding of the frequency related nature of impendance and parasitic coupling. There is no simple way to use this knowledge without the math.
Besides being a analysis tool, math is also a very concise language that is extremely well suited to describing some very complex concepts that are important to understanding how things really work. If you don't understand the words then there is no way for anyone to really explain the concepts to you.
I guess the point I'm stumbling around is why do DIY Audio? Nice esthetics are great to look at (Peter your enclosures make me crazy with envy), but are not really the heart of the matter. The point at least for me to DIY audio is that I know that it is possible to get better results than most of the commercially available gear, for at least two reasons. First my time is free and second I'm not that worried about trimming another $1.27 from the bill of materials cost. This means that I can pursue some extreme concept that are not commercially feasible to manufacture, but can result in superior performance. But extreme audio concepts are also not normally available predigested or in a kit form.
I design high volume electronics products for a living and trimming $1 from the parts cost of a board is considered a real achievement. It is very relaxing to me to be freed from the tyranny of "value engineering" and DIY audio gives me that freedom. But what is the point if all you are doing is cookbook designs or building kits without any real understanding what it is that makes one circuit better than another.
The following comments are possibly going to offend some people and please forgive me if they do, because that is not my intent. This entire thread speaks to what I see as a great divide in the DIY community.
In my opinion, a lot of the tweeker activity and "platinum plated" mentality I have seen in DIY audio is really just due to enthusiasts who want to make a difference but lack the mental tools to understand what is actually happening inside their gear. After all it doesn't take much thoughtful analysis to plug in a new interconnect cable or replace all of the caps in a piece of high end gear with order of magnitude more expensive parts. Never mind if the change is actually doing anything helpful. After all if something has been "tweeked and moded" it must be better, especially if the new parts cost a lot.
The reality is that despite the lack of "pedigree" for the parts replaced, at least those parts were chosen based on the requirements of the circuit. I very rarely see any real analysis done of the impact these changes make on the circuit so at least in my mind it's a toss up if the changes made things better or worse. The standard refrain is "try it and let your ears judge". Never mind if for example the change in a caps ESR or frequency response totally screws up the design criteria and reduces reliability or function.
I am not saying that caps and cables don't matter and I am also not saying that only an electrical engineer can make meaningful improvements to audio gear. What I am saying is that if you don't work to develop analytical skills, then one line of reasoning will often sound just as good as another and then you are the natural prey for the con men and hucksters in audio.
Any good design is a balancing act that maximises desirable goals while minimising negatives. You can just pile stuff up and for a while it might not collapse, but if you want to really design a robust, high quality piece of gear then it has to be a "design" not just a pile of parts thrown together by the winds of fashion or luck.
Phil
Mmm, at last a fresh uncoloured view on matters
Ever thought that the 1 $ that was trimmed by the engineers is put back again in it by the tweaker ? Isn't "value engineering" not the reason for some to open their devices and wanting to change it ?
That requirement being: lowest possible cost.
And all this science I don't understand
I have advocated theory basic electronics as well as provided links for the more esoteric subjects like diode reverse recovery noise. There are plenty of technically inclined people on the forum.
My problem is the "engineers" who look with total skepticism on tweaks without any investigation into material science, physics, RF design, or anything outside the most pedestrian electrical engineering approaches to audio design. Vibration damping of electrical circuits and parts come to mind as an example of this.
I have also tried to suggest that people read the data sheets when using a part. I have also proposed the use of Spice modeling as a learning tool for issues like phase margin.
When I posted a link for an article on phase margin Mr. Rollins had a fit. I presume it was for advocating the very viewpoint of learning some math and theory that you propose.
haldor said:
I have advocated theory basic electronics as well as provided links for the more esoteric subjects like diode reverse recovery noise. There are plenty of technically inclined people on the forum.
My problem is the "engineers" who look with total skepticism on tweaks without any investigation into material science, physics, RF design, or anything outside the most pedestrian electrical engineering approaches to audio design. Vibration damping of electrical circuits and parts come to mind as an example of this.
I have also tried to suggest that people read the data sheets when using a part. I have also proposed the use of Spice modeling as a learning tool for issues like phase margin.
When I posted a link for an article on phase margin Mr. Rollins had a fit. I presume it was for advocating the very viewpoint of learning some math and theory that you propose.
jean-paul said:
Mmm, at last a fresh uncoloured view on matters
Ever thought that the 1 $ that was trimmed by the engineers is put back again in it by the tweaker ? Isn't "value engineering" not the reason for some to open their devices and wanting to change it ?
That requirement being: lowest possible cost.
Hi jean-paul,
You've mistaken my point. I think you left the second half out of of your last sentence. Value engineeing means "Lowest possible cost that gets the job done". More expensive parts that are not appropriate for the task are not going to work better.
I agree that putting that $1 (or $10 or even $100) back into a circuit can be rewarding, but only if the more expensive part has the characteristics required by the design and only if the better characteristics matter to the specific role being performed by the part.
For example, a particular component may have been selected because of it's specific frequency response. Perhaps the circuit needs a response roll off above 100KHz at this point in the circuit and by using a "lousy" part adding more components to perform the roll off was avoided. This is called design elegance in my book. When that "lousy" part is replaced with a "better" one the circuit may no longer behave properly. You could argue that adding more parts is better, but there are lots of people who will argue the opposite (the less number of parts the better).
That is the point of my message. Building and improving gear requires that you take the time to understand how the gear works. If you don't understand what the part is doing in the circuit then how can you judge if one part is "better" than another? Ignorance with a big bank account is only going to end up with better results by accident.
I recall a neighbor of mine who once in all seriousness asked my wife if a particular plant was a member of the weed family. He was under the impression that useful plants are scientifically catagorised differently than weeds. The reality is that good or bad is totally dependent on what you are trying to use something for. Wheat is a weed when it's on the golf course, but that doesn't mean we shouldn't use it to make our bread.
I have often seen displayed here a tendancy to catagorize certain commonly used, lower cost components as being members of an electronic "weed" family. If these parts are used in applications that require performance they are not able to provide then they are bad. But just because they can be misused, does not mean that there is anything wrong with using them appropriately. A tantalum cap in a circuit does not automatically indicate the presence of evil (just don't pass audio through it).
In an absurd example, the temperature stability and high frequency response of silver mica caps are not crucial to the task of making power supply filters for power amps. I don't need to spend the money on 0.2F worth of silver mica caps to verify this and if someone claims that I do then my respect for their opinions is decreased. There are other aspects of audio design that are not as clear cut and I am not adverse to experimentation, but I will let my training and my intellect guide that exploration for me.
Phil
Re: And all this science I don't understand
Hi Fred,
I have no argument with your statements. It is amazing to me how often I stumble across flaws in commercial designs that obviously came from not reading (or understanding) the specifications for the parts. I can't recall how many times I have seen designs that depend on "typical" or better performance from key parts in them to work.
It is also possible to get into trouble because you have too good a supplier. My company recently changed over to a "global" procurement system to take advantage of our company wide volume for better pricing. We had been using a particular brand of precision resistors in our designs for years without any problems and as part of this cost optimisation purchasing switched to a lower cost supplier. Turns out the parts we had been buying were actually performing much better than specified (part was specified for 50 ppm temperature stability but actual performance was closer to 5 ppm). The new cheaper parts met the 50 ppm paper spec, but didn't work correctly in some of our designs. I'm guessing now because the original designers are no longer here, but I'll bet we never identified the need for a tighter spec in this particular usage and just lucked out due to our good supplier. The whole exercise was a real eye opener for me. Cost quite a bit of money to correct as well.
Good designs include margins for robustness and to permit use of less expensive parts (which often vary more than better parts). In designs with lots of margin you can often get away with blindly changing components and the thing will still work acceptably. The problem is if you don't analyse what you are doing then you have no way of knowing how close to failure you are getting. The part of value engineering that I dislike the most is reducing design margins. Often the more cost optimised a product is the more carefully "tweaking" must be done. This is unfortunate, because these are also the same designs that will often gain the most from the right tweaking.
That is one more thing to be said in favor of buying lower volume products. Odds are the design margins haven't been cost optimized to the razors edge of failure. You can't justify spending $10,000 in NRE to save $1 a unit, if you only sell 1000 of them a year. Most of the products I design in my day job end up being made in greater than 100K annual quantities so the pressure to save that $1 in unit cost can be huge.
Phil
Fred Dieckmann said:
Hi Fred,
I have no argument with your statements. It is amazing to me how often I stumble across flaws in commercial designs that obviously came from not reading (or understanding) the specifications for the parts. I can't recall how many times I have seen designs that depend on "typical" or better performance from key parts in them to work.
It is also possible to get into trouble because you have too good a supplier. My company recently changed over to a "global" procurement system to take advantage of our company wide volume for better pricing. We had been using a particular brand of precision resistors in our designs for years without any problems and as part of this cost optimisation purchasing switched to a lower cost supplier. Turns out the parts we had been buying were actually performing much better than specified (part was specified for 50 ppm temperature stability but actual performance was closer to 5 ppm). The new cheaper parts met the 50 ppm paper spec, but didn't work correctly in some of our designs. I'm guessing now because the original designers are no longer here, but I'll bet we never identified the need for a tighter spec in this particular usage and just lucked out due to our good supplier. The whole exercise was a real eye opener for me. Cost quite a bit of money to correct as well.
Good designs include margins for robustness and to permit use of less expensive parts (which often vary more than better parts). In designs with lots of margin you can often get away with blindly changing components and the thing will still work acceptably. The problem is if you don't analyse what you are doing then you have no way of knowing how close to failure you are getting. The part of value engineering that I dislike the most is reducing design margins. Often the more cost optimised a product is the more carefully "tweaking" must be done. This is unfortunate, because these are also the same designs that will often gain the most from the right tweaking.
That is one more thing to be said in favor of buying lower volume products. Odds are the design margins haven't been cost optimized to the razors edge of failure. You can't justify spending $10,000 in NRE to save $1 a unit, if you only sell 1000 of them a year. Most of the products I design in my day job end up being made in greater than 100K annual quantities so the pressure to save that $1 in unit cost can be huge.
Phil
Haldor:
I'm not sure I agree with your thoughts on low volume design. A high-volume design can afford to spend more on ensuring that the design works, and works reliably. I have a Tek571 curve tracer that is frighteningly fragile - look at it in a funny way and the collector supply has a bonfire. When I finally prised the manual out of Tek, I found that this supply was festooned with add-on CR networks desperately trying to make it stable. I concluded that because there's not a lot of call for curve tracers, they didn't spend the money on design in the first place because they wouldn't have been able to recover it.
By the way, my first step in fixing it was calculate all the DC voltages within the circuit, then calculate the gain. Armed with these pencil annotations, I set about finding/replacing faulty components. (An op-amp with a bulging/burnt underbelly, a charred mess that later turned out to be a resistor, five transistors, a DAC, and a possibly innocent multiplexor.) My heart is in my mouth whenever I test MOSFETs. Tek win a prize for their statement, "Most of the protection of the 571 has been implemented in the software."
Sy:
Maths is a tool, and a very powerful one. Neglect it at your peril. However, it's easy to get so wrapped in the maths that you lose track of what the problem actually was. Regarding complex numbers, I remember being told that, "Maths is just a game, when we can't play it anymore, we invent new rules until we can." I've found that electronics students will swallow complex numbers quite cheerfully if they're previously had to solve problems graphically. The graphical introduction means they understand what's happening, and the complex algebra enables them to do it in half the time.
In electronics, the total quanity of maths and smoke is fixed. Increase the maths, decrease the smoke...
I'm not sure I agree with your thoughts on low volume design. A high-volume design can afford to spend more on ensuring that the design works, and works reliably. I have a Tek571 curve tracer that is frighteningly fragile - look at it in a funny way and the collector supply has a bonfire. When I finally prised the manual out of Tek, I found that this supply was festooned with add-on CR networks desperately trying to make it stable. I concluded that because there's not a lot of call for curve tracers, they didn't spend the money on design in the first place because they wouldn't have been able to recover it.
By the way, my first step in fixing it was calculate all the DC voltages within the circuit, then calculate the gain. Armed with these pencil annotations, I set about finding/replacing faulty components. (An op-amp with a bulging/burnt underbelly, a charred mess that later turned out to be a resistor, five transistors, a DAC, and a possibly innocent multiplexor.) My heart is in my mouth whenever I test MOSFETs. Tek win a prize for their statement, "Most of the protection of the 571 has been implemented in the software."
Sy:
Maths is a tool, and a very powerful one. Neglect it at your peril. However, it's easy to get so wrapped in the maths that you lose track of what the problem actually was. Regarding complex numbers, I remember being told that, "Maths is just a game, when we can't play it anymore, we invent new rules until we can." I've found that electronics students will swallow complex numbers quite cheerfully if they're previously had to solve problems graphically. The graphical introduction means they understand what's happening, and the complex algebra enables them to do it in half the time.
In electronics, the total quanity of maths and smoke is fixed. Increase the maths, decrease the smoke...
A point I've made several times in this thread!
I've done designs for low volume (one-offs to a few dozen) and high volume (millions) production; your point is quite valid, each has their own necessities and trade-offs. I am, frankly, disappointed to hear that Tek would release an instument so flakey. Every piece of Tek gear I've ever used was super-solid and I had gained (apparently falsely) a complete confidence in their products.
Tek trouble
I am told that Tek went through a bit of a bad patch about eleven years ago, and had to use a lot of euphemisms (downsizing etc). The 571 is from that era. Incidentally, 24** series oscilloscopes and 1741/51 waveform monitors are also from that era. If an output transistor on the Y amplifier lets go, it's terminal because the company that supplied the hybrids to Tek went out of business, and they didn't find a second source (other than cannibalising old kit). A year ago, I persuaded them to fix a 2465B, but the same fault has just reappeared.
In case anyone thinks this is gratuitous Tek-bashing, I used to have a 485, and now have a TDS3032.
I am told that Tek went through a bit of a bad patch about eleven years ago, and had to use a lot of euphemisms (downsizing etc). The 571 is from that era. Incidentally, 24** series oscilloscopes and 1741/51 waveform monitors are also from that era. If an output transistor on the Y amplifier lets go, it's terminal because the company that supplied the hybrids to Tek went out of business, and they didn't find a second source (other than cannibalising old kit). A year ago, I persuaded them to fix a 2465B, but the same fault has just reappeared.
In case anyone thinks this is gratuitous Tek-bashing, I used to have a 485, and now have a TDS3032.
EC8010 said:
That is a shame about the Tek. I own an old 465 and use the newer TDS scopes at work. They have all been very reliable for me. I guess even the good guys can screw up.
I will stand by my general observation that low volume designs will tend to be more conservative with bigger operating margins, because it costs money to see just how cheaply it is possible to make something. There is always the possibility that the designer didn't really have a handle on the requirements for the circuit. In that case the low volume product may not have been tested adequately enough to shake out problems. I guess my position is more true for relatively straightforward products that are well understood (and where there are reference designs from manufacturers to borrow from).
Phil
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.