|
Home
/ FAQ
/ News Classifieds / Events |
Audio Asylum Thread Printer |
Get a view of an entire thread on one page |
68.0.154.81
I would soon like to begin matching the impedance of all data carrying lines (traces) in my new CD player. For convenience in this project I would prefer to use surface mount R's. Opinions please on who makes the highest quality SMT R's ?P.S.
Anyone know what the average propagation speed of data traces on FR-4 pcb material is ?
The characteristic impedance also depends upon the track width and it's spacing from the ground plane, if any, and from stuff close enough for it to fringe to.I've got info on track Z and can dig it up for you if you wish, but I bet you can Google it up about as fast. Since matched busses are all the rage these days it should be easy to find.
Resistor wise I doubt that type matters, SMT parts tend to have good bandwidth because as they don't have lead inductance. If you've got a fast scope try looking at the signals and see if you can spot any reflections and ringing.
I've never looked at mine, didn't cross my mind. I'm curious why you are concerned, do you think it's causing jitter? Perhaps I should be worried about it too...
Good luck,
Rick
Thanks for your reply. My concern with the propagation speed of traces on F4 pcb material stems from an observation some years ago by John Algaier, then a digital engr. with Sun Systems, regarding the speed of data in digital circuitry : The data lines must provide not only for the speed of the data being sent, but also for their EDGE SPEED (rise time) which can be as high as several mega hertz or more. This is also borne out in a book by James Buchanan (Advisory Engr., westinghouse Electric Corp.) "CMOS/TTL Digital Systems design": If a digital data line cannot achieve the necessary speed of the data being sent it will result in false toggling (too soon or too late) which has the same effect as "jitter". Buchanan states that the ideal impedence of data lines is approximately 60 ohms and that data lines' impedence should be matched to 80% of that figure i.e. 50 ohms. It would seem logical that if the data line cannot provide for the edge speed of the data being transmitted that too could result in false toggling. Buchanan also points out that most of the radiation (crosstalk) from traces is from their edges and therefore data lines should be "guarded" by a ground trace close by each side of the data line, or the data should be transmitted by twisted pair wires with one of the pairs being grounded at each end. In the event, it may well be desireable to send critical data via twisted pair wires since no CD player I have seen has data lines guarded by ground traces on either side. I have thought that such data might better be sent using Cat5e twisted pairs (rated for 350 MHtz transmission speed)). However, in that case the problem of available space rears it ugly head. I'm convinced that no one yet has even an inkling of the sonics that CD players are capable of. Oh yes, they work (they make noise) but nothing near what they are truly capable of producing. Tarzan was right : "It's a jungle out there !"
Well, there are a few vines hanging around, but it's not quite a jungle. The saving grace is the systems are clocked. Let's say that you have a DAC, overtly all the data line(s) need to do is settle within the voltage range for their logic states for the setup time before the clock hits to latch in the correct code. But secondary effects can rear their ugly heads sometimes such as noise from data transitions showing up on the DAC outputs due to internal crosstalk.However it's usually the clock lines that causes problems. Minor variations in data signal voltages caused by crosstalk or reflections are simply ignored as long as the signal stays within it's rather large window. The same variations in the clock signal however cause timing jitter. This can be especially bad if the coupled energy is correlated to the amplitude of the signal as it produces nasty FM artifacts.
So if you decided to try to improve you player by tweaking the signal paths, you may want to concentrate your efforts on the clock lines.
With respect to CD players in general, I think that nowadays the factors that affect the sound are quite well understood. Of course that doesn't mean that all players fully optimized...
Rick
I totally agree. I think that, ideally, the clock should have it's own well designed stiff power supply and the clock lines should be twisted pairs, as noted above. Also, the xtl's case should be grounded directly to the ground plane of the pcb and the xtl wrapped with several layers of lead tape to shield it from vibration. Lead tape is available at any golf store or pro shop where it is used to increase swing weight of golf clubs. All IC's should also have their ground pins grounded directly to the pcb ground plane since many ground busses that they are connected to wander long lengths before making ground. Ground busses in general are bad news.Onwards Tweakers, Onward !
That all sounds good except for the XTAL, I think that may not be the right direction, here's why...Way back in history I used to work for an outfit that built FM telemetry transmitters for rockets. A problem was that the rockets were really rough when the motors were running and that vibrated the crystals causing incidental FM which robbed us of useful deviation. So, the solution was to encase the crystal in a little cube of custom molded, very low durometer silicon rubber. Their leads were cut short and brought out along with a case ground using Litz wire. They were then largely isolated from shock and Vib.
Perhaps a similar approach would be of value for audio. Using lead tape seems to me more likely to lower the resonance rather than to provide isolation.
Good luck in your quest...
