On a recent post, hugobross showed us a PCB for a high-power amp. Very nice work, reminded me a discussion I'd like to start.
I would like opinions about the effect of rounded corners in signal-carrying traces.
Have read in many places that rounded corners produce better sound than square ones, but I haven't got the opportunity to check.
Anyone with experiences in this area?
I would like opinions about the effect of rounded corners in signal-carrying traces.
Have read in many places that rounded corners produce better sound than square ones, but I haven't got the opportunity to check.
Anyone with experiences in this area?
I don't know about sound quality, but square corners can cause arcs at high enough voltages. However, the voltages involved are far, far beyond anything you'd use in a solid state amp. I'd say it's mainly an esthetic decision. I would recommend using all square, all 45 degree, or all round within the same board, but just so as to make your eyes happy.
Grey
Grey
At RF freqencies, traces act more like transmission lines. You must use rounded corners, or your signal is apt to continue on into space as you make a sharp turn. 
I'm not aware of this being an issue at audio frequencies; rise times are extremely slow (hundreds of nanoseconds) and the systems remain lumped over long distances. (as usual, AFAIK)
Perhaps though this is another situation in which theory is questionable. Maybe rounded traces make the designer "happier." It certainly couldn't hurt much.
I'm not aware of this being an issue at audio frequencies; rise times are extremely slow (hundreds of nanoseconds) and the systems remain lumped over long distances. (as usual, AFAIK)
Perhaps though this is another situation in which theory is questionable. Maybe rounded traces make the designer "happier." It certainly couldn't hurt much.
If you consider it from an electromagnetic field point, any sharp turns are inherently bad. Having said that, most people don't seem to bother.
Whether it is worth your time or not, depends. I would think that large current situations are the worst. If you use regular components, you will have sharp turns regardless of what you do.
I feel good about making rounded corners. Your mileage may vary on that one!.
Petter
Whether it is worth your time or not, depends. I would think that large current situations are the worst. If you use regular components, you will have sharp turns regardless of what you do.
I feel good about making rounded corners. Your mileage may vary on that one!.
Petter
I have read differing opnions about both rounded and 45 degree turns.
Rounded turns are slightly longer, but usually not much, and unless you are using Surfacemount the Pad to component lead looks about 90 degrees to me
Vutrax (CAD) has a page that might be of interest
<a href="http://www.vutrax.co.uk/vbook3.htm">Track Widths and Clearances</a>
Some empirical data on the width of tracks required to handle various currents, plus track clearances required by high voltages.
Regards
James.
Rounded turns are slightly longer, but usually not much, and unless you are using Surfacemount the Pad to component lead looks about 90 degrees to me
Vutrax (CAD) has a page that might be of interest
<a href="http://www.vutrax.co.uk/vbook3.htm">Track Widths and Clearances</a>
Some empirical data on the width of tracks required to handle various currents, plus track clearances required by high voltages.
Regards
James.
Found one of the Links I was thinking of
<a href="http://www.ultracad.com/90deg.pdf">90 Degree Corners, The Final Turn</a>
The TDR test which included 45 and 90 degree turns showned no impedance difference with 17pS rise time pulses or any increase of radiated emmisions to 1.3GHz
dubious effects at Audio frequencies one would think?
Lots more data on PCB design at <a href="http://www.ultracad.com/tech.htm">UltraCAD Design, Inc's Technical/Design Notes</a>
James
<a href="http://www.ultracad.com/90deg.pdf">90 Degree Corners, The Final Turn</a>
The TDR test which included 45 and 90 degree turns showned no impedance difference with 17pS rise time pulses or any increase of radiated emmisions to 1.3GHz
dubious effects at Audio frequencies one would think?
Lots more data on PCB design at <a href="http://www.ultracad.com/tech.htm">UltraCAD Design, Inc's Technical/Design Notes</a>
James
That is a very interesting paper. I know Maxim's datasheets include some strong language encouraging the use of rounded traces and microstrip/stripline techniques. Probably this is because in video very small phase errors (>0.01 degree) or gain errors (>0.05%) can create display errors, so even minute impedence changes create problems.
One possiblity is that the effects of 90 degree bends are exacerbated in the absence of such techniques. However, I would still not be concerned about such effects on signals less than several MHz.
One possiblity is that the effects of 90 degree bends are exacerbated in the absence of such techniques. However, I would still not be concerned about such effects on signals less than several MHz.
Frequency above severall Mhz
Hi Trioth
You say the effects of 90° corners might become measureable above frequencys of severall Mhz, but an audio amplifier is much below that.
Now was I wondering, the closed loop bandwith of an amp is indeed very low (in comparison with severall Mhz). But the modern driver transistors go as high as 100 Mhz. This makes it possible to use alot of feedback to lower distorsion (and I know most do not like feedback). But isn't it so we do encouter these really high frequencys in the feedbackloop ? I know the feedbbackresistors are the most critical resistors in an amp. They are critical because of their "inductivity", so frequencys have to be very high.
Of course I could be very wrong about this, as my knowlegde of amplifier design is still quite limited (but I am working on it !! )
Any thoughts on this ?
Greetz
Wim
Hi Trioth
You say the effects of 90° corners might become measureable above frequencys of severall Mhz, but an audio amplifier is much below that.
Now was I wondering, the closed loop bandwith of an amp is indeed very low (in comparison with severall Mhz). But the modern driver transistors go as high as 100 Mhz. This makes it possible to use alot of feedback to lower distorsion (and I know most do not like feedback). But isn't it so we do encouter these really high frequencys in the feedbackloop ? I know the feedbbackresistors are the most critical resistors in an amp. They are critical because of their "inductivity", so frequencys have to be very high.
Of course I could be very wrong about this, as my knowlegde of amplifier design is still quite limited (but I am working on it !!
)Any thoughts on this ?
Greetz
Wim
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.