Pcm2702 Usb-dac - Page 3

00940 · 27th February 2004, 10:57 AM

I've done my homework.

Here's a single sided pcb, analog and digital grounds connected. Power supply traces are 40mil, signal is 16 mil. Clearance is 8mil. Grid is 1 inch. I hadn't a correct layout for the inductor, so I used a cap layout. The computer ground is not directly connected to the groundplane but to the chip only.

Gyula · 27th February 2004, 02:29 PM

Hi Rookie!

In this way, everything is coupled capacitively to your ground plane, and it's Full of ground loops!

Good luck,
Gyula

Gyula · 27th February 2004, 02:31 PM

Oops!

I have to apologize from Rookie, I wrote the reply above to 00940.

Adam Reed · 27th February 2004, 03:22 PM

Gyula--I would have to say that with this single sided board the capacitive coupling from a trace to the ground plane is next to nil. Capacitance between two plates (trace and ground plane for example) is proportional to the area of the plates in parallel, and traces side by side with a ground plane on the same layer are coupled only by the thickness of the 1 ounce copper, NOT the width of the traces.

We do this all the time on low noise photoamps here at work. it works fine, no ground loops. And it's not really a problem with 2 or 4 sided boards either. Calculate the capacitance with 0.062 FR-4 separating two plates, for example (2 sided standard board). It's pretty small. I have in production 4 layer boards (0.015 inch insulating layers between 1 ounce copper traces, 2 signal layers with a ground plane and power plane) with 10 megohm feedback resistors in current-to-voltage converters where 1 picofarad is a big deal. These boards work fine, and the typical parasitic capacitance is on the order of 1/4 picofarad. Not enough to create much of a ground loop....

cheers

Adam

Rookie · 27th February 2004, 05:19 PM

Here is my PCB. According to Gyula, it is fool of ground loops too.

Adam Reed · 27th February 2004, 05:42 PM

That looks pretty nice, Rookie!

One other comment I forgot to add to the previous post: Ground loops are created when various parts of a circuit have different return paths to ground with significant resistances to those paths. In an extreme case, where a circuit drawing significant current causes a measurable voltage drop across a return trace and shares that ground return with, say, a low level analog signal, real havoc can result--oscillation, noise, etc. As far as the analog circuit is concerned, its ground reference is bouncing up and down as the load varies.

The main reason a fill, or ground plane, is used is so that all the ground return paths are through a large area and thus low impedance conductor, thus insuring that all return paths have a low impedance to ground. And still, if you have a significant current draw (hundreds of milliamps or more), you should have a separate ground plane for the high current circuit only, tied (as I described in the previous post) to the other ground planes where power comes in to the board.

There is a more elegant way to do this--a star ground, where **every** ground path has its own trace back to a single grounding point. I have a picture of a preamp designed by Dan Schmalle (Doc Bottlehead) where every signal or power path has a separate wire, and all the return wires lead to one single point on the chassis. It's definitely a 3-D rats nest and nearly impossible to do properly on a PC board...but technically speaking that is the BEST way to design a circuit. But then again that is why PC boards have ground planes instead...much easier to implement and nearly as effective.

cheers
Adam

00940 · 27th February 2004, 08:53 PM

If the first wasn't good, would this be better ?

Gyula · 27th February 2004, 10:31 PM

Hi Adam!

Had wrote that observation because I think the tracks and the ground plane are too close to each other.
The capacitance depends on the Area and the Distance of the conductors at a given permittivity.

Gyula

Gyula · 27th February 2004, 10:41 PM

Hi Adam!

The ground loop is coming from the parasitic impedance and capacitive coupling. These are the two terms of oscillation. This is not coming from the resistive voltage drop as you wrote, the current division is an another phenomeon. Every path has her own LC, and begin to oscillate with each other.

Gyula

00940 · 1st March 2004, 10:34 AM

no comment on the new layout ?

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Another PCM2702 DAC project.	Dxvideo	Digital Line Level	1	4th March 2009 07:05 AM
Problems with my PCM2702	Mikael Abdellah	Digital Source	4	7th October 2007 10:51 AM
Is anybody can help me with my PCM2702	burderly	Digital Source	4	5th August 2006 03:51 PM
Keces.com PCM2702	jmar	Digital Source	23	20th February 2004 04:08 PM
Pcm2702 Usb Dac	sugano	Digital Source	9	27th May 2003 06:37 AM

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

27th February 2004, 10:57 AM	#21
00940 diyAudio Member Join Date: Sep 2003 Location: Paris	I've done my homework. Here's a single sided pcb, analog and digital grounds connected. Power supply traces are 40mil, signal is 16 mil. Clearance is 8mil. Grid is 1 inch. I hadn't a correct layout for the inductor, so I used a cap layout. The computer ground is not directly connected to the groundplane but to the chip only.

27th February 2004, 02:29 PM	#22
Gyula diyAudio Member Join Date: Feb 2004 Location: Kunszentmárton, Hungary	Hi Rookie! In this way, everything is coupled capacitively to your ground plane, and it's Full of ground loops! Good luck, Gyula

27th February 2004, 02:31 PM	#23
Gyula diyAudio Member Join Date: Feb 2004 Location: Kunszentmárton, Hungary	Oops! I have to apologize from Rookie, I wrote the reply above to 00940.

27th February 2004, 03:22 PM	#24
Adam Reed diyAudio Member Join Date: Jan 2004	Gyula--I would have to say that with this single sided board the capacitive coupling from a trace to the ground plane is next to nil. Capacitance between two plates (trace and ground plane for example) is proportional to the area of the plates in parallel, and traces side by side with a ground plane on the same layer are coupled only by the thickness of the 1 ounce copper, NOT the width of the traces. We do this all the time on low noise photoamps here at work. it works fine, no ground loops. And it's not really a problem with 2 or 4 sided boards either. Calculate the capacitance with 0.062 FR-4 separating two plates, for example (2 sided standard board). It's pretty small. I have in production 4 layer boards (0.015 inch insulating layers between 1 ounce copper traces, 2 signal layers with a ground plane and power plane) with 10 megohm feedback resistors in current-to-voltage converters where 1 picofarad is a big deal. These boards work fine, and the typical parasitic capacitance is on the order of 1/4 picofarad. Not enough to create much of a ground loop.... cheers Adam

27th February 2004, 05:42 PM	#26
Adam Reed diyAudio Member Join Date: Jan 2004	That looks pretty nice, Rookie! One other comment I forgot to add to the previous post: Ground loops are created when various parts of a circuit have different return paths to ground with significant resistances to those paths. In an extreme case, where a circuit drawing significant current causes a measurable voltage drop across a return trace and shares that ground return with, say, a low level analog signal, real havoc can result--oscillation, noise, etc. As far as the analog circuit is concerned, its ground reference is bouncing up and down as the load varies. The main reason a fill, or ground plane, is used is so that all the ground return paths are through a large area and thus low impedance conductor, thus insuring that all return paths have a low impedance to ground. And still, if you have a significant current draw (hundreds of milliamps or more), you should have a separate ground plane for the high current circuit only, tied (as I described in the previous post) to the other ground planes where power comes in to the board. There is a more elegant way to do this--a star ground, where every ground path has its own trace back to a single grounding point. I have a picture of a preamp designed by Dan Schmalle (Doc Bottlehead) where every signal or power path has a separate wire, and all the return wires lead to one single point on the chassis. It's definitely a 3-D rats nest and nearly impossible to do properly on a PC board...but technically speaking that is the BEST way to design a circuit. But then again that is why PC boards have ground planes instead...much easier to implement and nearly as effective. cheers Adam

27th February 2004, 10:31 PM	#28
Gyula diyAudio Member Join Date: Feb 2004 Location: Kunszentmárton, Hungary	Hi Adam! Had wrote that observation because I think the tracks and the ground plane are too close to each other. The capacitance depends on the Area and the Distance of the conductors at a given permittivity. Gyula

27th February 2004, 10:41 PM	#29
Gyula diyAudio Member Join Date: Feb 2004 Location: Kunszentmárton, Hungary	Hi Adam! The ground loop is coming from the parasitic impedance and capacitive coupling. These are the two terms of oscillation. This is not coming from the resistive voltage drop as you wrote, the current division is an another phenomeon. Every path has her own LC, and begin to oscillate with each other. Gyula

1st March 2004, 10:34 AM	#30
00940 diyAudio Member Join Date: Sep 2003 Location: Paris	no comment on the new layout ?