Design of PCB layout

I am going to use Crystal Oscillator and hence PLL is not used

What frequency range is allowed for crystals.

Would it be fine if I make a break out board similar to this... http://onlinetps.com/shop/images/Items_Images/ATMEGA128_Breakout_board.jpg obviously with correct pin count

If possible I would try to include the decoupling capacitors, crystal circuit and the ADC resitors on the break out board.

Would there be any problems?

Hi,

There is some PLL circuitry you should try to place as close to the ADAU as possible, see the datasheet. Also, depending on the lenght of the lines you use to control the ADAU from an MCU (IIC, SPI) and how sharp are the edges of those signals in your specific implementation, you might need to add some serial terminators on those lines.

Also, try to keep the analog lines (ADC, DAC) as close to extenal the analog circuitry as possible. Usually there are some HW design recommendations from ADI's datasheets, worth reading them.

Cheers,

JaaC

Even when you are using a crystal instead of a direct clock, the PLL Is still used. Your crystal clock frequency should be chosen to match the appropriate PLL mode, as shown in Table 12 in the datasheet. 12.288 MHz, or 256 x 48 kHz, is a common audio clock and for that setting you would set the PLL_MODEx pins to their 256 x fs settings.

If you are building the ADAU1701 onto a breakout board then all of the components you've mentioned - decoupling capacitors, crystal circuit, and ADC resistors - should absolutely be on that board with the SigmaDSP. Depending on your trace and connector length, then you may also need serial termination resistors on any digital data signals crossing the connector, as Jaime mentioned.

Thanks for the information.

Instead of breakout board I am thinking of implementing the circuit given in Figure 3 of the Selfboot note you have written with some changes.

I will remove the EEPROM connections and connect them outside the breakout board (just in case I need to change to microcontroller boot/USBi). Also I'd like to use BC557 transistor for the regulator circuit instead of FZT953. I am not sure of the transistor although it has enough power dissipation and current gain.

I would bring out all the analog/digital connections and booting connections outside and then to another board, which I can later modify as and when required, so as to test prototype circuits.

Would that be ok? Also what type of capacitors should I use? Or any type would do?

Page 17 of the ADAU1701 datasheet gives the required specs of the external transistor for the regulator circuit. h_fe should be at least 100, and it should be able to dissipate at least 90 mW (if the SigmaDSP core is fully-utilized). The BC556 transistor you mentioned meets these specs, so it should be okay to use.

Hi Jaime,

Would you care to elaborate on "serial terminatars" please - values, which end? I've built a 1701 board with a breakout ribbon which doesn't appear to be causing me any problems but maybe that's because it's short. The longest being about 4" from the chip. It would be useful to have some ball park values in case it has to get longer.

Cheers

Gordon

Hi Gordon,

The idea is to try to avoid reflections, which are caused by impedance missmatches between the source+line and the receiver end, and which at its time generates ringing, overshoot/undershoot -possibly compromising the electrical tolerance of the devices-, and EMI issues, besides the fact of probably not working reliably.

Electrons travel at a distance of around 6 inches per nanosecond, and the important part here is the rising/falling edge, more than raw switching speed, in other words, it is the same thing from a 10 MHz signal with 1ns rising/falling edge, as from a 1 MHz signal with same 1ns rising/falling edges. So, if for instance, your trace length is 6 inches or more and your rising/falling times are 1ns or less -faster-, then you are likely to have all the trouble mentioned above. In that case, source series terminators, which is, resistors placed as close as possible to the source end, will help you mitigate the mentioned problems. The resistors should have a value which when added to the impedance of the driving end -the signal source-, equals the impedance of the transmission line; this way the resistor will dampen or absorb the reflection from the receiving end and will mitigate the undesired effects.

If you are using connectors, then you also need to account for the connector impedance and other factors, but again, it all depends on the rising/falling edge and trace/wire length.

You can get started by checking the waveform you see at the receiving end with a high speed, high bandwidth oscilloscope, and see if you are experiencing any ringing, or if your signals are clean enough.

If you want to go deep into this topic I would recommend you these books:

- High Speed Digital Design: A Handbook of Black Magic by Johnson/Graham. recommended on every ADI DSP manual!

- Signal Integrity Issues and Printec Circuit Board Design by Douglas Brooks.

- High-Speed Digital System Design—A Handbook of Interconnect Theory and Design Practices by Stephen H. Hall, Garrett W. Hall, James A. McCall

Regards,

JaaC

Well, that didn't work too well. Tried replying from Thunderbird as per "

To post a reply to the thread message, either reply to this email or visit the
message page: https://ez.analog.com/message/7211#7211"; and wound up with a blank
message in the forum. Ah well.

Jaime Andrés Aranguren Cardona wrote:

Hi Gordon,

[...]6 inches or more and your

rising/falling times are 1ns or less -faster-, then you are likely to have

all the trouble mentioned above.

That's a useful benchmark, thanks.

In that case, source series terminators,

which is, resistors placed as close as possible to the source end, will help you mitigate the mentioned problems.

It was actually the "series" bit that puzzled me slightly. Is that assuming zero source and destination impedance? There don't appear to be any figures for output impedance generally. Is it a convention that if not stated it's close to zero?

You can get started by checking the waveform you see at the receiving end with a hish speed, high bandwidth scope, and see if you are experiencing any ringing, or if your signals are clean enough.

Having read a little further into the thread it seems parleam08 and I aren't doing quite the same thing. I'm only bringing out the GPIO's and then only as inputs so it probably isn't an issue in my case. All useful information though as I'm using the CLKOUT of the 1446 in the next prototype.

Much obliged

Gordon

This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

Thank you,
EZ Admin