[sdiy] Digi-Mod schematic

[Seb Francis]

Hi Achim,

Many thanks for your comments.  My answers below ...


> page 2:
> - it looks like your schematic shorts C42
>   

No sure what you mean.  This cap is meant to provide a slowly rising 
!RST voltage to keep the CODEC in reset until everything else settles.  
The diode protects the !RST input when the power is turned off.


> - combination of 1uF/100nF in X7R does not make much sense unless you 
> can place the 100nF much closer to the IC. I don't expect the ESR and 
> resonances of these two values to be very different, combinations of 
> electrolytics and ceramics are a different story.
>   

I'm going by what the CODEC datasheet says.  The 100nF will be placed 
close as possible, but the 1uF will also be fairly close  The datasheet 
doesn't say anything about needing electrolytic or higher ESR for the 
1uF.  You think it will be better to have an electrolytic instead for 
the 1uF?


> - check whether you can hang the enable of X1 into the air or if it is 
> better to use a pullup
>   

It's ok for the oscillator I was going to use, but I guess I should tie 
it high in case other oscillators need this.


> page 3:
> - it is probably easier on the PCB layout if you use a 4 or 8 channel 
> TVS array, perhaps in combination with a resistor array.  The 4ch 
> devices are commonly used for USB and should be widely available.  
> Use schottky diodes if possible.
>   

Yes, I'll play with this when doing the layout.


> page 5:
> - since you do have 5V on board, I don't see why it shouldn't be 
> possible to drive the output from 5V as all other MIDI interfaces do.  
> The input to the LVTTL driver works nicely with just 3.3V swing.
>   

The +5V on board from U5 is used only for the clean VA for the CODEC so 
I preferred not to run any digital stuff off it.  Seems to me that 
switching a 5mA current loop could inject some glitches, no?

The 5V power input (regulated down to +3.3V) is not certain to be 
present and may be 15V for systems that do not have 5V power supply.

Only problem I could see with running the MIDI on +3.3V is when 
connecting to some (non-standard) device that draws its power from the 
MIDI bus.


> page 6:
> - it would be vastly easier to derive the 3.3V and 5.0V by a POL 
> switching regulator.
>   

I had given this some thought and decided I didn't want to risk the HF 
switching noise getting into the CODEC, but it keeps being mentioned, so 
maybe I should revisit the idea.

Is it really possible to get a clean power with minimum of components 
and zero experience with switching regs?  From what I've read they are 
fine for running uProcs, but to get rid of the switching noise, 
additional filtering and/or linear post-regulation is necessary.


> page 12ff:
> - I wasn't sure what you'd want to use the -1.65V for, but if you only 
> use it for level shifting to mid-supply, I would rather derive it 
> from the 3.3V directly via an inverting buffer.
>
>   

It's for shifting the -5V to +5V CV range into 3.3V to 0V for the ADC 
(see pages 12+)

Any reason why you'd rather use an inverting buffer to get the -1.65V?  
Doesn't seem any less components (would save 1 resistor, but have to use 
an opamp instead of the 3-pin voltage reference)


Seb