#10953 NORM -: XO-1.75 External Mic sense does not work

[Zarro Boogs per Child]

#10953: XO-1.75 External Mic sense does not work
-----------------------------------------+----------------------------------
           Reporter:  wmb at firmworks.com  |       Owner:  wad    
               Type:  defect             |      Status:  closed 
           Priority:  normal             |   Milestone:         
          Component:  hardware           |     Version:  1.75-A3
         Resolution:  fixed              |    Keywords:         
        Next_action:  never set          |    Verified:  0      
Deployment_affected:                     |   Blockedby:         
           Blocking:                     |  
-----------------------------------------+----------------------------------

Comment(by wmb at firmworks.com):

 Replying to [comment:2 wad]:
 > There are two problems with this circuit.   The first is the voltage
 divider on the gate of Q9, which means that while we have 2.6V of bias
 voltage, we are only applying 1.3V of it to the gate.   This ensures that
 the output voltage will never be higher than 1.3V.

 I'm confused by this analysis.  I could see how that might be the case in
 a source follower circuit, with the source connected to a resistor to
 ground.  But in this case, the source is either grounded through the mic
 jack or pulled up to 3V3.

 It appears to me that the intended circuit operation is a grounded-source
 current switch, in which case 1V3 on the gate should be enough for
 sufficient LED current - although boosting it to 1V8 wouldn't hurt.


 > A second problem is the voltage drop across the LED, roughly 1.8V when
 no current is flowing.   Even correcting the voltage divider, it is hard
 to get a voltage that I would feel comfortable feeding into a 3.3V GPIO.
 >
 > In 1.5, the signal was current sensitive.   Either it was tied to
 ground, or it was pulled up by another pullup.
 >
 > The solution for 1.75 A3 should be to add a 47 Kohm pullup resistor to
 +3.3V to the EXT_MIC_PLUG signal.   I can't quite explain why the pullup
 has to be so low --- 15K pulls up to +3.3V, 120K only pulls up to 2.1V ---
 I suspect the input impedance of the SoC is relatively low (150Kohm ?).

 How about a Schottky diode with anode connected to Q9 source and cathode
 connected to EXT_MIC_PLUG?  That would prevent any reverse conduction
 through body diodes.

-- 
Ticket URL: <http://dev.laptop.org/ticket/10953#comment:3>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system