0

I noticed Everlight EAPLRAA4 receivers go into oscillation if there is no incoming light signal. Frequency depends, varies with the part, either 8..9MHz or hops to 32MHz. The supply is as clean as it can be - SMT 0.1u caps right on the pins of the receiver, ferrite bead on power, one more 0.1u cap after the bead. 22p is on the output as per the datasheet. Happens in a test breadboard setup and in production built audio switcher. When the incoming signal is present all works as expected. After the light is off, the oscillation builds up in less than a minute.

Has anyone seen this oscillation? Is this normal, perhaps specific to everlight?

barbazoo
  • 222
  • 1
  • 7
  • It’s likely due to the high Q of your Power source. With LC and too low Rs. Does the resonance match the filter? Is there a 30 pF load on the output? – Tony Stewart EE75 Jun 20 '21 at 23:36
  • There's 22pF load on the output. Happens with no ferrite bead as well, frequency is about the the same. There are 3 receivers on the board, they all go into different resonant frequencies (8.16Mhz, 7.88MHz, 32.3MHz). Power supply is 7805. – barbazoo Jun 20 '21 at 23:47
  • The usual debug method is to find how to change the resonant frequency from a driving point and then analyze the ability for positive feedback from stray reactance or resonant paths of mutual coupling and try to dampen the effects with resistance or suppression methods of isolation or attenuation or loading. – Tony Stewart EE75 Jun 21 '21 at 06:16
  • The same Everlight part is present in 3 builds - all with different layouts and different power supplies. All are going into oscillation ~8MHz after some time when no light present. I agree with @TonyStewartEE75 assessment in general, however the consistency of the behavior across 3 different designs and the different frequency when 3 receivers on the same board suggests it's a faulty part. I have ordered a different model to replace it. – barbazoo Jun 21 '21 at 15:00
  • Power // Gnd planes below IC prevent the parasitic inductance of >=10 nH which creates an Vcc impedance of >= 500 mOhms or so as 8MHz is below the expected unity gain breakpoint as 16MB/s can be achieved with max sensitivity and higher f starts to degrade this at 5V – Tony Stewart EE75 Jun 21 '21 at 16:38
  • Traces and part inductance is about 10nH/cm 2. To reduce the digital noises form the digital IC on the motherboard, the planar capacitance formed by an isolated Vcc and Gnd planes is critical. The POF device must be mounted directly on these two planes to reduce the lead parasitic inductance. 3. The isolated Vdd and Gnd planes must be connected to the main Vcc and Gnd (digital) planes at a single point using ferrite beads. The beads are used to block the high frequency noises from the digital planes while still allowing the DC connections between the planes – Tony Stewart EE75 Jun 21 '21 at 16:39
  • The Sharp/Vishay part should be better – Tony Stewart EE75 Jun 21 '21 at 16:41

0 Answers0