LT1328 powering & photodiode multiplexing

Hi Pavel:  I don't see a supply rejection vs frequency plot in the data sheet.  I never had any supply issues with it over my 24 years of looking at the part.  But the input stage is quite noisy, and it sometimes spews output junk because the noise is overcoming the intentional offset.  The way to increase the offset is to put a weak pullup on the BIAS pin to 5V; 3.01Meg is a good start at overcoming its own noise.  In noisy environments (like automated test equipment) some customers have experienced grief.  But in those cases, success was always achieved simply by strengthening the pullup current.

The problem with adding extra stuff on the front end is that it adds capacitance.  The demo board comes with a fairly large photodiode, 70pF as I recall.  And it's a wide angle photodiode.  If you used multiple SFH213 with high optical gain and low capacitance, you might get good performance.  But in your case, it would be the MUX capacitance in parallel with one photodiode at a time.

By the way, I have successfully used the DN399 technique to reduce input noise on LT1328.   Are you planning to do 4ppm?  Using full rate 4ppm, the LT1328 becomes sensitive to different power levels at different codes.

Hi Pavel,

The LT1328's high-pass filter rejects low frequency noise, however the pin 1 bias voltage is still susceptible to drift from variations to the supply voltage.  Please refer to second TOC on page 3 of the datasheet (Change in Voltage on Pin 1 vs Supply Voltage).  High frequency noise on the supply can lead to jitter on the TTL signals.  If a low noise LDO is not used, I would recommend making sure that the supply pin bypass filter caps can handle any high frequency noise on the supply line.

Regarding multiplexing, I would recommend multiplexing the digital outputs rather than the analog inputs.

Regards,

Omar

Thanks Glen,

The application isn't infrared, it's in visible band. The coding isn't 4ppm, but S/PDIF (for transmitting audio), the datarate is about 1Mb/s. The photodiode we plan to use isn't selected yet, there are about 15 in the shortlist with 0V terminal cap varying in range 50pF ... 1500pF. According to the simulation in LTSpice, 1500pF is still Ok. That's why there was question on analog mux (in order to avoid using multiple LT1328 to test different photodiodes). Concerning powering, at your glance is it Ok to use one 500mA normal LDO (without particular noise ratings) for powering multiple LT1328 (perhaps filtering power interconnections with ferrite beds) ?

Hi Omar,

Thanks. So you think it would be wise to use low-noise LDOs ... like these ones (LDLN025 from ST and LT3045 from Analog Device) ? What about one LDO per one LT1328 ... or one LDO for multiple LT1328 will be also Ok ?

ONojan said:

multiplexing the digital outputs rather than the analog inputs

Ok

Sincerely,

Pavel

Hi Pavel,

Adding LDO will help to improve the noise performance. Whether using one or eight LDOs in your new design depends on your noise requirements for your system and the noise level from your supply. One LDO can supply eight infrared receivers but might not be optimal for noise performance if their locations are far away from each other.

Best,

Jane

Hi Jane,

Thanks. Eight LT1328 (as well as theirs photodiodes) will be quite close each other on PCB. Perhaps using one LDO in star configuration (i.e. VCC of LT1328 goes directly to LDO output) for eight LT1328 using ferrite beds for isolation will be Ok ?

Hi Pavel, ferrite bead(s) can be added if the system is required, typically placed before an LDO. 

Hi Jane,

I thought place ferrite beads between LDO output and VCCs of eight LT1328.

Hi Pavel,

An LDO is sensitive with a ferrite bead. If you need support for LT3045, we can forward your question to Power Management forums. However, it doesn’t hurt to add footprints for ferrite bead(s) on your prototype board and verify whether you need it or not.