LOG114: CSO?

Hi Stephan, 

Yes, the LOG114 is manufactured under multiple CSO. This information is reflected in the datasheet to better clarify the changes that occurred from the product change notification announced back in 2025 (PCN20250326006).

We are happy to review your application if you have any concerns on the internal op amp implementation. 

Best regards,
Ashley

So my application is fiber optics testing equipment... My industry deals with a lot of cables and connectors that have multiple fiber optic strands in them, and the purpose of our product is to test all of them at the same time to save technicians time in the field. One of our tests is ensuring that the connectors are terminated correctly and nothing is swapped, so the transmit side of the tester modulates a short message on to each of its outputs to identify each fiber. The receiver side needs to measure the optical power and also demodulate the message. We've been cheating and using SFPs to do this but we're now looking to reduce cost and board area, and improve measurement accuracy.

The actual signal is manchester code at 2 MBaud. This was a compromise between how fast our embedded microcontroller could transmit and receive, and how slow an SFP (designed and intended for gigabit signals) could transmit and receive. We have to maintain compatibility between generations of our product so this requirement is now fossilized.

So what I've got going on here in this schematic is I've got an analog mux (elsewhere in the schematic) selecting a photodiode to get amplified by a TIA, then another op amp low-passes that signal to get an average power level, and a comparator extracts the modulated signal from that. Meanwhile the lowpassed signal goes to a log amplifier, so the output of that can go hit an ADC on my microcontroller.

I've discovered in simulation that I don't want to simply use the log amp as the TIA, because when you log a modulated signal, the logged output has much lower troughs below the actual average power, than the peaks are above the average power, so the low-passed version of the logged signal will read lower than the actual average power. So I actually want to low pass the signal before logging it.

SPICE tells me that the 15MHz GBW of the lower-performing version of the part should be just good enough to pass a 2MBaud manchester code... Good enough for the comparator to recover it afterward. But I'll take any advice you have.

 Gerasimos Madalvanos The LOG200 looks like a great part but it is double the cost of the LOG114... If I need to use it I will but I've got enough sensitivity to cost in this project to prefer the 114 if I can use it.