Analog HFBR amplifier

[TMaxElectronics]

I have noticed quite a few new old stock analog HFBR receivers popping up on ebay for cheap recently, and decided to try using them instead of the TTL ones usually used for interrupter inputs.
They can't just be used as a drop in replacement unfortunately, as they have an off state voltage of around 2V and (depending on the received signal strength) a 500mV higher on state level, but a simple transistor circuit can turn that into a 0-5V inverted TTL output:


just thought I'd put this here incase somebody else wants to use them

[USAMAN]

Is this applied to the optical module?

[TMaxElectronics]

How exactly do you mean? The circuit gets connected between the output of the module and the circuitry that is designed to take the normal receivers.

[Max]

Hi everyone,


Since I've got a bunch of those receivers, too, I'm in the same situation. TMaxElectronics' schematic misses an important bit: The receiver needs 10R resistance between its VCC pin and 5V. More specifically, this resistance should be a combo of 2 100nF caps and 2 4R7 resistors. Apparently the HFBR-2416 is quite noise sensitive (or more specifically, it's integrated preamp). Broadcom (Avago) even recommends inductors in the supply and in the ground line. For details, have a look at C1/C2 and R1/R2 in the schematic below

That aside, I had a look at their application notes and what they recommend for "arbitrary duty factor data" (meaning the signal can be anything from 0% to 100% duty for any amount of time) and turns out, there's a dedicated document just for this: AN-1121. Based on that I designed my own adapter board. With 25mm x 18mm it should be small enough to work as drop-in replacement on the common UD2.7 boards. It takes up the space reserved for the second optical receiver (crude paint preview attached).
It uses a comparator configured as edge detector. This makes it suitable not only for interrupter signals of any pulse width, but also for standard serial transmissions up to a few Mbaud/s without having to use manchester encoding or similar techniques (yes I know they have other advantages and if you already use them, you probably have no reason to stop doing so).
Since the requirements for this application are lower than what the appnote required it could probably be simplified, removing some of the passives around the comparator.

I'll order and test the circuit soon and report back whether it works or not.


Kind regards,
Max

[klugesmith]

Looking forward to seeing your results.  Please tell us about your circuit board fabricator.  Can you give us some numbers for the amplitude of AC-coupled HFBR output, and the hysteresis voltage of the comparator circuit?  Thanks!

[Max]

Hi klugesmith,

I looked at JLCPCB and Aisler for manufacturing the (two layer) PCBs:
JLC: 25 PCBs for 5,24€ + 12,11€ FedEx = 17,34€ and about 10-15 days til delivery.
Aisler: 24 PCBs 12,48€ + 9,51€ UPS = 21,99€ incl. VAT and 11 days til delivery.
With only 5€ difference I decided to go for the more local company, Aisler. I don‘t quite understand why their only shipping options are UPS and UPS Express; with let‘s say DHL for <=5€ they‘d be on par with JLC. Granted, they‘re only competitive for non-custpmized 2-layer PCBs. 4 layers is at least double the cost and you still don‘t have nearly as many options than you get with JLC - which offers 4 layers with no extra costs plus many customizations.
The PCBs haven‘t arrived yet, so not much more to tell.

As for the HFBR-24x6Z receivers, most of your questions are answered by the datasheet, page 22.
The receiver’s output voltage sits somewhere between 1.8V and 3.6V. On top of that you get your signal which is amplified such that you get about 7mV output per 1uW optical input (min. 5mV/uW, max. 10mV/uW). From the same datasheet we find that the matching transmitters + cables will typically give you 63uW of optical power. This results in a 440mV (315mV-630mV) signal on top of the 1.8V-3.6V offset.
Once I get the thing I can actually measure it but I don‘t expect any surprises.

Edit: Hysteresis I don‘t know. I pretty much copied the circut from their appnote which contains surprisingly little info about it.
The circuit can however be simulated in LTspice.


Kind regards,
Max

[Max]

Ran the simulation again, looks like 5mV hysteresis.

Kind regards,
Max

[klugesmith]

Very nice.  I'd never thought of AC coupling + hysteresis as a way to pass digital signals with arbitrarily low frequency.  It depends on having hysteresis much larger than the channel's noise level (RMS).  I guess many applications can tolerate rare glitches, even output toggles that stick, due to noise.

[Max]

PCBs and parts arrived today. The short reply: it works. But...

PCBs are mostly fine; see other thread. Schematic and layout work, too, though I did goof up the silkscreen in one place; I attached the fixed version. The downside is that the output voltage of the comparator (I used a TL3016) only ranges from 0.7V to 4V. That should still work with most logic chips but it's on the edge; 74HC series f.ex. requires Vlow to be <0.8V.
The reason is R8+R9 which put the outputs under a constant load of about 8mA. I didn't think too much about this since these values come straight from the appnote but for the TL3016 that's enough to raise/lower the output levels as measured - and following the LTspice simulation the same is true for the LT1016 (the part used in the appnote). Since the appnote doesn't say anything about where these values come from and how they need to play together I'm not entirely sure how much I could raise R8 and R9.

Before raising your eyes about the solder joints, it looks worse on the picture than in reality and it was my first time soldering 0805 sized SMD components. With enough flux and a wide chisel (so that you can heat up both pads at the same time) it's not too hard - and much faster than THT.

I attached two versions of the PCB, one gives an inverted output (which matches the open collector of the digital HFBR receivers), the other gives a non-inverted output.


Kind regards,
Max