Hi All,

I am looking for some advice (or an example circuit) for the design of an analog high frequency LED and driver circuit design. I am hoping to build a fibre optic link to enable noise free measurement of high frequency voltage signals from the low voltage output of a capacitive divider. The signals are between +-10V with a bandwidth of 1-50MHz. It is important that I replicate the input waveform shape, hence why I would prefer to avoid ADC. I have initially had some trouble in finding a suitable LED due to the rise and fall times so possibly a laser is required. My electronics knowledge is a little rusty so please be kind . Any help would be greatly appreciated.

Jon

 

Advice - look for datasheets and Application Notes for HFBR1528/1528 and 2523/1533 (Avago)
They has wonderful optomech connector to and from free choiced glass fibre, and are fast enough let (as in my case) transfer accurately the 10 ns strobe pulse over 50 meter distance in ultra-high EMI (from laser operation) space with sub-ns accuracy. Cheap, simple, powerful.
Circuitry - for receiver I used the 74LS136N straight to the feet 1 of HFBR2528, what shivered 74HCT75N and then other quarters of 136N was used as gate drivers. However, with eyes of today, probably, I would use better instead the 74AS02D and SN74LVC2G74 - what are about one order faster. For small gate driver at such high speeds sorry there are no many good alternatives, I was able to found exclusively one ISL55110 but it works well only up to 1000pF what is sure too small. So I stacked it parallely and provided with pushpull booster cascade.
For transmitter - use just a bjt with speed characteristics You need, only be warned - LED there inside burns-off in miliseconds - Your signal may not be uninterrupted. If it may be such, then current must be harshly diminished, thus there the speed characteristics will be lost (see datasheet). Therefore I would say this is a heavy argument to shift things toward the D-class amplifier ideology in the entrance of system.
So, if You need a correct shape of signal - HFBR is able to work in linear mode, but then how to manage the outcoming signal after it`s genuine opamp? One way is to use the comparators to chop signal into uniform fragments what may be later conditioned digitally. Then the keywords ought be AD96687 and ADCMP553B - both are sub-nanosecond players and cost surprizingly tiny, only one-digit price. Or other way is to use that tiny brilliants 2N2222 what are able to work rather near to GHz range. Or just jump into SMD pool where there are plethora of better GHz candidates up to, if needed, 45 GHz. I mean using the lineary range of transmitter LED.

 

See

 

See
View attachment 141683

Thank you for the example circuit, very much appreciated, I will test it out. Just one quick question, is that a standard/downloadable spice model or did you create it yourself? I have searched but couldn't find it. I am just a little confused as to what library it will be placed in as the parameters Pnom Inon Ith etc. are not common to those listed for diode models (for example .model LXM2-PL01-VFBin_C D(Is=9.9054E-16 Rs=0.3672 N=3.2997 Iave=350m mfg=Lumileds type=LED)). Apologies for my inexperience, I had used spice only briefly in my undergrad several years ago however I am pretty sure I just selected components from a standard library.

 

Advice - look for datasheets and Application Notes for HFBR1528/1528 and 2523/1533 (Avago)
They has wonderful optomech connector to and from free choiced glass fibre, and are fast enough let (as in my case) transfer accurately the 10 ns strobe pulse over 50 meter distance in ultra-high EMI (from laser operation) space with sub-ns accuracy. Cheap, simple, powerful.
Circuitry - for receiver I used the 74LS136N straight to the feet 1 of HFBR2528, what shivered 74HCT75N and then other quarters of 136N was used as gate drivers. However, with eyes of today, probably, I would use better instead the 74AS02D and SN74LVC2G74 - what are about one order faster. For small gate driver at such high speeds sorry there are no many good alternatives, I was able to found exclusively one ISL55110 but it works well only up to 1000pF what is sure too small. So I stacked it parallely and provided with pushpull booster cascade.
For transmitter - use just a bjt with speed characteristics You need, only be warned - LED there inside burns-off in miliseconds - Your signal may not be uninterrupted. If it may be such, then current must be harshly diminished, thus there the speed characteristics will be lost (see datasheet). Therefore I would say this is a heavy argument to shift things toward the D-class amplifier ideology in the entrance of system.
So, if You need a correct shape of signal - HFBR is able to work in linear mode, but then how to manage the outcoming signal after it`s genuine opamp? One way is to use the comparators to chop signal into uniform fragments what may be later conditioned digitally. Then the keywords ought be AD96687 and ADCMP553B - both are sub-nanosecond players and cost surprizingly tiny, only one-digit price. Or other way is to use that tiny brilliants 2N2222 what are able to work rather near to GHz range. Or just jump into SMD pool where there are plethora of better GHz candidates up to, if needed, 45 GHz. I mean using the lineary range of transmitter LED.

Thank you for your detailed advice. I have actually been looking at the HFBR series of transmitters/receivers, however even the 125MBd (HFBR15X7Z) version has an optical rise time of 12ns, which I originally assumed (without knowing a great deal about the characteristics of these components) would result in some sort of distortion of the waveforms I wish to send, having rise times as low as a few ns? The HFBR range you mentioned has a 10MBd rate, did you have any issues with that for 10ns pulses?

 

Thank you for the example circuit, very much appreciated, I will test it out. Just one quick question, is that a standard/downloadable spice model or did you create it yourself? I have searched but couldn't find it. I am just a little confused as to what library it will be placed in as the parameters Pnom Inon Ith etc. are not common to those listed for diode models (for example .model LXM2-PL01-VFBin_C D(Is=9.9054E-16 Rs=0.3672 N=3.2997 Iave=350m mfg=Lumileds type=LED)). Apologies for my inexperience, I had used spice only briefly in my undergrad several years ago however I am pretty sure I just selected components from a standard library.

I strongly doubt that the LED will operate at a frequency greater than 1MHz (not to mention 50MHz). The model, the LED that you showed is defective (there are no parameters that determine the frequency properties (inertia)).
I made the laser model myself. You can download all my models. Look I strongly doubt that the LED will operate at a frequency greater than 1MHz (not to mention 50MHz). The model, the LED that you showed is defective (there are no parameters that determine the frequency properties (inertia)).
I made the laser model myself. You can download all my models. Look
https://forum.allaboutcircuits.com/...nents-models-of-ltspice-free-download.133690/

 

I strongly doubt that the LED will operate at a frequency greater than 1MHz (not to mention 50MHz). The model, the LED that you showed is defective (there are no parameters that determine the frequency properties (inertia)).
I made the laser model myself. You can download all my models. Look I strongly doubt that the LED will operate at a frequency greater than 1MHz (not to mention 50MHz). The model, the LED that you showed is defective (there are no parameters that determine the frequency properties (inertia)).
I made the laser model myself. You can download all my models. Look
https://forum.allaboutcircuits.com/...nents-models-of-ltspice-free-download.133690/

Thank you Bordodynov, both for your advice and your contribution with all of the models you have freely provided. I look forward to using them now and in the future. I already had doubts about the effectiveness of LEDs at the speeds I require. I will instead focus on laser diodes.

Thanks again.