UWB Pulse Generation (Avoiding SRD)

I'm working on a project that will use UWB pulses for time of flight applications. I was planning on using a step recovery diode as many papers discuss using a SRD circuit. However, SRDs are a little hard to come by (can't just order them from digikey), and so I'm starting to give thought to other ways to generate UWB pulses.

In my receiver, I was planning on using an ADCMP572 comparator from Analog Devices to detect when the UWB pulse arrives. It is able to detect a minimum input pulse of 80 ps and has a 35 ps output rise/fall time. Would it be possible to generate a UWB pulse by using this comparator in the transmitter? If I send a 10MHz pulse to the comparator's P input and a 100ps delayed version to the N input, I would have the comparator output high for just 100ps until the delayed pulse reaches N and sets the output low again.

It sounds like that would work, but the ADCMP572 outputs in CML. I'm not very familiar with CML. Can I just slap a capacitor at one of the CML outputs (which has a 400mV rise but it's centered around VCC) to get a positive AC pulse centered around GND and feed that into a differentiator (to get a monocycle) and then an amplifier and antenna for broadcasting?

Any advice or suggestions about UWB are appreciated.

the adcmp573 shows the output as having two independent emitter follower stages. So you can use pull down resistive loads and get a single ended output from each pin. You do realize that the voltage swing will only be .6 v pk-pk, so it will not go very far as a transmitter. I suspect that the unbuffered adcmp572 would have some odd ringing/time delay problems if you misbalanced the output to be single ended.

You might do a little better using both outputs and some sort of coplanar feed uwb antenna. That way you will get a full 1.2 v pk-pk voltage swing into the antenna.