10:18 <azonenberg> So, browsing the amplifier section of Analog Devices

10:19 <azonenberg> I see some really cool ultra wide bandwidth RF amplifiers

10:19 <azonenberg> But most/all of them are single ended

10:21 <azonenberg> Look at the HMC460LC5 for example

11:24 <azonenberg> The ADL5580 looks good but the datasheet is a bit iffy regarding common mode ranges

11:24 <azonenberg> Making an ac coupled diffprobe is not beyond the realm of plausibility

12:21 <azonenberg> lain: So i spent a while more thinking about how to make the AKL-AD1 better

12:21 <azonenberg> My new design calls for a cascade of two amplifiers and is AC coupled

12:21 <azonenberg> First we have an ADL5580 to convert the differential input to a single ended output, and terminate the other ADL5580 output

12:21 <azonenberg> This will then get fed into a Mini-Circuits EHC-24L+ single ended RF gain block

13:11 <azonenberg> https://www.antikernel.net/temp/ad1-sim.png S-parameter sim of the proposed amplifier cascade

13:11 <azonenberg> including the actual ac coupling cap i plan to use, measured data from an AKL-PT2 standing in for the differential probe input

13:11 <azonenberg> then an ADL5580 -> EHC-24L+

13:11 <azonenberg> fed by a TCBT-14+ bias tee

13:12 <azonenberg> If i can design a bias tee that goes closer to DC without losing high speed performance that would be great but i feel like beating something available COTS from minicircuits will be hard

13:13 <azonenberg> The bias tee issue will go away, and low-end bandwidth will drop to a few kHz instead of 10 MHz, if I'm OK with a lot less gain

13:13 <azonenberg> Without the second amplifier i'd be at about -15 dB

13:14 <azonenberg> The other option of course is to cascade two ADL5580s to get more gain

13:16 <azonenberg> Which might actually work nicely

13:17 <azonenberg> this eliminates the need for the bias tee

13:20 <azonenberg> With two ADL5580s, I get about -3 dB of nominal gain out to 3 GHz, then it starts peaking. But i suspect the peaking will be largely compensated for by losses elsewhere in the system

13:20 <azonenberg> That, and too much gain is an easy problem to solve with a LPF of some sort

13:24 <azonenberg> Actually wait, i forgot the 5580 had adjustment for peaking

13:30 <azonenberg> With Cpeak=3 and two cascaded 5580s, I get decently flat response out to 4 GHz then it bumps up just a little but i think pcb losses in the amp will eliminate that