LTC5569 IF output options

Question

The LTC5569 has a differential IF output while I need a 50 Ohm unbalanced line. I can work with IF frequencies in the 400-800 MHz range and don't really have a problem with anything in there. My IF bandwidth is 10 MHz and I was considering using a SAW filter at 782 MHz, but this is not a hard requirement. 
 I would like to understand what my options are with respect to the output of the mixer. 
 Options I see: 
 1) Use the TC8-1X transformer from the reference design - this is easy but bulky and expensive 
 2) Follow the suggestions on page 15 of the datasheet to change the required transformer from 8:1 to 4:1, then use a 200:50 Ohm balun in a chip like the Murata DXP18BN5014TL, replacing the discrete RF transformer with it. 3) Use a discrete SMD RLC network. 
 Questions! 
 Is 2) even possible? I have never done it before. It does look very appealing. Is there any reason why the DXP18BN5014TL shouldn't work as a replacement for the bigger RF transformer? It is lacking a DC feed, but maybe DC blocking the inputs of the transformer might work? Alternatively the 1720BL15B0200 seems to have a DC-feed port! I didn't find much to go on about this topic online unfortunately... 
 About 3), this seems like a pain mostly due to tolerances and IF bandwidth. The IF bandwidth might be fine as it should double and be around 20-30 MHz based on my pixel counting on the datasheet, but I believe that with 2% tolerances I would have very bad yield (0.02 * 800 MHz = +- 16 MHz, that seems to way too big to be usable with such narrow IF bandwidths). 
 Still about 3, I couldn't get the formula in the datasheet to give results matching the table. From page 16, L1 = (1/(2 * Cif * Wif * Wif)). Cif should be given on page 14, that is 1.3e-12 for 380 MHz, Wif is 380e6. Plugging those numbers in gives (1/(2 * 1.3e-12 * 380e6 * 380e6)) or 2.66 uH, a very long way off from the suggested 68 nH. 
 Thank you and have a nice day.

ADIApproved · Accepted Answer

Hi epos2 , 
 It's not totally clear to me what center frequency you'd like your IF to be at. You mention at least 3 different IF frequencies: 
 
 Anything between 400-800MHz 
 782MHz 
 380MHz 
 
 Responses: 
 1) Use the TC8-1X transformer from the reference design - this is easy but bulky and expensive 
 This is the easiest solution, as shown on the datasheet. It's very flexible, wideband, and high-performance. 
 
 2) Follow the suggestions on page 15 of the datasheet to change the required transformer from 8:1 to 4:1, then use a 200:50 Ohm balun in a chip like the Murata DXP18BN5014TL, replacing the discrete RF transformer with it. 
 I don't see any reason that the DXP18BN5014TL wouldn't work for this application. The only thing to be aware of is that it may not be able to handle having DC current run through it. If not, then the bias current for the IF pins of the LTC5569 will need to be provided using the tuning inductors (shown as L1/L3 on page 14) and DC blocking capacitors should be inserted between those and the transformer. You may want to request some s-parameters for the transformer and use the ones in the LTC5569 datasheet to simulate the IF match before building up a board. 
 
 3) Use a discrete SMD RLC network. 
 This approach could work for you, but you've identified some of the drawbacks to using a lumped element balun. They have their place, but can be a little bit finicky. Alternatively the 1720BL15B0200 seems to have a DC-feed port! I didn't find much to go on about this topic online unfortunately.. 
 Similar to the DXP18BN5014TL, the 1720BL15B0200 should work as well. This part is rated to have DC current flow through it, so it can be used without much modification. Again, it would be prudent to get the s-parameters from Johanson and simulate the design before building up a board. There are many SMD baluns that are similar to the ones mentioned, but most will be at comparatively higher frequencies due to their small size. Check with Murata, Johanson, Mini-Circuits, Anaren, MACOM, etc. for other possible options. 
 Still about 3, I couldn't get the formula in the datasheet to give results matching the table. From page 16, L1 = (1/(2 * Cif * Wif * Wif)). Cif should be given on page 14, that is 1.3e-12 for 380 MHz, Wif is 380e6. Plugging those numbers in gives (1/(2 * 1.3e-12 * 380e6 * 380e6)) or 2.66 uH, a very long way off from the suggested 68 nH. 
 I believe the error you're making here is that you're only inputting 380MHz for ω when you need to multiply it by 2*π since ω = 2*π*f. Making this change gives you a value of 67.47nH. 
 
 Best regards, 
 
 Weston Sapia 
 Senior RF Applications Engineer 
 Analog Devices