That dip on the high side of the passband was bothering me. The GPLA crystal design program predicted it, and indeed, when I measured the performance of the actual filter, there it was. The rig was working fine -- the receiver sounded fine and everyone tells me that it sounds great on transmit. But still, it bothered me.
So I started working out with the various crystal filter software packages.
This filter was -- sort of -- a Cohn Min-loss filter, but I had built it with four crystals and three shunt caps (80 pf each) and no series caps at the input/output. This morning I decided to see what would happen if I put the series caps in there. Here is what Wes's GPLA predicted:
Wow, that looked a lot nicer. And the 150 ohm terminations seemed to be just about perfect for the BITX design -- no need for impedance transformation. I heated up the soldering iron and searched the junkbox for suitable caps (I found 2 82 pf caps -- close enough).
Here is what the results looked like (I didn't measure insertion loss so the top of the curve is just the peak of the response curve).
.JPG)
Exactly as predicted! Thanks Wes! And thanks to Farhan for encouraging me to characterize my crystals and to use the available software
Now I have to go back and de-ripple the 5 MHz filter in my BITX 17.
Our book: "SolderSmoke -- Global Adventures in Wireless Electronics" http://soldersmoke.com/book.htm Our coffee mugs, T-Shirts, bumper stickers: http://www.cafepress.com/SolderSmoke Our Book Store: http://astore.amazon.com/contracross-20
Posts
You just measured voltages with your scope again didn't you? Did you take the log of the voltage readings for the y axis?
ReplyDeleteNice looking response! Looks like very minimal ripple.
Of course, if it turns out you have a huge insertion loss we'll all mock you (just kidding).
Bert: Yes I put a signal from my DDD signal source through the amp that precedes the filter, the filter, and the amp that follows. I had 6 db pads (50 ohm) at either end. I measured the voltage out on my Rigol digital scope, then converted to db down from peak using a spreadsheet. I suspect the insertion loss would also be as predicted by GPLA. 73 Bill
ReplyDelete