PLL unlock problem - help needed!!!
I have an evaluation board of Hittite with hmc783 pll. When using the eval board on its own (with the attached software, controlling its registers by usb controller) I succeed in locking the PLL. Frequency is 11.5GHz, oscillator is 50MHz.
I've designed a different board, using this PLL (I wanted to try it with an eval. board first to see how it goes) but the PLL cannot be locked. Frequency is the same (11.5GHz), oscillator is 125MHz.
The PLL is fractional, but can be used also as integer. The s/w that controls it comes with predefined registers for integer and fractional type PLL so I can load it, also I can change almost anything on the PLL using the GUI of the PLL s/w.
Both boards (mine and the eval.board) are pretty much the same, from the PLL point of view. Avcourse mine has extra features, but regarding the PLL, the design is pretty much the same.
The eval. board has all different loop filters implemented in it, using resistors/capactiors not placed, i.e. defined as NA in the schematics, so one can choose to implement all types of loop filters, moving and replacing the caps and res as he wishes. I did the same on my board.
Both boards have an amplifier used for the active type loop filter. My amplifier is different than the one used at the eval.board, but the un-locking occures at passive filter and also at active filter, so this is not the reason.
When using the passive loop filter approach, I cannot lock, but I can see the LO signal in the spectrum at the right frequency. The signal looks bad. It's wide, about 8MHz bandwidth, and it's looked as if it is modulated or something. Has 2 peaks, and in between it varies like a sinus signal. I'll try to upload a picture later on. When I change the frequency using the s/w, the LO signal is changed to the frequency I changed, but it's still looks bad (~8MHz bandwith, etc.).
The s/w shows unlock state, probabely because it's not as it should be...
I measured the Vtune voltage with an oscilloscope on the eval.board, and wrote down the Vtune values for few frequencies. Then I disconnected the Vtune input to the PLL in my board and injected the same Vtune values from an external P/S. I could see a solid, stable, tone at the desired frequency on a spectrum analayzer, though I had a slight deviation in the voltage needed for the particular frequency. When I closed the loop in my board (disconnected the external Vtune injection), the PLL cannot be locked.
Any help will be granted, as I'm currently stuck here.
Thanks...
every couple months someone comes on here with same question on plls. You should search past troubleshooting recommendations.
In general the 1st thing to check is if the registers are getting programmed properly. Big difference between using debugged hittite software and your own controller...one bit off and you are lost.
Is big ground pad on bottom of PLL chip soldered? did you leave ANYTHING off between the eval board and your board, such as blocking caps, etc?
also, try to put the same reference oscillator as on the eval board. Sometimes PLLs are sensitive to noise and cannot lock properly. You can also try to connect the control lines from eval board to your board. If both suggested things don't help, there might be something really wrong - the chip that you use, or the design itself.
focusing on the last part of your post - why don't you try the same vtune filter and schematic as in eval board. sometimes the application engineers knows tricks from the kitchen that they do not publish, but if you call them they will say : use the schematic we provided as reference, it is "optimised" for our chip.
If you disconnect the VTune from the PLL, you couldn't get a solid and stable frequency output. The 8 MHz perhaps is the loop bandwidth of the PLL. You can place the first resisitor in the loop filter (looking from the CP) as a few hundred Ohms. If you have programmed it correctly, you will get the locked frequency output.