LMX2492: Cannot change VCO Frequency or start ramping

Darren,

Sounds like you just want to start with a locked PLL, so lets go for that ...

1. Establish communication -- Done
The fact that you can send serial commands to power the device up and down and toggle I/O pins establishes that you are able to send commands to the device and the device is able to understand them.

2. Load default mode
We're in the debug process, no need to add any additional factors until you can get the PLL to lock.

3. Check Input Frequency -- I think this could be the issue
Ensure that you have 100 MHz input frequency. Yes, there is an XO on the board, but it is NOT powered up or hooked up by default. Ensure you see the 100 MHz input at the pin. So start with a 100 MHz input to the OSCin input SMA and verify that it is making it to the pin on our device.

4. Check VCO Output
Realize that the LMX2492 does not create an output, but only steers the VCO output. I don't know the exact range of the VCO, but it is wider than 9.4 to 10.8 GHz, as they have to have some margin. There is also a divide by 2 output. So if you see 4.46 GHz at the divide by 2 output, this means 8.92 GHz at the VCO, which sounds like the VCO is producing its lowest output frequency. I would check the VCO tuning voltage and I bet it's 0V.

5. Check VCO tuning voltage
Now try to make the VCO frequency and tuning voltage move. Try toggling the phase detector polarity on the TICSPro GUI from "-" to "+". This typically makes the VCO tuning voltage to go to the maximum value and also I would expect the VCO frequency to move. If you don't see the tuning voltage move, something is suspicuous. Maybe the loop filter is disconnected, shorted to ground, charge pump is tri-stated, ect.

6. Read back the N divider and R divider outputs.
On the MUXout (or other I/O pins), you can see the PLL R divider and PLL N divider. On this pin you can see the output of the N and R dividers as the chip interpretes them to be. Ensure that this frequency is correct and that it is not too low (sensitivity problem) or too high (sensitivity or hamrmonic problem)


Regards,
Dean

Dean,

Great to have your help!

I did 2. For 3 I see power going to the XO chip, and I can measure Voltage on the output of the XO, so I think I'm ok there.
For 4, ya, I was thinking something like that. I measure Vtune as .6V and it doesn't change even when I set the GUI VCO frequency at something different.
5: Toggling the CPPol doesn't do anything, which I believe is what you meant. So now we're getting somewhere. How would I test those other possibilities you mention?
6: I'm not sure how to interpret what I should see on these outputs. Whereas a flag has either a High or Low state, the dividers should have some sort of number. Am I looking for a count on the pin, that is, P number of pulses to indicate the R/2 is at value P? What would I be triggering on to see this. Or maybe more basically, what mode should the I/O pins be in for me to be able to "see the output" of the dividers?

Thanks,
Darren

Darren,

I'm assuming you are using our TICSPro software, but it would be good to know.

Here's some things:

After setting default mode, cycle power, toggle SWRST bit, and push Ctrl+L to make sure that we start off in a known state.

2: OK. As we are missing something, we can be absolutely sure of this with step 6. So lets go to the next one.

6: For this one, we want to see what signal the part believes is at the OSCin pin. To do this, do the following:
a. Set MUXout_MUX = "Output R/2" state 16
b. Set MUXout_PIN = "Output PP" State 2

The output is the frequency after the PLL R divider then divided by 2 again (divide by 2 makes a nice square wave); it's not a number of fixed pulses as you say above.

Now look at the MUXout pin on the oscilloscope and you should see half of hte input frequency. So if the reference is 100 MHz, you should see a 50 MHz square wave. If you a higher multiple, it could be a haronic. If you see nothing or noise, it could be a sensitivity problem and this implies the signal is not getting to our pin. Realize also that if you have no signal going to this pin, it can self-oscillate, although it will be way off from the target freqeuncy. If you are absolutely sure that there is indeed a signal at the OSCin pin, yet the part says nothing is there, consider some ideas. First of all, signal should be AC coupled and bias level on OSCin pin should be Vcc/2. If it is some issue with the pin, try driving the other compliementary output pin and see if there is some issue with the pin.


5: Now assuming that we are sure we have the reference and the device sees it, focus back to the charge pump. Try toggling the charge pump polarity and then try an output frequency of 100 MHz and 20000 MHz and see if this moves the VCO. If not, consider these possibilities:

a. Loop filter is shorted to ground. If so, you should see impedance =0 and short on board. If you see current to the board increase when you change the charge pump current, this could be an indicator.
b. Loop filter is disconnected from VCO. This would mean bad connection for resistors R3_LF or R4_LF on the board.
c. Charge pump Pin is damaged or shorted to neighboring pin. Hard to prove this, but if it doesn't toggle, something is off.
d. If you did anything fancy with the loop filter or used any kind of active filter, then this could be an issue.


REgards,
Dean

Dean,

I've asked Jeston to set up another phone call, because I'm not getting anywhere.  Hopefully we can debug it over the phone.  

Yes I'm using TICS Pro.  I loaded all the defaults like you instructed.

I redid step 3 from your earlier post, and found there is no clock coming out of the XO on the EVM board.  Power is going in, but no clock.  I cannot tell why it isn't connected, on the schematic and the board itself it seems connected.  Ok, so I connected a 100 MHz clock into the OSCin SMA like you suggested.  Then I connected my scope to the MuxOut pin, set to R/2.  

I KNOW I'm sending in a 100 MHz clock, but I can't see anything on the MuxOut R/2 pin.  I have tried it on the other I/O pins as well.  I also connected the pin to my Spectrum Analyzer and could not so the signal at 50 MHz.

So somehow the board cannot see the 100 MHz signal properly.  Hopefully a phone call can clear this up.  
Talk to you soon,
Darren

Darren,

If you cannot get the right frequency out of the R divider "Output R/2" from the MuxOut pin, then this would cause issues getting the right charge pump voltage. Although you are putting a signal to the OSCin pin, I don't trust it is getting into the part. If anything is blow, it sounds like the OSCin pin.

To test the VCO path, try this:
1. Toggle the VCO phase detector polarity, it should cause the VCO to jump frequencies. But some parts use the OSCin pin to clock in programming bits, so if it doesn't work, it doesn't fully convince me that it still cant be related to the OSCin pin.
2. Set the charge pump to "Disabled" in the TICSPro GUI. This disables the charge pump. Then apply a voltage of say 1 V and see if you can move the VCO voltage this way.

I think that the "write error" is a red herring. This error has nothing to do with the actual device, but rather something with TICSPro. TICSPro is basing this error in any way based on how the device responds.

REgards,
Dean

Dean,

I ran your two tests and the first one did not accomplish anything.  The second one, I am able to change the VCO CW spike when I apply different voltages to VTune.  What does this indicate?

Thanks,

Darren

I also just checked after C9, so at the 2492 OSCin pin, and I very clearly see a 100 MHz signal. So when you say the OSCin pin might be blown, do you mean INSIDE the 2492?

Dean

Power Supplies:
A) The power pin of the 3.3 SMA pad is not driven. But I think that's because R49 is unpopulated. The Test Point at VccPLL is at 3.3V
B) This is at 4.92V
C) This is at 4.92V
D) This is at 3.3V
Just for ducks, I also checked that voltage after R30 and R32 are in fact 3.3V, and they are.


I don't have a 50 Ohm load on OSCinN because the pad isn't populated with an SMA. But at this point, I actually took off my Signal Generator and can see the 100 MHz coming from the XO at the C9. I could connect an SMA on the pad and try and drive OSinN, but how would I turn off OSCinP?

I think your assessment of the VCO railing sounds right. And then when I applied 1V and 2V to the VTune, it jumped up by maybe 40s of MHz.

I think all of your summary is correct. So what should I do next?
Thanks,
Darren

Hi Darren,

Did you ever get it locked? Please check below.

When the board is powered up, the current is about 170mA.

After programming, it should lock. Current will go up to 225mA. LED will lit (the quality of the LED is not good, it may not work. so check MUXout to ensure you see 3.3V). Output is 4.8GHz at RFout/2 port. Vtune is about 2.44V.