AD9516-0 with passive loop filter and external VCO

Hi Giuliano,

Sorry for the delay because I have to get registration in this website to make any responses. AD9516-0's 5V mode charge-pump can support external VCO's tuning up to 4.5V (0.5V at low side), but Fig 12 on Page 19 of datasheets does not include full temperature effect, therefore (since the charge-pump switches and current sources were all made by CMOS transistors), I recommend you back up 300mV for each rail. For example, 0.5V at low side should add 300mV to it so the low side for full temperature range is 0.8V, and same for the high side (4.2V in the case). Please let me know if you have further questions. 

Hi yi2016wang,

thanks for the answer.

The PLL has been programmed to work at 860MHz and in this condition the VCO works at about 3V of tuning voltage. Therefore 4.2V should be ok for the control with a passive loop filter.

Unfortunately it loses lock at low temperature (in the range 0-10 °C).

Today I have changed the VCO with another one that can operate at 860MHz with a tuning voltage as low as 2.6V. In this condition the module can work at 0°C, namely the PLL at 860MHz locks once turned on.

According to these measurements, it seems that there is an issue about the ability of the AD9516 to drive VCOs with tuning voltage higher than 3 V about.

Let me know your feedback.

Thanks

Giuliano

Hi Giuliano,

I don’t think it is AD9516 charge-pump issue because it is clear that AD9516 is specified for 0.8V~4.2V tuning operation over temperature range. I suggest you select the VCO with less temperature variation. If you have a VCO datasheets, I can check it for you (but you need to include temperature data in the VCO datasheets). The low frequency tuning voltage shifting is caused by VCO’s temperature coefficient, not the charge-pump on AD9516.

Hi yi2016wang,

If the VCO worked with a tuning voltage higher than 4.2V at low temperature I should record a tuning voltage at the top of the AD9516 spec (0.8V~4.2V), therefore 4.2V. Instead, at low temperature, the tuning voltage measured with the multimeter does not exceed 3V. Furthermore, always at low temp, if I reprogram the AD9516 at a lower frequency, e.g. 850MHz, the PLL locks and the tuning voltage is about 2.7V. This latest test seems exclude also loop filter.

Hi Giuliano,

Please confirm what I read from your email:

You said “at low temperature, tuning voltage measured is <3V”, correct?

If that’s true, then something else here is not right, because the charge pump should work at 3V at low temperature.

Can you verify supply on all supply pins at low temperature? Can you measure the tuning voltage when AD9516 is unlocked at low temperature to see what is?

Yi

Hi yi2016wang,

The AD9516 looks correctly supplied.

I have modified the loop filter setting C1 (capacitance to gnd) at 470pF, and the series C2-R
(to gnd) at 33nF – 1K. I have set the 10h register to 2C: in this way, cp
current is at 1.8mA. According to this setup, the loop bandwidth is about
30KHz.

At ambient temperature, the PLL locks correctly at 860MHz (output
215MHz= 860MHz / 4) (Vtune = 3V about); it also locks to 870MHz (output
217.5MHz= 870MHz / 4) (Vtune = 4V about). Decreasing temperature downto 0°C the
PLL loses lock and the Charge Pump output cannot exceed 2.9-3V; actually at
860MHz setup the PLL is in a condition near to lock; at 870MHz setup the output
frequency does not exceed 862-863MHz.

I have set the 10h register to 0C, 4C, 7C: in this way, cp current is at
0.6mA to 4.8mA. According to this setup, the loop bandwidth change in the range
15KHz to 70KHz about with a degraded phase margin. For all the condition I
observe a correct behavior at ambient temperature but unlock at low
temperature.

I have set the 10h register to 24h: in this way, cp current is set to
1.8mA and in Pump Up mode. In this condition at all temperature the cp output
is set to +5V (the same of Vcp) and the VCO output is at 880MHz about.
Therefore in Pump Up mode the AD9516 works correctly and it is correctly
powered; in Operative mode (10h register at 2C) the AD9516 does not seem
working well  decreasing temperature.

As I already wrote, what I see is like the cp clips or saturate to a
voltage that is about 4.2V at ambient temperature and decreases reducing
temperature, down to 3V at 0°C. This decreasing is gradual in temperature.

I wonder if it is possible in Analog laboratory on a AD9516 with
external VCO demo board to repeat a temperature test in order to verify if the
cp can be critical at low temperature.

In any case I am solving the problem changing the VCO with a 850-900MHz
one, that operates at 860MHz with a Vtune in the range 1.5-2V.

Let me know your comments and update.

Thanks in advance

Hi Giuliano,

Based on what you said, the charge pump can go up but not go down, correct? For example, in your case, it can’t go below 3V, correct?

It would be easy to check if you can adjust the reference input up and down. For example, if you lock the PLL at 860MHz with Vtune=3V, then just adjust the reference input down little to see if Vtune would move down below 3V (assume VCO is positively tunable).

Possible issues could be:

1.       VCO output is railed

2.       This particular AD9516’s charge pump has problem

Hi yi2016wang,

No problems for the cp to go down...... I can set up for example the PLL to operate at 850MHz and in this case the Vtune is about 2V. It seems the cp has problem to go up 3V at low temperature, when it works in operative mode, even if Vcp=5V. In Pump Up mode it can reach 5V.

I have verified this behavior on 2 boards.

I am trying to solve the problem changing the VCO with a 850-900MHz
one, that operates at 860MHz with a Vtune in the range 1.5-2V.

Regards

Giuliano

Hi Giuliano,

So it is the problem if Vtune moves high (above 3V). Have you tried to adjust reference input to move Vtune up?

AD9516 is 100% tested in production for its charge-pump for 3.3V and 5V supply. It must be something wrong that we don’t catch. Would you let me know the VCO you used, Mini-Circuit or other ventors, model/part number? I can have someone to check that if I know the exact VCO.

Have you ever checked VCO’s output level going into the AD9516’s CLK input pins? Do you use differential input of VCO, or single-ended? How’s match done? The CLK input receiver is a differential receiver with self-bias internally. For a VCO signal, a properly designed match circuit is necessary.

Hi,

At low Temp, I have tried to program the PLL to a higher frequency, at 870MHz, where the Vtune should be 4V. After this frequency change the Vcp did not change significantly above 3V.

We use single ended input.