[SOLVED] question about frequency synthesizers

[ginebra]

Hi. I've been reading and researching about how to design a microwave frequency synthesizer. The books of rohde and manassewitsch have helped me a lot in understanding the theory behind the design.

However, up to now, I am confused on how I would actually test my design because I have decided to create the synthesizer via a single-loop pll. What I mean to ask if it is really the standard procedure to just simulate the design (via SimPLL) to reach the desired specs and then fabricate the whole board of the synthesizer? Or is there a way to test if my pll design works as planned before fabrication of the entire board?

I'm asking this because of my tight budget. I cant really afford to fabricate multiple times considering the prices of the components.

Thanks a lot in advance.

 

[tony_lth]

You can use Analog Devices free software ADISimPLL to simulate the PLL. The important is select OPAmp, it's better to select a single power supply OpAmp (such as ADA4841), to suppress the spurs caused by negtive power supply.
It's better to use LDOs for Freq Synthesizer, OpAmp, and VCO separatedly.
If you select AD synthesizers, you can simulate the whole PLL phase noise and spurs include all things.
And, the PLL ref voltage should have bypass capcitor with 10uF Tan, 100nF etc.
The loop PLL capacitors should use plastic film capacitors for anti temperature varied.
Cheers.

 

[ginebra]

Thanks tony_lth. I just want to ask if i'm in the right path with regards to how I would design the microwave frequency synthesizer. Here are the steps I have thought of to realize my goal.

1. Based on target frequency, choose appropriate VCO
2. Choose other components of the single-loop PLL i.e. divider, phase detector, op-amp
3. Analyze PLL characteristics via simulation from SimPLL
4. Transform single-loop pll design to actual PCB layout
5. Fabricate and test actual synthesizer characteristics.

Am I skipping an important step? Just want to improve my plan as much as possible before I start so that problems can be lessened.

Thanks again in advance.

 

[G0HZU]

You need to consider how easily the microwave VCO can be pulled by varying load so you usually need to buffer the output to prevent spurious responses due to load change. Also, use a VCO that can start up reliably with zero volts on the tuning pin (or add a startup offset)

Also, if it is a microwave PLL design you need to take care in how you feed back the signal from the VCO to the PLL divider or prescaler. Make sure any VCO harmonics are supressed at least 10dB down and make sure the feedback path doesn't provide voltage gain at the harmonic frequencies. The best way is to terminate the line with a suitable broadband load.
Make sure the feedback level is well within the recommended range on the datasheet.

Use a good quality reference as the noise of the reference will multiply up in your loop by 20logN.

Include some ground pads at key points in the RF sections to allow you to tap into the PCB design with test equipment. Also include a test point for the Vt volts and make sure the artwork around the loop filter is rugged and will withstand a few component changes if you want to tinker with the loop filter components.

Make sure the VCO is well grounded with lots of ground via holes and the same goes for any prescalers or PLL chips.

Observe the P*P-P rules if you use a DM prescaler.


Finally, have a look to see if you can obtain an evaluation PCB for your chosen PLL chip and look at how the PCB is laid out. Also, it might have a VCO that is close to your own design requirements so you could learn from playing with it on test equipment.
HTH

 

[Mazz]

The procedure is correct.

A couple of notes:
1. You need to know and properly set the specification of any single block of your design. D'ont forget to check stability, noise, settling time.
2. You need RF pcb design experience to avoid many common mistakes that can degrade dramatically your PLL.

I hope it can help.

Mazz

 

[biff44]

Assuming you have access to the test equipment needed....the best way to go would be to buy a prototype board with the PLL chip you want attached to it, hook it to the VCO and frequency reference you want to use, program it with a laptop, and test it thoroughly. The protoboards typically have room for an op amp loop filter to be installed. If you are new to synthesizers, it is highly likely that you will not get the performance you expeced (phase noise, spurs, settling time, etc), and will have to significantly modify the schematic.

 

[tony_lth]

1. maybe you need to make freq plan, such as what is your freq point, what is freq spacing, that would affect the PDF(phase detect freq), and that would help you to select freq synthesizer.
2. select TCXO specifications, if your system need 1ppm freq stability, then your TCXO also need. The TCXO normally should be amplified by transistor.
3. There are two ways to calculate the phase noise, one is from TCXO phase noise, the other is from the phase noise floor of the synthesizers, both way should be calculated. If you use ADISimPLL, it would calculate all the results, you need to analysis the report.

 

[ginebra]

Thank you all guys. I am really a newbie with regards to this matter so I really appreciate all the info you've given.

Actually it's really the translation from the block diagram to the actual schematic that is very problematic for me. I dont have any RF PCB experience to lean on such that I still doubt whether I can actually build a synthesizer.

I would just ask how long it took for you to actually build a synthesizer? Also is knob-based control of the output frequency really easier than controlling it via digital interface?

Thanks again for your replies.

 

[G0HZU]

It depends what the spec is for the synthesiser.

I've designed loads of synths over the years. The most recent was the 'easiest' because I used a ADF4350 all in one PLL/VCO/DIVIDER chip.

If you use one of these then you just need to be able to learn how to program it via the SPI interface. This chip will generate a synth output over 137MHz to over 4GHz.

It's also very small. I used it as part of a tuned detector system. At the other end of the scale I've designed synths using discrete devices and this takes a lot more experience with repect to PCB layout and also circuit design to avoid pitfalls.

Will the ADF4350 do what you require? If so you could buy an evaluation PCB.

The only downside is that the chip comes in a fine pitch SMD package that will be difficult to fit to a PCB by hand.

I'd post a link to the ADF4350 datasheet but the forum has blocked me from posting any links yet

just google

adf4350 pdf

 

[ginebra]

Thank you all for your responses. I just have a last question to ask. Do synthesizers in the microwave range really require Rogers boards for good performance? From what I've read there's a trade-off between using FR4 and Rogers boards which include no. of layers for the board, complexity of design and cost.

Its just that budget is very tight this days such that this issue has come up. Hope some can share their experience using these type of boards in synthesizer design (esp. Rogers Boards). Thanks in advance.

 

[tony_lth]

Maybe it's up to your requirement. What's your freq? Rogers is better at higher freq.
I used FR4 up to 4GHz pLL VCO, and VSWR maybe worse because FR4 Er changed more than Rogers Er at Temperature cofficients.
But Fr4 have stronger copper layer adhere strength than rogers.

 

[ginebra]

Thanks tony_lth. I'm designing my synthesizer for the 7-9 GHz range. I just dont have any experience with acquiring Rogers Boards and I wonder how long the procurement process would be in additional to my budget concerns.

 

[biff44]

there is no way you can use an fr4 board to 9 ghz. 4 Ghz is about the tops, and it is pretty lossy there!

 

[G0HZU]

You are certainly starting at the deep end with a 9GHz design as your first PLL! The highest I have done is a PLL that tuned 4-8GHz. Once you get to frequencies this high you also need to know how to lay out the RF signal path to maintain a flat and reliable response after the PLL or you can throw away lots of signal level. Are you feeding this PLL to another circuit or is it just output to a connector? (eg SMA)

You can get a PCB made in a few days or a few weeks depending on how much you are willing to pay. However, it's best to check the PCB manuf already has your chosen type of Rogers in stock otherwise it can take several weeks longer.

I usually prototype synth PCBs 'in house' by getting them milled on a PCB router eg LPKF or T-Tech. However, at 9GHz you might be better going straight to a PCB manuf as you will be using devices that require very good grounding (lots of tiny via holes etc)

The best advice I can give is to get your PCB design (and your schematic and PLL structure) peer reviewed by LOTS of RF people before submitting it. That way you are most likely to have a successful design first time.

 

[ginebra]

Thanks. The synthesizer I'm making is just output to a connector. I guess that's the next agenda. Improve on RF design skills to have a very good design.

Thanks again for all your replies.