MRF300AN: questions about datasheet to the real world

cancel
Showing results for 
Search instead for 
Did you mean: 

MRF300AN: questions about datasheet to the real world

675 Views
QRN
Contributor I

I would like to build up a couple of RF amps using the MRF300AN, this looks like a really great device.  I have a whole series of questions about going from the suggested applications in the datasheet, to building actual working circuits.  I'm going to start off with a couple questions about the PCB's specified:

The PCB's specified in the datasheet give a part number, and the manufacturer as MTL.  I see that MTL is an outfit providing PCB fabricating services.  Am I correct that the boards may be ordered from them, by merely supplying this part number, and they will already know what it is without having to send them files or anything?  If so, this is excellent!  While I'm not new to PCB fabrication, I've never had to make plated-through holes before, and researching this implies that it will be far more trouble than it's worth for only a couple of boards if they can be obtained at reasonable cost from MTL.

My second question involves the difference in PCB used for the datasheet's 230mhz design, this is the only board that differs from the others.  While the first RF amp that I want to build will be on 50mhz (therefore the design seems to be pretty straightforward and turn-key from the datasheet), the second will be on 222mhz.  This frequency range is a transition, where stripline techniques start to become more practical and "normal" LC elements start to become less practical.  Is there a particular reason for using stripline techniques instead of air-wound coils for the 230mhz design?  Or was it done merely to illustrate usage of striplines in the input and output matching?  Likewise, the board material is different.  At this frequency, is there really any significant advantage to using AD255C instead of FR4?

 

I ask this second question because I would prefer to build the 222mhz design with the smaller FR4 board and air-wound coils used in the lower frequency designs.  It seems like there would not be any significant loss of efficiency or added difficulty in construction in using air-wound coils instead of striplines at this frequency.  Likewise, it would seem like any losses in FR4 PCB material would also be insignificant at this frequency (am I wrong about this?).  To me, the advantage of reduced size and cost (especially considering the heatsink and a copper block heat spreader) would outweigh a loss of .001% efficiency in air-would coils or a .001db loss in the FR4 board material. 

 

These questions are first, because I am planning the chassis layout within the cabinet, including the airflow path for the forced-air cooling system, so will need to know which set of dimensions I'm designing around.  My next set of questions will involve component choices, again from the perspective of going from what the datasheet says to a real-world implementation of a working circuit.  TIA for any useful replies!

0 Kudos
2 Replies

644 Views
QRN
Contributor I

Ok, great, thank you for the response!  So then my conclusions are correct about the board itself.  But your response prompts the question, is it possible to order just the bare PCB only, rather than the complete "evaluation board" or "kit"?  I originally checked with both Digikey and Mouser (which are your major distributors in the US, and I have been ordering parts from both of them for decades), and only the complete "evaluation board" is available.  Likewise, I do not see any reference to just the bare PCB in the "RF Circuit Collection".  The only reason I would even order the bare PCB rather than replicate the layout and etch it myself is that it has to lay flat against the copper heat speader, and the necessary method of plated through holes is far more trouble than is worth to do myself for only two boards (I'm also going to build one for 50mhz).  So would NXP or MTL sell me a couple of the bare PCB's?  How would I go about ordering them?

 

My next comment and question involves some of the component choices.  That Bourns trimpot specified in the datasheet is not a real world component that is readily available.  Both Mouser and Digikey sell variations of that same part, but the physical configuration is different, such that it will not fit onto the board correctly.  This appears to be a non-production part that appeared on someone's bench as part of a sample kit from Bourns.    And for the available variations, almost $5 for a trimpot?  Seriously?  That's an awfully expensive trimpot, when parts with the exact same specs are available from the surplus houses for well under $1.  For my implementations of the circuit, I'll use one of the trimpots that I already have laying around on hand, locate the bias voltage divider on the power supply board (fed from an IC regulator instead of a zener), and just bring the bias in from off the board as a separate power lead.  It will require an extra feedthrough capacitor and piece of wire, but I already have those on hand, and it sure beats paying almost $5 for a trimpot!

 

My next question is about some of the capacitors used.  Many of them are rated at 500v.  Is a 500v rating really necessary for some reason?  At this power level, drain voltage, drain impedance, and load impedance, wouldn't the minimum "safe" rating be somewhere closer to 100-160v?  Again, I have many of the values needed on hand, but at voltage ratings of less than 500v.  Also, some of the exact capacitors specified are no longer available (at least not from Mouser and Digikey), so substitutions will become necessary in some cases regardless.

0 Kudos

657 Views
LPP
NXP TechSupport
NXP TechSupport

1.
We did not expect customers to place orders directly to MTL. Instead, one should order the boards through our distributors or using "RF CIRCUIT COLLECTION" service on the NXP site:

https://www.nxp.com/design/development-boards/rf-circuit-collection:RF-CIRCUIT-COLLECTION

The boards can be provided as fully assembled and tested circuits or as a kits including PCB and all the components required. Delivery options can be discussed (for example, the number of device samples).

2.
Your general understanding is correct.

Microstrip design reduces the BOM and simplifies assembly. In the case of microwave design, using microstrip matching is indispensable.

At low frequencies, using microstrip matching is limited by the size. Typical microstrip element size needed to match any impedances is Lambda/4. If using capacitors at the ends, the length of microstrip segment can be shortened to Lambda/12. Shorter segments could only be used for partial tuning.

So, there is a natural low frequency bound for microstrip matching design. It is 300-400MHz for typical board dimensions. At lower frequencies, this technique does not give any advantages.

3.
The board material determines dielectric constant (wavelength->TL dimensions) and losses. The losses must be taken into account and they are critical in some applications.

- Low noise amplifiers. Input losses increase noise factor by the same amount.
- High Q designs (filters,transformers). The losses limit maximum quality factor of the filters.

In some cases, the substrate losses can be beneficial since they improves stability of an amplifier.

A designer must choose between cheap FR4 material and costly ones after evaluating the performance loss caused by the material. Most RF power amplifiers below 1GHz can be made using FR4. Microwave designs generally use high quality RF substrates.

4.
222 Mhz RF power amplifier can be implemented using FR4 material and discrete components for impedance matching.

Have a great day,
Pavel
NXP TIC

 

0 Kudos