Recommend me a 1GHz buffer that drives 50 ohm load

[Henry-cool]

Dear all guys,

I will use an off-chip 1GHz Buffer.

This buffer should drive 50 ohm load. Input P1dB > 5dBm, gain=1

Do you know the product which satisfy this?

Thanks!

 

[RF-OM]

Re: 1GHz Buffer

If you want to build it as discrete component device probably the least expensive solution will be to use emitter follower. Check Infineon RF transistors like BFP540 which is very inexpensive and you may have excellent buffer. Or, if the cost is not an issue, you may check RFMD, Minicircuits, WJ and other companies for RF transistors and gain blocks. In this case you get ready to go internally matched device.

 

[FvM]

Re: 1GHz Buffer

As you specified a unity gain buffer, I suspect you may intend a high (> 50 ohm)input impedance. In this case the input impedance specification would be a very important parameter.

In any case, with matched output impedance, a buffer would have an effective -6dB gain. Do you intend a -6dB buffer or an unity gain amplifier? An unmatched output buffer is more a theoretical option at 1Ghz bandwidth, it would have an attenuation due to it's output impedance anyway.

 

[RF-OM]

Re: 1GHz Buffer

As I understood the author wants to build narrow band buffer with 1 GHz frequency. If it imust be broadband then amplifier with feedback must be used. In this case BFP540 or BFP 650 can be used and it is easy task for single transistor design with 50 Ohm input and output match. RC feedback from collector to the base and about 10 Ohm emitter degeneration will do the job.

 

[Henry-cool]

Re: 1GHz Buffer

Thank you for reply~!

I check the products which you recommended.

They are good products. But, I need another.

I need high imput impedance(>>50ohm), 50 ohm ouput impedance Buffer.

In 1GHz, most buffer's output impedance is 50 ohm.

But input impedance is also 50ohm.(not high impedance)

I don't know how I do.

 

[FvM]

Re: 1GHz Buffer

So I guessed right. Knowing the said specification would allow to select a suitable part (if any).

It didn't come clear yet, what gain = 1 means for you? To reproduce the input voltage at the load, a v = 2 buffer would be needed. Is an inverting amplifier, as suggested, an option for your application (apart from the required high input impedance)?

P.S.: For applications with similar specifications, a wideband OP as TI THS3201 may be a solution. You should check the application circuit at the datasheet title page, it may already fill your needs. https://www.ti.com/lit/gpn/ths3201

 

[Henry-cool]

Re: 1GHz Buffer

Thank you for reply~!

Inverting amplifier is a good solution if input impedance is high.

But, is it possible to make inverting amplifier in 1GHz?

 

[FvM]

Re: 1GHz Buffer

Yes. Would be probably an inverting amplifier as suggested by RF_OM with a preceeding emitter follower as a buffer.

 

[madengr]

1GHz Buffer

Why don't you just use a voltage divider followed by a gain block? If you have a 20 dB gain block such as the AD8353 you can make a voltage divider with a ~5k input resistance and still get unity gain. Of course there will be some parasitics in the divider but it's simple.

 

[Henry-cool]

Re: 1GHz Buffer

Why don't you just use a voltage divider followed by a gain block? If you have a 20 dB gain block such as the AD8353 you can make a voltage divider with a ~5k input resistance and still get unity gain. Of course there will be some parasitics in the divider but it's simple.

Thank you for reply~!

I check the AD8353. The buffer is a good ability!

But, the buffer has a single input(50 ohm), single output(50 ohm)

I don't understand how I can make a voltage divider with a ~5k input resistance.

If it is possible, can you explain it in detail with picture?

 

[madengr]

1GHz Buffer

Just put a 5k resistor in series with a 50 ohm resistor, and connect the amp between the two resistors. The amp sees a 50 ohm source and your circuit sees a 5k load. The 1/100 dived ratio and 20 dB amp gain give you unity gain over a wide bandwidth. The parasitic, parallel capacitance of the resistor is what will limit the divide ratio. 5k may be high; I have not done a simulation.

 

[RF-OM]

Re: 1GHz Buffer

It is still unclear what bandwidth do you need. How you can expect right answer if you did not provide right information?

May be cascode amplifier CE-CB is the choice? It will have high impedance and high gain. You can use voltage matching instead of conjugated and put resistive voltage divider at the input. It will bring input impedance to the required level and reduce extra gain. Noise will increase but it may be well inside the limits. Such technique may help to get right input and output impedance (when resistive divider also installed at the output too) as well as more stable operation across wide band. Linearity also can be improved. One of the grate advantages is that this configuration does not invert the phase twice and have high one stage stability with two stage gain value.

 

[madengr]

Re: 1GHz Buffer

Sorry I had my (in the head) calculations wrong. It will be a 2 kOhm resistor and you will need 40 dB of gain stage. Here is a simple probe that will provide about +/- 1 dB over 3.5 GHz bandwidth. The simulation uses Modelithics models so the chips resistor parasitics and pads are included; they must be included since they affect your roll-off. The 3 dB pad between the AD8354 provides stability and also helps equalize the gain due to the parasitic capacitance of the divider. The extra cap at the input may be tuned for additional equalization.

If you want a buffer though this is overkill since it uses allot of current. You would be better off trying what the other's have suggested.