jfet buffering for oscilloscope

[amin5659]

hi every one.
I wanna to buffer a signal from a high impedance source and shift the level of that.
the frequency is about 0 to 50 mhz
the circuit i found is like this:

in soft ware it is ok but in practice 3db point of this circuit is about 7 mhz :|
since im new in analog design any correction for this circuit or any new circuit will appreciated
important factor's are:
flat frequency response
unity gain
ability to shift the signal's level

thanks

 

[crutschow]

I assume you mean MHz, not mhz.

That circuit should go much higher than 7MHz.
How was the circuit built "in practice"?

Such a high frequency circuit must be built on a board with a ground plane and very short connections between all nodes.
Also required is decoupling of all power directly at the component power leads using 0.1µF ceramic caps with as short a lead as possible (surface mount to the ground plane is preferred).

Anything less will likely lead to poor circuit performance.

 

[barry]

I'm not sure where the input to your circuit is. It looks like what should be the input is grounded. I would look at your common mode input range; according to the data sheet it's +/-1.9 with a +/-5V supply; you're using -3.3. Also, you're layout could have a big effect.

 

[dick_freebird]

You can buy FET probes (not cheap, at retail, but idiots
sell electronic surplus on eBay cheap if you're patient).
Some are good to pretty high RF.

You might have to accept a nonlinear transfer function
and a serious input voltage range constraint (so, cal-map
it yourself, and feed the 'scope trace to Excel for post-
correction, for the nonlinearity bit; hope your signal is
within sensible range for FET operation).

Think about the signal range. Think about using the FET
as a simple follower with Zout << regular 'scope probe
Zin, supplies set by something that senses the high and
low and adds headroom at both ends, and then use as
little circuitry as possible in the signal path proper and
take off the buffered output by a regular 'scope probe
which can tolerate wide input range?

 

[FvM]

The discrete FET buffer doesn't make much sense these days, you can achieve better performance with JFET or CMOS OPs.

 

[amin5659]

Yes but in my frequency range I couldnt find a good op amp which be appropriate for me

 

[FvM]

I've used e.g. 500 MHz FET OP OPA656 for similar applications.

 

[crutschow]

It would appear that the OPA656 op amp as FvM suggest, used as a non-inverting follower, will do what you want without the need for the added JFETs.
That makes a very simple and small buffer circuit that could be put in a small tube to make a probe.
It would also need a short ground clip at the probe tip to minimize high frequency signal ringing.
The layout rules I suggested still needs to be followed for good performance at high frequencies.
A ±5V supply will give a ±3.5V nominal peak signal range. You could go to a ±6V (±6.5V max.) supply for a ±4.5 volt range.

 

[amin5659]

It would appear that the OPA656 op amp as FvM suggest, used as a non-inverting follower,

The discrete FET buffer doesn't make much sense these days, you can achieve better performance with JFET or CMOS OPs.

thanks. but what about DC shift?
it seems if i want to shift the level, i have to use inverting follower.am i right?

 

[crutschow]

What do you want to shift the DC level?
If you want to block any DC bias on the signal, then just capacitively couple the probe input.

 

[amin5659]

What do you want to shift the DC level?

because adc's D.R is about 0 to 2V.

 

[KlausST]

Hi,,

What ADC? The scope´s ADC? I assume the scope input should be positive and negative.

Klaus