BUF602: Question about simulation results.

Hi Rodrigo,

seems that you short circuit the output of BUF602 with the too low impedance of attenuator?

Kai

I am surprised the simulation is doing that well, you should carefully look at the output VI graph, back then we were doing 4 quadrant curves, but you just want the 50ohm load line, At zero volt out, it does spec 350mA peak, but that is not what you are doing. 

Hi. Please, see the waveform below with a 100ohm load, without the attenuator.

It is not symmetrical.

Rodrigo

Hi Rodrigo,

Looks like you are still hitting the voltage swing limit.

Best,
Sam

I have the same behaviour when I change the amplitude from 5 to 2Vpp.

The output requires a 1.1V headroom with 100ohm load, so on 6V, max of 4.9V swing. Here is +/-6V supply  +/-5V, yes it will round off at the limits, 

You can run this to say 6.4V, 

Here that is, probably just seeing loss due to output impedance of 

If I look inside the TINA model I am using, very confusing, this is the same as the TINA library model, but I got this on the product folder. 

If we test the output impedance in the model, we get this, At DC, this is 4.68dB or 1.7ohm, at 10MHz, it is 4.81dB or 1.74ohm. driving +/-5V input into 100ohm should be showing 5*100/101.74 =+/-4.91V - we get a little less, 

The no load gain is a bit less than 1, here at 10MHz it is -0.034dB, or 0.996. Including that the output should go from +/-4.91 to +/-4.89. Getting closer

Thanks for the analysis Michael.

Do you suggest any other TI part for this scenario?

Rodrigo

can you use a dual? +/-5V through what max frequencie? Gain of 1 flat as far as possible and exact? +/-6.5 or higher supplies available? 

I can deal with any supply value up to +/-12V.

The freq. operation is from 1 to 200MHz and the max. signal amplitude is +/-5V.

Working with unit flat gain is preferable.

Rodrigo

That is quite demanding, 

1. 200MHz 10Vpp is asking for 6280V/usec

Probably the only part to look at is the THS3491, 

You may have to run inverting to tune the response in flatter, but here is a 1st try. using +/-10V supplies, this THS3491 TINA sim was not finding the right DC point, perhaps a TI person could fix this for you to go on, 

THS3491RGT mode gain of -1l.TSC

Michael, the THS3491 input impedance is 15ohms, right? I would like to "isolate" the input stage witn a high impedance device, like a buffer. 

Regards,

That is an open loop number, I thought you wanted to terminate in 50ohm input, inverting mode on a CFA op amp can do that with an extra shunt to ground. 

You could also run non-inverting but I was suspecting flat response at this low a gain would be tough, inverting has slightly higher slew rate and takes the input buffer out of the path (a gain error term). 

Ok, yes, you are right.

I've added the T-Pad attenuator to the design that you sent me and the output signal is very good (VF1). By the calculus the output should be 1.7Vpp, but I am reading 2.1Vpp (VF2). I need to investigate why.

I could not get the sim to work, can you attach the file so I can tune it in, 

Yes, the file is attached.

Tks,

THS3491RGT mode gain of -1l_Att.TSC

So I just kind of guessed that 750ohm would be about right for this noise gain, pretty darn flat it seems, 

That 750ohm input R is part of your input impedance, to match to 50ohm you need to adjust the R to ground to get that, exact would be 53.57ohm so 53.6ohm E96 standard, 

Here is a +/-5V at 200Mhz, you do need to let a few cycles run along to get the initial dV/dT to settle, but this looks pretty perfect 

Thank you to tune the input impedance.

But by my calculus I have to attenuate -14.92db (-15.4dB - (2x-0.24dB from the switchs)) to get the 1.7Vpp. I don't know the T-Pad is not achieving the value.

Hello Rodrigo,

   Were you able to follow Michael's solution to achieve the attenuation? Or is the board differing from the simulations?

   What values are you looking to achieve for each of your attenuation stages? 

Thank you,

Sima