Op amp for fairly high speed

[jss87]

First of all, I don't know what I'm doing; so feel free to disparage my
attempts. I'm just trying to lash together some old and new equipment.

I have some pulses that are "triangular" with a flat top; the time from
start to top is about 5 to 20 microseconds and the flat portion about 0 to 5
microseconds. The timing will be constant for a given setup, but I'd like to
be able to handle them all. The input is 0 to 1 V, and I'd like a gain of 8
to 10. The pulses come in random intervals.

I've tried an AD829 which should have plenty of speed. The output is about
as wide as the input, but it has three states depending on the value of the
feedback resistor: flat, variable over a narrow range, but the output does
not seem to depend on the input voltage, or pegged.

I would welcome any suggested schemes for this amplification, or tips based
on the behavior I'm seeing.

Thanks,

John

 

[Jon]

John,
The AD829 should do the job. What is your power supply voltage? What
is your load impedance? With +/- 5V supplies and a 50 Ohm load, the
nominal output voltage swing capability of the AD829 is only +/- 1.4V.
Another problem may be that you don't know how to spell your first name

Regards,
Jon

 

[jj]

With +/- 5V supplies and a 50 Ohm load, the
nominal output voltage swing capability of the AD829 is only +/- 1.4V.
Another problem may be that you don't know how to spell your first name

Regards,
Jon

I have +/- 12 and I've seen the output go to 10 V or so, but that was with
the output going to a scope, not the equipment. I'm not too sure of the
impedance; most of the gear is connected by 50 ohm coax. I may revert to an
inverting config so I can have 50 ohm input impedance.

I'll work on the spellling,

jahn

 

[Tim Williams]

I have some pulses that are "triangular" with a flat top; the time from
start to top is about 5 to 20 microseconds and the flat portion about 0 to
5 microseconds. The timing will be constant for a given setup, but I'd
like to be able to handle them all. The input is 0 to 1 V, and I'd like a
gain of 8 to 10. The pulses come in random intervals.

Ok, so the maximum criteria is:
- 5us shortest rise/fall time (pretty slow as things go, FYI
- 10V (p-p?) output
- 100kHz maximum rate
?

10V in 5us is 2V/us slew rate. Even the lowly 741 does 5V/us, IIRC. (Oh
nevermind- it's a tenth of that! Well, discrete circuits I've built easily
handle this.) Well, the LM833 does >5V/us, anyway.

I would welcome any suggested schemes for this amplification, or tips
based on the behavior I'm seeing.

First of all, what is your signal, and what is your goal with it?

If you just want straightforward *amplification* of a trapezoidal wave (that
is to say, square shaped pulses or whatever, but with sluggish rise and fall
time slope), a reasonable speed op-amp wired for a gain of 10 or so will do
it right there. If you want to *generate* the slopes, accurately, from a
relatively square signal source, you need to look at slew-rate-limited
amplifiers. There was a discussion on this a few months ago IIRC.

Tim

 

[[email protected]]

John,

I have +/- 12 and I've seen the output go to 10 V or so, but that was with
the output going to a scope, not the equipment.

If you have the 50 Ohm termination, you just can't get +/-10V output
swing
out of the 829.
Actually you can't get +/-10V at 2K if your power is +/-12 V, at least
you will not be guaranteed to get it. You can see all this stuff well
documented in the graphs of the 829.
You would have to look into the compensation, choose the non-trivial
one (I do not remember its name now, it is explained in the application
text of the data sheet, basically a much lower cap than otherwise
between compensate output and inverting input).

Dimiter

 

[jss87]

First of all, what is your signal, and what is your goal with it?

If you just want straightforward *amplification* of a trapezoidal wave
(that is to say, square shaped pulses or whatever, but with sluggish rise
and fall time slope), a reasonable speed op-amp wired for a gain of 10 or
so will do it right there.

Yes, just amplification, and per Dimeter, I'm looking at adding a follower
to drive a 50 ohm load. What is bugging me the most is the lack of linear
amplification. It is almost like I have a comparator that puts out one of
two levels depending on the gain (at the rail, or somewhere lower, jumping
suddenly as the feedback value is changed).

Thanks for the advice.

John

 

[jss87]

Dimiter,

Thanks for the help. I'll look at a follower to drive the load.

John