Inverting amplifier / positive pulse problem

O

O. P. E.

Guest
I built a pulse generator that produces ~1.5 ns pulses that go from
0-3V. (Use Linear app note 98 if you want to build one.) I would
like to put it into an RF amp to produce the 7V pulses needed for my
application. My problem is that my amp is an inverting amp. Thus,
when I put in my positive pulse (actually attenuated down for the
amp), it produces a negative pulse of ~4.5 V (corresponding to the
bottom rail).

I thought that I could switch the polarity of the SMA connector on my
board -- attaching ground to the middle pin and the pulse output to
the case. However, I don't think this will work as the ground from
the amp and scope will then have to float. I thought of using a NOT
gate, but wouldn't that just give a pulse that goes from -5 to 0?
Also, the gate would likely increase the pulse width, which we would
like to keep as short as possible.

What would be ideal is to have some simple device that takes a
positive voltage and flips it to negative, while keeping the pulse
width short. At low frequencies, an op-amp could do it, but I'm not
sure of any op-amp that will do it at these frequencies. Does anyone
know something that will do it? Has anyone done something like this
before?

Obviously, it would be better to just use a non-inverting amp. The
spec sheet on the amp (JDSU H301) doesn't even mention that it is
inverting. Are all Optical Modulator Driver amps inverting, so that
it is just an understood?

Thanks
 
O. P. E. wrote:
I built a pulse generator that produces ~1.5 ns pulses that go from
0-3V. (Use Linear app note 98 if you want to build one.) I would
like to put it into an RF amp to produce the 7V pulses needed for my
application. My problem is that my amp is an inverting amp. Thus,
when I put in my positive pulse (actually attenuated down for the
amp), it produces a negative pulse of ~4.5 V (corresponding to the
bottom rail).

I thought that I could switch the polarity of the SMA connector on my
board -- attaching ground to the middle pin and the pulse output to
the case. However, I don't think this will work as the ground from
the amp and scope will then have to float. I thought of using a NOT
gate, but wouldn't that just give a pulse that goes from -5 to 0?
Also, the gate would likely increase the pulse width, which we would
like to keep as short as possible.

What would be ideal is to have some simple device that takes a
positive voltage and flips it to negative, while keeping the pulse
width short. At low frequencies, an op-amp could do it, but I'm not
sure of any op-amp that will do it at these frequencies. Does anyone
know something that will do it? Has anyone done something like this
before?

Obviously, it would be better to just use a non-inverting amp. The
spec sheet on the amp (JDSU H301) doesn't even mention that it is
inverting. Are all Optical Modulator Driver amps inverting, so that
it is just an understood?

Thanks
Click to expand...
Most packaged rf amplifiers are not characterized as inverting or not
inveting, just as an amplifer. ie gain input vswr output vswr and 1 dB
compression point, plus "bandwidth".. You can make a modest gain 10-20
db amplifier with a single active decie, some input match, bias and
output match. with a single devcie if you wiggle the base (gate) and
take the signal off the collector (drain) you have an inverter. you
coul use two in series!

For a 1-2 nS pulse you could use an avalanche tx and hen atenuate the
signal down to 7 volts, as the avalacnhe tx will produce about 100 Volt
1 ns rise 1.5 ns fall the signal has a nice impulse shape..

Marc
 
On May 3, 11:41 am, LVMarc <LVM...@att.net> wrote:

Most packaged rf amplifiers are not characterized as inverting or not
inveting, just as an amplifer. ie gain input vswr output vswr and 1 dB
compression point, plus "bandwidth".. You can make a modest gain 10-20
db amplifier with a single active decie, some input match, bias and
output match. with a single devcie if you wiggle the base (gate) and
take the signal off the collector (drain) you have an inverter. you
coul use two in series!

For a 1-2 nS pulse you could use an avalanche tx and hen atenuate the
signal down to 7 volts, as the avalacnhe tx will produce about 100 Volt
1 ns rise 1.5 ns fall the signal has a nice impulse shape..

Marc
Click to expand...
Perhaps I should have said "GHz amp" instead of "RF amp." For most
radio signals I doubt polarity matters. However, since this is an amp
to interface a signal source with something else, the polarity does
matter. Doing some more research, it looks like the amp expects a
signal from -.7V to -2V. So, I think it may be expecting some kind of
ECL signal. I hope that I can find some kind of translator circuit.

I looked at a few circuits with avalanche transistors. However, with
the delay line/AND gate circuit you can change the pulse width by
turning a pot, and set the rep rate by using the square wave of an
external function generator. The design I found with avalanche
circuits with variable pulse widths required a BNC cable cut to
length, and had much lower rep rates to allow capacitors to recharge.
Also, I worry about the noise 100V pulses create, as I am getting
enough with TTL.

I'll have to look into your active device suggestion. That sounds like
a circuit I looked at that was posted on Microwaves101. However, that
circuit was to produce pulses of RF, while I am just looking to
produce square/guassian pulses.

Thanks for your response.
 
On May 3, 4:54 pm, "O. P. E." <mle...@gmail.com> wrote:
Perhaps I should have said "GHz amp" instead of "RF amp." For most
radio signals I doubt polarity matters. However, since this is an amp
to interface a signal source with something else, the polarity does
matter. Doing some more research, it looks like the amp expects a
signal from -.7V to -2V. So, I think it may be expecting some kind of
ECL signal. I hope that I can find some kind of translator circuit.

I looked at a few circuits with avalanche transistors. However, with
the delay line/AND gate circuit you can change the pulse width by
turning a pot, and set the rep rate by using the square wave of an
external function generator. The design I found with avalanche
circuits with variable pulse widths required a BNC cable cut to
length, and had much lower rep rates to allow capacitors to recharge.
Also, I worry about the noise 100V pulses create, as I am getting
enough with TTL.

I'll have to look into your active device suggestion. That sounds like
a circuit I looked at that was posted on Microwaves101. However, that
circuit was to produce pulses of RF, while I am just looking to
produce square/guassian pulses.

Thanks for your response.
Click to expand...
--------------------

I found a nifty circuit for doing this while looking for something
else. HP did it by using a simple transformer, then, of course,
discontinued the product. Here is a link, and the data sheet shows
how they hook it up.

http://www.home.agilent.com/USeng/nav/-536902328.536881961/pd.html?pageMode=OV

I think HP has discontinued more useful equipment than 90% of
companies have ever made.

Hopefully, this will help anyone in the future who follows this
thread, and is looking for a solution. I'm going to look for some
transformers to hook it up. As I've never worked with RF
transformers, does anyone have suggestions?

Thanks
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top