Fast rise time at low frequency .. is it possible

Thread Starter

captoro

Joined Jun 21, 2009
10
Hello,

Need square wave output with fast rise time.
I have the LT1818 amplifier chip, It has a rise time of less then 1ns. I just realize that this chip does not make really good square wave under 10khz.
I need to pulse 8Hz but with 1ns rise time. Is there a chip out there that does that ?
MOst chip I saw so far that handles low frequency has rise times in the milliseconds !
regards
Ken
 

AlbertHall

Joined Jun 4, 2014
8,551
I am not familiar with the LT1818 but I don't see how it could anticipate that the next edge was a long time away and so increase the rise time for the current edge.
 

ronsimpson

Joined Oct 7, 2019
190
Can you post your circuit?
What is the output voltage?
I can't see how you are going to get 1nS edges but 5nS more likely. (maybe with a gain of +1, it is fastest)
I don't see why the edges are slow at or below 10khz.
I have never used a amplifier with mS rise times. Well maybe, I did use a amp to measure fractions of pico amps. It was slow.

I need more information. How are you measuring, what signal source, what scope, what probes. Give me all you can.
 

Ylli

Joined Nov 13, 2015
799
The LT1818 will operate down to DC. Are you using something like this?:
index.png
Try adding a small capacitor (100 pF) across R2. Doubt that you will get 1 ns rise time, data sheet shows 5 ns more likely.

Otherwise, you might feed the oscillator into second LT1818 configured as a Schmitt Trigger.
 

MrChips

Joined Oct 2, 2009
19,425
Rise time and frequency bandwidth are interrelated.
Approximately, 1/frequency = rise time

If you want short rise time you have to seek a component with wide bandwidth.
 

danadak

Joined Mar 10, 2018
3,630
What pk to pk V do you need at the 1 nS Tr, Tf ?

What is load in C and R values ?

VHC or ECL logic can produce very fast Tr, Tf, but again that
depends on what you are loading it with and if you can live
with 5V or less pk-pk values.


Regards, Dana.
 

Thread Starter

captoro

Joined Jun 21, 2009
10
Hi,

I am feeding the output to a ISL55110 mosfet driver. I am not getting a perfect square wave too as you can see.
 

AnalogKid

Joined Aug 1, 2013
8,157
There is nothing in the datasheet, including the internal schematic, that prevents normal operation down to DC.

However ...

The circuit you posted is a square wave oscillator. That type of circuit overdrives the device output stage, which can cause all kinds of interesting things to happen that are not a part of the linear operation specs.

What scope are you using? What is its vertical amplifier bandwidth? Is the scope probe properly compensated? How long is the probe ground lead? This is critical when dealing with such high frequency signal components.

Also, the output waveform will not be a true 50/50 duty cycle.

ak
 

danadak

Joined Mar 10, 2018
3,630
Hi,

I am feeding the output to a ISL55110 mosfet driver. I am not getting a perfect square wave too as you can see.
Why do you feel you need 1 nS Tr and Tf ?

The datasheet, to get max specs, is asking for < 2 nS, even that, for the application, does not seem
necessary. Also you only need 3 - 5 V drive to the 55110. As far as source of square wave seems like
a simple PLL based OSC and then a fast logic gate would suffice to get the Tr, Tf.

http://www.ti.com/lit/ds/symlink/cdclvc1102.pdf Look at this catagory of buffer to get the 1 nS
type performance.

Regards, Dana.
 
Last edited:

ronsimpson

Joined Oct 7, 2019
190
With no load the 55110 might get to 1.5nS. With a MOSFET connected (depends on size) it will have more like 10nS rise or fall times. Because the 55110 has gain I think driving it with 20nS sedges will be ok to get faster edges back out.
I am posting my circuit and the oscilloscope output.
From what I see on the scope you have several possible problems all happening.
*Are your scope probes balanced? Calibrated. Does not look balanced.
*Looks like the 1818 is driven into the rails which makes it sticky or slow to come back off the power supply rails.
*There is a chance the power supplies sage with load but probably not.
*If you are looking for nS edges you can't use a wire for ground. Three inches of wire will ring at that frequency.

I would use a digital counter to get 8hz from a faster counter. Maybe a F or S version of TTL. Or CMOS like the 4060 counter. If you need fast edges you can buffer with a fast buffer before the 1818.
 

Thread Starter

captoro

Joined Jun 21, 2009
10
I am using the same circuit with a LM741 OPAMP, I am getting almost perfect square wave, but the rise time is very slow. Thought when changing the OPAMP, it will increase the rise time, It did but I am getting distorted wave.
 

ronsimpson

Joined Oct 7, 2019
190
scope probe compensation: Have you compensated you probes to that scope?
1572010753259.png
1572010844208.png
When measuring 1nS edges a normal scope ground clip will not work. The length of the ground wire needs to be small. See picture.
1572011023616.png
 

Thread Starter

captoro

Joined Jun 21, 2009
10
How distorted ?? You can look at the picture I uploaded yesterday. I know the LM741 is old, thats why I asked in my first post if anyone know an IC that could accomodate what I need. The LT1818 is only good for frequencies above 10khz. I aiming between 4 to 20Hz.
 

Ylli

Joined Nov 13, 2015
799
The LT1818 is only good for frequencies above 10khz.
Where do you get this info from? Op amps are good to DC. From the LT1818 datasheet:

"The LT1818/LT1819 are single/dual wide bandwidth, high
slew rate, low noise and distortion operational amplifiers
with excellent DC performance."
 

Thread Starter

captoro

Joined Jun 21, 2009
10
I was looking at the figure ' Input Noise Spectral Density ' on page 6 of the datasheet.
Maybe I am not reading the graph correctly.
oscilloscope_.jpg
 
Top