Log Amp (AD8319) Output Issue - How to invert?

Hello everyone,

I'm working on a system that accurately records (4) points along the rising edge of the envelope of an RF pulse. It must be able to accomplish this goal over a wide bandwidth. To do this, I decided to use the AD8319 log amp, and use the high-speed LT1721 quad-comparator.

I want to sample a positive RF pulse, but unfortunately the AD8319 has a negative slope, so the input RF pulse is inverted. I corrected this using a AD8058 op-amp in an inverting amplifier configuration with a gain one 1. The problem is the AD8319 produces a DC component with an AC signal on top of the DC (DC-coupled), so this DC component is inverted along with the AC signal, producing a signal that has the right characteristics but is shifted approximately -0.7VDC down.

Is there any easy way to fix this without affecting the performance of the logamp? I want an AC coupled signal fed into the quad comparator, so I can set trigger voltages at, say, 0.2V, 0.4V, 0.6V, and 0.8V.

I tried using a minicircuits coaxial DC block, but it distorted the envelope output. Perhaps it does not have the right frequency characteristics since the max frequency output of the AD8319 is appx. 50 MHz, and that DC block is design for operation up to 18 GHz. Should I just try a normal series capacitor to act as a high-pass filter? If so, what value should I use?

Any ideas are welcome. Thank you in advance!

It's hard to understand, why you want to invert the log output signal instead of adjusting the thresholds and inverting the comparator outputs.

When AC coupling a signal, you have to decide about the intended cut-off frequency first. Also a high pass has a characteristic impedance. A 50 ohm DC block is most likely unsuitable.

Yes, you have to design the high pass according to your needs, and a simple series capacitor (together with the necessary bias resistor) will act as the most simple 1st order variant.

I doubt however, if AC coupling the signal complies with your intention to accurately determine signal levels over a wide bandwidth. To decided about, you have to know the signal characteristic.

FvM, thank you for the reply.

The reason I don't want to invert the comparator outputs is it requires more IC's (4 inverting opamps as opposed to 1). I'm also aiming to measure the timing difference between two rising pulses with sub-nanosecond accuracy using time-to-digital converters (ACAM GPX module). The differential propagation delay of comparators will add more variability in timing.

I'm a bit rusty in HPF design. Could I just use the laplace transform of circuit components and design in that manner (i.e capacitor = 1/s)? I'd imagine this is a textbook design.

As for the signal bandwidth, the AD8319 outputs a spectrum ranging from 0 to 50 MHz.

EDIT: I need to hook up the output to a spectrum analyzer and see the spectral content of the signal I anticipate and ensure that there aren't significant low-frequency components that could potentially be cut off.

Any suggestions are welcome. I know some of this seems elementary, but I'm mostly a RF electromagnetics student so I don't deal with low-frequency analog design too often .

Added after 25 minutes:

Also, does anyone know if there is a software that can simulate this log amp? I've been trying to find a SPICE model or something that would work in Agilent ADS, but haven't had any luck.

I ended up just adding a negative DC offset to the input of the inverting amplifier. This achieved the desired result. Unfortunately, I only have positive voltage regulators on-hand (i.e. LM137). I wish Radioshack offered the LM337!

The reason I don't want to invert the comparator outputs is it requires more IC's (4 inverting opamps as opposed to 1).

Click to expand...

In my opinion, inverting a comparator output can be simply done by exchanging incerting and noninverting input.

I wish Radioshack offered the LM337!

Click to expand...

It's hard to find an "electronics" shop, that has less parts in stock than Radio Shack. It's just a legacy from their older days. You possibly get more parts from any TV repair shop, not to mention serious DIY and professional electronics shops and catalog distributors.

You are absolutely right about switching the comparator inputs FvM. I feel silly for missing that simple concept.

It turns out, however, that I need a VGA stage between the logamp and the comparators since the peak-to-peak voltage swing (from 0V to Vmax) varies from 200mV to 650 mV based on the input power to the AD8319 log amp. Vout to vary from appx. 0V to 1V.