Precision Rectifier Clarification and Problem

Thread Starter

Erasimus

Joined Mar 29, 2009
5
I'm currently working on a sensor project for a colleague in Mechanical Engineering and part of the circuit design that he gave me from his adviser's graduate thesis contains a precision rectifier like the one shown below.


http://sound.westhost.com/appnotes/an001.htm

I found this circuit on this website as well in another thread.

The circuit should respond at 100 kHz and I'm using the LF356 Op-Amp. Does anyone have any recommendations for high speed diodes? Thanks very much in advance!

-Erik

edit: I am looking at digikey and mouser but I'm not entirely sure which parameters will give me the response I'm looking for. I found the one below... but not sure if it will work.

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=MA27P0100LCT-ND
 

SgtWookie

Joined Jul 17, 2007
22,230
Look at Trr (Time for Reverse Recovery).
Ironically, there is no Trr specified in that datasheet. Perhaps they don't want you to know how fast (or slow) it is?

A standard 1N914 or 1N4148 switching diode has a Trr of around 4nS. Let that be your benchmark.
 

SgtWookie

Joined Jul 17, 2007
22,230
After playing around with it a bit in a simulator, looks like an LF356 may have a bit too much propagation delay for your 100kHz input. An LT1220 works much better.

The opamp selection is currently a much more severe problem than the diode's switching time.

Another item to look for in the diode's specifications is capacitance. Lower = better. The 1N4148/1N914 has 4pF.
 

Thread Starter

Erasimus

Joined Mar 29, 2009
5
Thanks for the help Sgt.

I was definitely thinking about replacing the LF356 due to the propagation delays... It has worked for me in preliminary tests, but I can see a bit of distortion and I think I'm at the breaking point. Thanks for the advice about the diodes as well... at least I now have an idea of what to look for.
 

SgtWookie

Joined Jul 17, 2007
22,230
Just so you can see the difference in the simulations I ran, have a look at the attached.

Notice the difference in the delay, and particularly the difference in distortion near the zero crossing point.

The 1N4148/1N914 diodes behaved essentially identically, nearly like ideal diodes. Just for grins, I tried using 1N4001's; it turned the output into a series of shark's fins. :p

It still could be improved by using even faster opamps, of course. Stray capacitance will be a killer for you; breadboarding isn't in the cards because breadboards will have too much capacitance. I suggest using the "dead bug" prototyping method on a copper clad board (Google is your friend here). Be sure to use bypass capacitors on the opamp's Vcc/Vee pins, but keep other components/leads away from the ground plane.
 

Attachments

Thread Starter

Erasimus

Joined Mar 29, 2009
5
Thanks for the simulations. I understand why to not breadboard based on the tie points within the breadboard... additionally, I already know what dead bugging is for the circuit. I've actually worked up a PCB for the circuit that has been working so far. However, now that you mention stray capacitance I worry that I didn't design to prevent it well enough... Do you or anyone else have any tips or tricks to laying out these high frequency boards? I've attached a PDF printout of my current PCB and I can already see that the footprints for the large capacitors that I had planned on using is definitely going to create some capacitance... (those are the large square green areas for reference :p)

Any other advice is certainly welcome, thanks again.
 

Attachments

Last edited:

SgtWookie

Joined Jul 17, 2007
22,230
Well, capacitance for Vcc/Vee to ground is good. You need a 0.1uF decoupling cap to ground for Vcc/Vdd/Vee on each opamp or other IC to help keep noise under control. The caps should be located as close to the IC's pins as possible.

You have routing problems. You should start off using one layer for running the Y axis, and another for the X axis. If you need to change directions, put a via through to the other layer. Otherwise, you box yourself in a corner and run out of layers very quickly.

For the moment, re-route the board so that everything runs either vertically or horizontally. Routing diagonally or at odd angles can make things tough on you, especially when you're new at it. The yellow traces you have running all over the place.

It also helps limit capacitive coupling if you cross signals on opposite sides of the board at 90° to each other; it minimizes the size of capacitance.

Keeping the board size small will also help to limit parasitic inductance and capacitance.

Check out this site: http://www.smps.us/pcb-design.html
 

Thread Starter

Erasimus

Joined Mar 29, 2009
5
Sorry! To clarify, the yellow routing lines that are still visible are from the ground pour that's on the bottom of the board and are removed once the ground pour is in place... I didn't include it because it makes it difficult to see the other layers... but I seem to have confused things worse. The red routes are on the bottom of the board and the green are the top. I've disabled the ground net so that those yellow traces aren't there anymore... refer to the previous schematic to see where things are netted to ground, and this second schematic for a more accurate representation of what the board will look like.

It's an interesting idea to have vertical traces in one direction and horizontal in another.... I had been operating under the assumption that right angles as well as numerous vias in general were something to avoid due to reflections, is this a false assumption? The op amps (8 pin soic) are coupled with .01uF capacitors right next to the Vee and Vcc mostly through a via directly beneath the pins.

-- It also helps limit capacitive coupling if you cross signals on opposite sides of the board at 90° to each other; it minimizes the size of capacitance.

This is a very good point, I'll try to keep this in mind as I move forward.

One specific question that I have is in regards to the big green pads in the north east quadrant... If there is a ground plane underneath them that seems like a relatively large amount of capacitance despite the relative thickness of the PCB and the somewhat low voltages (DC at that point since they're involved in a low pass filter) that they should be enountering. Is this something I should be worried about and in the future if I do have to have these large footprints... how can I mitigate their effect?

Once again, thank you so much for your responses, I just found this site through google... but I will now be doing all that I can to help people here because I want other people to receive the same help that I did...
 

Attachments

Last edited:
Top