Hi, I've just started getting into electronics and I stumbled across this forum and was very impressed by the knowledge and enthusiasm of the community. I'm excited to learn from you guys!

I was reading a design application on generating sine waves: http://www.national.com/an/AN/AN-263.pdf

On page 8, there's a circuit describing how to approximate a sine using logarithmic shaping. I think I get most of the circuit, but the part I don't understand is the diode section. I can sort of see how it functions as a clamper, but I have no idea why it is "temperature compensated", as they say in the paragraph above.

I'm also confused as to why there appears to be an extra diode in the clamping section (circled). At best, it seems like it will make the clamping section asymmetrical, and at worst it seems to me that when the lm311 swings high, lots of current will flow through that diode to ground. (going right through another diode, then left through the zener, then down through another diode to ground).

Thanks for your help!

Kyle

 

The LM329 is temp compensated.

The current would be limited by the LM311 ouput stage and the 1.2K resistor

 

I'm also confused as to why there appears to be an extra diode in the clamping section (circled). At best, it seems like it will make the clamping section asymmetrical, and at worst it seems to me that when the lm311 swings high, lots of current will flow through that diode to ground. (going right through another diode, then left through the zener, then down through another diode to ground).

The diode served a different purpose here. It is not an extra but essential instead to provide separate paths for the positive and negative feed into voltage clamp.

It has to be connected like that because the LM311 has open-collector output so it cannot source current but can only sink current. Without the diode the path would be +15V into 2x1K2 in series for positive feed whereas the path is -15V via a single 1K2 for the negative feed to the voltage clamp.

I have highlighted the two different paths in the diagram so you can better visualise the process.

 

It has to be connected like that because the LM311 has open-collector output so it cannot source current but can only sink current. Without the diode the path would be +15V into 2x1K2 in series for positive feed whereas the path is -15V via a single 1K2 for the negative feed to the voltage clamp.

Oh! That makes so much sense! I didn't realize that the LM311 was configured in that way - I was assuming that it would act like an opamp and be able to source current as well. Things are so much clearer now!

So let me see if I can put in my own words why the diode is needed, to make sure that I understand. When then transistor in the LM311 turns on, the current through the LM329 will be:

[0 - 0.7 - 6.9 - 0.7 - (-15)] / 1k2 = 5.6mA

When the transistor turns off, the current will be:

[15 - 0.7 - 0.7 - 6.9 - 0.7 - 0] / 1k2 = 5mA

If the diode wasn't there, the current when the transistor turns on would be the same, but the current when the transistor turned off would be:

[15 - 0.7 - 6.9 - 0.7 - 0] / 2k4 = 2.8mA

So I take it that the different between the 5.6mA and 5mA isn't enough to cause a significant change in the clamping voltage, but the 2.8mA would. Am I understanding this correctly?

 

Oh, also one more question - what is the function of the 5pF cap? My only guess was to filter out any thermal noise from the zener diode. How did they decide on the value of 5pF? (Caps like this always mystify me)

 

So I take it that the different between the 5.6mA and 5mA isn't enough to cause a significant change in the clamping voltage, but the 2.8mA would. Am I understanding this correctly?

You calculation is correct. Are you sure sure you are just start getting into electronics?

I think the idea is to make the zener reference current as stable as possible because zener voltage varies with current. One can even trim one of the resistors or place another diode in series with 1K2 in order to "balance" the two current paths.

About the 5pF capacitor I have little understanding of how its value is calculated. All I know is the higher value one uses, the slower the output would switch states, i.e. the slew rate of the output voltage changes be slower. Perhaps its there to prevent "ringing" on the comparator output.

 

I think the idea is to make the zener reference current as stable as possible because zener voltage varies with current. One can even trim one of the resistors or place another diode in series with 1K2 in order to "balance" the two current paths.

That makes sense - or perhaps it is less to keep the zener reference current stable (because according to the datasheet it looked really stable for currents above 1mA) and more to keep the currents stable for the forward biased diodes.

About the 5pF capacitor I have little understanding of how its value is calculated. All I know is the higher value one uses, the slower the output would switch states, i.e. the slew rate of the output voltage changes be slower. Perhaps its there to prevent "ringing" on the comparator output.

Preventing "ringing" on the comparator output makes a lot more sense to me than my thermal noise idea. Good thinking.

Thanks for the answers everyone!