LM319-N: LM319 output capacitance

Hi Niels,
The output capacitance is not specified in the datasheet and for an older part like this, there would not be any internal data available to supply either.

Can you monitor the LM319 output directly with the rest of the connections removed (to investigate if your issue is related to the device or what it is driving)? If you can do that, I'd recommend you try it.

According to Figure 4 of the datasheet, as long as your input overdrive is >2mV, you should get the output response in ~100ns. What pull-up resistor and supply voltages are you using? What is the input CM level? What is your input overdrive? Can you supply the schematic?

Regards,
Hooman

Hi Hooman,

Thanks for your reply. I've not yet measured the LM319 disconnected from the rest of the circuit. The only thing on the output of the comparator is an FPGA, which does make a really small difference to the RC time if I disconnect the FPGA from the rest of the board. So there are I think two sources for parasitic capacitance: either it's the board itself (maybe poor routing) or the LM319. 

The pull-up resistors that we use are 4k7, which is high compared to the 500 ohm I saw in the datasheet, but I think this was a design choice made to reduce current consumption on the board.

The circuit is as follows:

I used a signal generator (square wave, with min 0V and max 5V) on pin 9. Pin 5 is connected to the wiper of a potentiometer, which is connected at one end to a reference voltage of 5V and at the other end to ground. There is also a capacitor of 100nF between pin 5 and ground. The output is connected via a 4k7 resistor to a 3V3 supply voltage. 

Regards,

Niels

Hi Niels,

Sounds like you are running the LM319 with a total (V+ - V-) supply voltage of 3.3V, right? There is no spec for propagation delay / response time when you use this device below 5V total supply voltage. So, I'm not going to be able to tell you whether the 500ns you are measuring is to be expected or not?

Please let me know the supply voltages you are using and other requirements (e.g. open collector output, offset voltage, propagation delay, power consumption, number of channels, package type, temperature range, etc.), and I might be able to point you to a device with specified response time at I'm assuming 3.3V supply voltage.

BTW, a scope waveform would be very useful, if you can provide one? I like to know if the response time you are measuring is the inherent delay within the part (like Figure 4, and 5 copied below), or is it the rise / fall time of the output waveform. The choice of pull-up resistor only affects output rise time and would not influence the part's response time / delay. Have you tried lowering the 4.7k output pull-up to see if you can improve the response time?

The pin numbers you've noted (pin 9, and pin5) are for two different channels / devices within the package.  I assume instead of pin 5, you had meant to say pin 10. Please double check.

Regards,

Hooman

Hi Hooman,

The LM319 is connected to a 5V supply, because input signals can be anywhere in between 3V and 5V. However, the pull-up of 4k7 at the output is connected to a 3V3 supply, because the output of the comparator is connected to an FPGA with 3V3 signal levels.

For the current design we have to stick with the LM319, because the boards have been made already. You can see an example of the problem in the picture below. The input is driven by a 3V3 pulse, while the other input (reference voltage) is at about 2/3*3V3. You can see the propagation time mentioned in the datasheet (which is not a real issue) plus an additional RC-curve, which causes problems.

If I solder another 560R on top of the 4k7, I get the following result:

So lowering of the resistor will be a good option for the current board, but I have to investigate if that disturbs other blocks on the board. There are 20 of these comparators on the board, so current consumption might become an issue if I lower all of the resistors by a factor of 10.

For the next generation of the set-up, we are going to implement even more comparators so any suggestions for a better device are appreciated. What we need is a comparator, with adjustable tripping level, input voltage 0...5V5 and output voltage of 3V3. It would be nice if there is a faster device available. Totem pole output would be nice, but I think we would run into a problem because the input voltage can be as high as 5V5, while the output must be 3V3.

Regards,

Niels

Hi Niels,

So, your concern is nearly all related to the load capacitance working with your pull-up resistor (4.k or 560ohm).

From your scope photos estimated RC time constants (~400ns for 4.7k pull-up), I'd say that you have 80pF total load that your pull-up resistor sees.

Looks like response time is not your issue and LM319-N will be fine to use at 5V total supply. So, no need to change your comparator device for your current design.

How about your output trace on the PCB? It could be that you're seeing a lot of load capacitance because of trace capacitance. You could verify this by leaving the trace tied to LM319-N, but remove the subsequent stage from board. You could remove the ground plane under your trace and make sure it's as short as possible, if there is one next to it?

For your future needs you've listed, I'd recommend the LMV7235 which has a faster propagation delay, RRI capability, open drain output to interface with 3.3V logic. Please take a look.

Regards,

Hooman

Hi Hooman,

Once again thanks for your reply.

I've been busy working on other projects, so that's why I didn't reply to your latest post for some time. I followed your advice and removed the subsequent stage and measured the time again. This hardly made any change, so the capacitance I am facing must be either the board capacitance or the output capacitance of the LM319.

I am now starting the design for the new version of our board. I ordered some of these LMV7235 and I will consider these for the new design. This made me wonder if you also have an alternative for the LM139 we also use in our current design? We use it for the same purpose as we use the LM319 for, only difference being that we can apply input signals up to 30V to the LM319.