CC2530 Analog Comparator is a Mystery

1) Leakage on the pins is dominated by other things on the pins and is in line with the other pins (ESD structures will most likely dominate leakage).  Data shows on the typical parts that 100 nA was seen at 85C input at mid-rail. 

2) This is a tough one to determine.  Since the output of the comparator only goes to the digital core, it is clocked at the clock speed and this dominates the delay time seen.  Effectively the comparator latency is lost in the clocking of the signal.  Remember though this is a nano-amp comparator so it is not going to be very fast.

Thanks Chris,

1) So, in reality the total current consumption will be around (230nA + 2x100nA)= 430nA. Then you need to add a voltage reference to (In-) pin to really do something useful, which would require a LDO-regulator about ~500nA (TPS78225 ) and a high impedance voltage divider (2.5V/20M-ohm) ~125nA. The total power in sleep mode with comparator working would be about ( 430 + 500 + 125 )nA = 1.055uA ....... and I thought it would only consume about 230nA :). I think I might go with LTC1540 in that case.

2) In my understanding, the clock does not work in the lowest sleep mode, so the total interrupt latency would be the Comparator's + wakeup-time, no ?

1) (facepalm).  Okay, I was wrong, I looked at the Op-Amp bias currents tab in the data not for the comparator.  Those are 800 pA (the op-amp has more going on on its pins so that is why it has higher current).  Yes, a reference is needed, and it looks like you are right with 500 nA for the LDO route.  If you look at the LTC1540, they have 450nA going in the resistive divider path so it is right around 750 nA typ operation.  So yes, about 100-200 nA less.  If you do not need 100% on time for the comparator, you could get away with less by cycling everything, but if you need 100% on, then you are correct as to the current usage.

2) I do not know, I worked on characterizing the op-amp and comparator and am not very well versed in the digital system that it sits in (I am an analog guy). 

I hope this this helps, and leakage current on the pins can be positive or negative, depending on who is leaking.

Thanks again Chris,

1) That makes more sense (0.8nA) for leakage current on comparator, nice catch. The required voltage reference is a killer though. I really want to use the CC2530 analog comparator, but if I need a voltage reference that will consume 500nA, this will be too much of power (in sleep mode) and additional cost for LDO. Is there a way I can use a zener diode for a reference voltage for the cost of ~100nA (from source=3.0v coin battery) ? I think in this case your Analog comparator would be a winner :). I just can't see how the LTC1540 can achieve such low power..

2) I don't think the propagation delay is much of a problem now, I can calculate it experimentally later. The voltage reference is the issue now.

I am sorry for the delay in responding, went looking around for zenors that would have a low enough current to work, but I could only find 2 uA minimum rated current for the output voltage.  I have to say that I was suprised at how few nanopowered references are out there (some "nanopowered things had 1uA or higher quescent current).  If you have better luck finding such a zenor, then I think you have a solution.  I guess since most of the references are so much more accurate than LDOs, they use more current (also that lowers the noise).  Good luck, if I find something I will post here.

chris

yes, same here. I found that LDO have lower power consumption, but I know what you're talking about when it comes to output voltage accuracy and noise, which is a big deal when you are dealing with high speed comparators. Indeed, the high speed TLV3501 gave me lots of headache because my input voltage and reference voltage was very noisy, so my output voltage ended up ringing at very high frequency, which effected the main supply and things just got worse from there. I tried increasing the hysteresis but it was no use.

I would like to try the LTC1540. I know it is a competitor chip, and an expensive chip, but it would solve two problems 1) ultra low power, and 2) voltage reference. I can even use the voltage reference from the LTC1540 for my higher speed TLV3501.

You are right though, I think there is a huge market for an ultra low power voltage reference <200nA. I don't know if this is reasonable, but I think it is possible.

Thanks Chris !

I have asked one of the micropower guys about this, hopefully he will have some insight to nanopower references.  Years back for another company he designed a comparator for 1 volt operation and had a 1 volt voltage reference design that they turned down.  We didn't understand what good a comparator is if you can not work a reference at that same voltage.

Let me know how that works for you, I forget if there are any noise specs on the LTC part, but I guess you can always add C or LC to quiet it some.

Good luck with your design.

Chris

Hi Chris,

As you ar the analogue expert on this you might be able to give more info on the speed/bandwidth of the comparator.

If I have a sine wave and want to wake up on amplitude level, what is the max frequency the sine could have:
I assume it it limited by op_amp bandwidth mostly?
100Hz, 100kHz, 500 kHz?

BR

Ørjan

The comparator is relatively slow as it only uses 230 nA or so.  I have not done tests to see what speed signal the comparator can capture.  The op-amp and the comparator are completely separate, so unless the signal is first going through the op-amp, then its speed should have no bearing on the comparator response.

Also, it really depends on how far beyond the comparator trip point the signal goes, and duration of time that the signal is above the threshold.  I am thinking a test with a variable pulse to see if the comparator triggers would be the way to check this out, I have some of the parts on the bench right now so I can give a look for typical numbers.

Thanx Chris,

Some typical number would be great!

BR

Ørjan

I looked into the design documentation of the comparator and found that the comparator has a simulated typical delay equal to 3.5us when the difference between the two input signals are 10mV and they are around avdd/2. For a larger overdrive the delay will be less.