INT- TLV3501 in low speed application

Hi Hawk,

It is not advisable to apply a slow slewing input signal to a very high speed comparators such as the TL3016 and TLV3501. The slow slewing waveform spends significant time relative to the comparators switching time passing through the hysteresis region. In the case of the TLV3501 image the output appears to be switching back and forth at a very high rate as the input waveform passes through the region. Any noise or nonlinearity of the input waveform can create an ambiguous condition where the comparator switches state. And the comparator is operating in the linear region at mid-scaleand any internally generated noise adds to the threshold uncertainty.

The TL3016 images shows a smaller amplitude signal appearing at the output when the input passes through the midpoints of the waveforms. The expanded edge of the pulse waveform shows a sine wave with a much higher frequency. Some comparators will oscillate if the threshold region is traversed very slowly. Remember comparators are somewhat similar to operational amplifiers in their internal design, but they are not intended operate in the linear region and they are not compensated like them either. Comparators are usually driven by waveforms with fast edge rates such that the input does dwell about the threshold for long. The oscillation doesn't gets underway, or its duration is so brief that its impact is minimal. The comparator output is intended to be at one output level or the other and spend the minimal time in between them.

Regards, Thomas

PA - Linear Applications Engineering

Thomas,

 it is good to see you here.

thanks your comments, however,  that is to say comparator output will depend on input signal rising edge plus amplitude, but amplitude only.  since the feeding signal is from high precision signal generaor, i suppose, it should be very smooth. and, my customer told me, if a higher input frequency, the output will be stable and clean.

here is my confusions as below.

 - why the comparator output could be affected by rising edge too ? the integration plus external Hysteresis circuit are both equiped.

- if a slow compartor is the better solution, what is the suggestion between input signal frequency ,or rising edge,between the response time of a comparator ?

thanks in advance.

Hawk Tong

Hawk,

- why the comparator output could be affected by rising edge too ? the integration plus external Hysteresis circuit are both equipped.

Are you sure that you test environment represents a good high-speed environment. When employing high-speed comparators the ground must be as close to zero impedance as practical. Otherwise you can get ground bounce. Similarly, you must have effective power supply decoupling at the supply pin that is has minimal ESR and ESL.

- if a slow comparator is the better solution, what is the suggestion between input signal frequency ,or rising edge,between the response time of a comparator ?

I do not know of any hard, fast rule, but I would suggest an edge rate that is tens, or hundreds of times, faster more than the slewing rate atthe threshold point. The high -speed comparators draw relatively large operating current and much can be saved by using slower ones. The TLV3401 and TLV3701 comparators are very low power comparators that may be completely acceptable for the application.

Regards, Thomas

PA - Linear Applications Engineering

Thomas,

 I double checked with my customer. the guy expect to convert sine wave to sequare wave ,  input frequency varies from 40kHz ~ 10MHz and amplitude varies from 1V to 5V that is why a high speed comparator is used here. if the input frequency is as high as 1MHz or more , it works normally. 

it seems neither TLV3401 nor can deal with wide input frequecy. Do you have other suggestion for this case ?

Best regards

Hawk Tong

Hello Hawk,

It appears that much of TI's comparator focus is on very low current comparators so there are not many high-speed alternatives. I suspect that most very high-speed comparators - regardless of manufacturer - will have difficulty with the slow edge of a 40kHz sine wave. It doesn't surprise me that  a 1 MHz, or higher, sine wave works nicely with these high-speed comparators.

SLL offers a couple of somewhat slower comparators that the TL3016 and TLV3501; they are the TL712 and LM306, but I am not certain that either one of them will remedy the fundamental issues we have discussed about using a high-speed comparator for very slow speed applications.

Regards, Thomas

PA - Linear Applications Engineering

Thomas,

I find a material about comparator application on EDN web, Comparing comparators: Measure signals, get results

http://www.edn.com/article/457866-Comparing_comparators_Measure_signals_get_results.php

My idea is how about to try high speed OpAmp for this case, such as OPA836. In doing so, the customer could benefit wide range of frequency and amplitude.

I try some simulation in TINA, it seems when the input frequency is high, the output wave will be decrease.

Another risk is the absoult input to Vin+ and Vin- may be too much for this part, if OPA836 is used in open loop mode.

Appreciated  your comments here.

Thanks a lot.

Hawk

Hi Hawk,

Indeed an operational amplifier can be used as a comparator, but as you know there design requirements are fundamentally different. I suspect that the controlled slew rate of the operational amplifier such as the OPA836 would be a good adavantage in this application. Since the OPA836 is an HSP operational amplifier I am not too familiar with its specifications, but do make sure that the maximum differential input voltage specification does not get exceeded.

I am sure the HSP group can guide you through any questions you might have with the OPA836.

Regards, Thomas

PA - Linear Applications Engineering