Looking for high speed (<50 nS) comparator capable of at least +/- 5 V rail to rail push-pull output (Or alternative device)

Many high-speed comparators have a separate GND, or differential/LVDS outputs.

It might be possible to combine a lower-voltage comparator and a higher-voltage buffer. What is the actual input voltage range?

Hi,

it heavily depends on the input signal and the actual circuit whether an OPAmp can be used as comparator or not. There are many tricks to make it run properly. But for this we need detailed information on your application.

But as Clemens already mentioned, I would also combine a fast comparator with a fast OPAmp as output buffer. But again detailed information on your application would be necessary.

Kai

I am trying to convert an approximately 0 to 5 V pulse train into a "precise" -5 V to +5V pulse train with minimal delay. The threshold is 2.5 V (Threshold does not need to be exact).

Can this be better/simpler done with something other than a comparator?

Since there are split supply gate drivers, do split supply buffer gates (logic gates) or level shifters exist? I don't need the high output current of a gate driver as I am dealing with signal not power here.

Hi,

you could do it this way:

hsin_sn10502.TSC

Kai

I will give this a try. But I have some questions.

Your suggested circuit looks like a non-inverting amplifier instead of a comparator configuration. So the output depends on the exact input voltage. However, what I an hoping to achieve, besides from level-shifting, is to clean up the imprecise input pulses into something that matches exactly the +/- 5V rail (or high precision +/- 5V pulses, which would be even better).

Will there be any problem if I use the SN1050X series OP-Amp as a comparator instead of an amplifier?

Thank you

Thanks for your post and everyone's suggestions.

I realize that you are seeking less than 50ns delay but I believe your best option is our TLV1805 which has a push pull output stage and can be powered direction from the split supplies of +/-5V.  This way you are only using one device to translate from 0 to 5 to +/-5V output and still use your 2.5V reference.

Chuck

There are no logic gates or level shifters that allow the output to go more negative. Your only choices are a comparator (but the TLV1805 is slow) or a gate driver. (If the gate driver's output edges are too fast, add a resistor.)

Hi.

HSIN-CHE HSIEH said:

Your suggested circuit looks like a non-inverting amplifier instead of a comparator configuration.

I wanted to focus on the output buffer.

I would add a comparator with hysteresis in front of the buffer:

hsin_sn10502_1.TSC

Kai

Today I tried to use the CD4050B. This is what I got.

Channel 1 is input (0 to 5 V), channel 2 is output (-5 to +5 V)

CD4050B seems to convert the voltage level as I desire. The threshold is not very critical for me. However, the fall time is too slow and the asymmetrical rise/fall time mess up my duty cycle.

I am currently using a non-TI part I found in my drawer. I know that TI manufactures CD4050B (or equivalent part). Would a TI part perform better ?(faster rise/fall)

Hi HSIN-CHE,

The CD4050B is not designed to output negative voltage. The TI device should clamp at -0.5V and we can't recommend operating a device outside of the absolute maximum ratings. 

As Clemens stated before, there are no logic gates or level shifters that allow the output to go more negative. Your only choices are a comparator or a gate driver.

Regards,

Sebastian

I don't think the way I use CD4050B would cause it to experience any negative voltage. The Vss of my CD4050B is tied to the -5 V bus and the Vcc is tied to +5 V bus. The input it sees would be pulses from +5 V to +10 V in reference to its Vss and the supply voltage would be +10 V. Everything appears to be within ratings.

I am thinking about redesigning my front end so it generates -5 V to 0 V pulses. The buffer or converter after that, with Vss tied to -5 V, would see pulses from 0 V to +5 V in reference to its Vss. Would this makes part selection easier?

Hi HSIN-CHE,

Thanks for clarifying you are using two supplies. I am looking into this and will get back to you tomorrow. 

Regards,

Sebastian 

Using the CD4050B with a low-level input voltage of +5 V relative to Vss would be outside its V_IL limit, i.e., there is no guarantee that it would actually switch. And CD4000 devices are very slow.

The CD4000 devices were acquired from Harris, which acquired them from RCA. What you can buy now from TI is likely to be the same die.

Using −5 V to 0 V pulses indeed would make part selection easier. However, there are no logic devices/translators that work at 10 V (except slow CD4000 series devices), so you would still need a comparator or gate driver.

You could then use cheaper gate drivers like the UCC27517, without a comparator in front.

Hi,

this would be the result with the SM74101:

hsin_tlv3601.TSC

The SM74101 allows true toggling from positive supply rail to negative supply rail, which the SN10502 cannot because of having an output stage containing BJT. This makes the circuit with the SM74101 a lot simpler.

I personally would go for the TLV3601 in combination with the SM74101. This is way better than to experiment with the old CD4000 series providing this weak and slow output stage.

Kai

I will try some of the gate driver options (e.g UCC27517 or SM74101).

Thank you