Comparator with driver output

Am 17.11.22 um 21:20 schrieb sonnic...@gmail.com:
> Hi all
> 
> A standard comparator f.x LM393, LM2903 etc commonly have a open-collector output.
> Does the group know about comparators with driver output, hence can pull up and down, and output like an ordinary opamp?
> 
LMV7219 M7

cheers, Gerhard

whit3rd wrote:
> On Thursday, November 17, 2022 at 4:56:17 PM UTC-8, Ricky wrote:
>> On Thursday, November 17, 2022 at 5:43:35 PM UTC-5, whit3rd wrote:
>>> On Thursday, November 17, 2022 at 1:06:03 PM UTC-8, Ricky wrote:
>>>> On Thursday, November 17, 2022 at 3:20:08 PM UTC-5,
>>>> sonnic...@gmail.com wrote:
> 
>>>>> An ordinary opamp does not fit my need, as my application
>>>>> could balance itself
> 
> (I interpret this to mean he wants to exclude midrange parking of the
> output... which generally just requires some positive feedback,
> creating hysteresis)
> 
>>>> This is when I use the Digikey search. If you type in
>>>> comparator it will take you to a search page...
> 
>>> Yeah, but... that's in-the-box searching, and it'll miss
>>> possibilities like using an op amp (or an unused section of a
>>> quad) or a discrete two-transistor circuit, or NE555, or TLV431,
>>> or power sequencer, all of which have comparator character. The
>>> "comparator" label isn't always the useful first-search-box
>>> criterion.
>> Not sure what your point is. If you want a comparator, Digikey is a
>> good way to search for one. If you want an alternative circuit,
>> then search for that. What do the two have to do with one another?
> 
> The application could be comparison to  midrange of power supply, and
> benefit from hysteresis, and require a high current output... so, an
> NE555 is a good off-the-shelf solution. Or it could just need a
> repeatable ON threshold, and a CMOS power monitor  would do that. A
> two-transistor Schmitt trigger also has bistable output and an input
> threshold. True comparators with input common-mode insensitivity are
> not the only choices in reality, but are the only ones that will show
> up in a search.
> 
>> So how do you use two transistors to form a reasonable comparator?
>> Is this something that only works in some special set of
>> conditions? Some sort of positive feedback, perhaps?
> 
> I often use resistors to add positive feedback to a comparator, and
> of course the Schmitt trigger is a useful two-transistor circuit,
> might be preferable.  Once you use positive feedback, an op amp is
> interchangeable with a comparator (no need for stability as with
> negative feedback, just gain).

Not so, but far otherwise.  Internally-compensated op amps trade off a 
lot, a lot of slew rate for feedback stability.

Plus external positive feedback is fairly slow, on account of input 
capacitance and internal delays, potentially leading to artifacts for 
wideband inputs.  Internal hysteresis can be nice and fast.

Oh, and of course it causes kickout from the noninverting input.

I used to really like the MAX900 quad 10-ns comparator, circa 1996--it 
had a latch output that could be driven from its own output (via an RC 
highpass) without misbehaving.  That made it easy to get a repeatable 
pulse without kickout.

Cheers

Phil Hobbs
-- 
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

On Fri, 18 Nov 2022 10:32:35 -0500, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

>whit3rd wrote:
>> On Thursday, November 17, 2022 at 4:56:17 PM UTC-8, Ricky wrote:
>>> On Thursday, November 17, 2022 at 5:43:35 PM UTC-5, whit3rd wrote:
>>>> On Thursday, November 17, 2022 at 1:06:03 PM UTC-8, Ricky wrote:
>>>>> On Thursday, November 17, 2022 at 3:20:08 PM UTC-5,
>>>>> sonnic...@gmail.com wrote:
>> 
>>>>>> An ordinary opamp does not fit my need, as my application
>>>>>> could balance itself
>> 
>> (I interpret this to mean he wants to exclude midrange parking of the
>> output... which generally just requires some positive feedback,
>> creating hysteresis)
>> 
>>>>> This is when I use the Digikey search. If you type in
>>>>> comparator it will take you to a search page...
>> 
>>>> Yeah, but... that's in-the-box searching, and it'll miss
>>>> possibilities like using an op amp (or an unused section of a
>>>> quad) or a discrete two-transistor circuit, or NE555, or TLV431,
>>>> or power sequencer, all of which have comparator character. The
>>>> "comparator" label isn't always the useful first-search-box
>>>> criterion.
>>> Not sure what your point is. If you want a comparator, Digikey is a
>>> good way to search for one. If you want an alternative circuit,
>>> then search for that. What do the two have to do with one another?
>> 
>> The application could be comparison to  midrange of power supply, and
>> benefit from hysteresis, and require a high current output... so, an
>> NE555 is a good off-the-shelf solution. Or it could just need a
>> repeatable ON threshold, and a CMOS power monitor  would do that. A
>> two-transistor Schmitt trigger also has bistable output and an input
>> threshold. True comparators with input common-mode insensitivity are
>> not the only choices in reality, but are the only ones that will show
>> up in a search.
>> 
>>> So how do you use two transistors to form a reasonable comparator?
>>> Is this something that only works in some special set of
>>> conditions? Some sort of positive feedback, perhaps?
>> 
>> I often use resistors to add positive feedback to a comparator, and
>> of course the Schmitt trigger is a useful two-transistor circuit,
>> might be preferable.  Once you use positive feedback, an op amp is
>> interchangeable with a comparator (no need for stability as with
>> negative feedback, just gain).
>
>Not so, but far otherwise.  Internally-compensated op amps trade off a 
>lot, a lot of slew rate for feedback stability.
>
>Plus external positive feedback is fairly slow, on account of input 
>capacitance and internal delays, potentially leading to artifacts for 
>wideband inputs.  Internal hysteresis can be nice and fast.

Some of the rrio opamps are pretty good comparators. OPA197 is a nice
ca 1 us comparator. It comes off rails clean and fast.

>
>Oh, and of course it causes kickout from the noninverting input.
>
>I used to really like the MAX900 quad 10-ns comparator, circa 1996--it 
>had a latch output that could be driven from its own output (via an RC 
>highpass) without misbehaving.  That made it easy to get a repeatable 
>pulse without kickout.

Nice trick.

MAX9690 was a wonderful 1 ns ECL comparator... until it wasn't.

ADI makes the only fast comparators that I know of. ADCMP562 has
programmable hysteresis. ADCMP582 is 37 picoseconds and has well under
30 fs rms jitter... hard to measure.

No spice models.

On Friday, November 18, 2022 at 12:56:17 AM UTC, Ricky wrote:
> On Thursday, November 17, 2022 at 5:43:35 PM UTC-5, whit3rd wrote: 
> > On Thursday, November 17, 2022 at 1:06:03 PM UTC-8, Ricky wrote: 
> > > On Thursday, November 17, 2022 at 3:20:08 PM UTC-5, sonnic...@gmail.com wrote: 
> > > > Hi all 
> > > > 
> > > > A standard comparator f.x LM393, LM2903 etc commonly have a open-collector output. 
> > > > Does the group know about comparators with driver output, hence can pull up and down, and output like an ordinary opamp? 
> > > > 
> > > > How do I search for such ones? 
> > > > 
> > > > An ordinary opamp does not fit my need, as my application could balance itself 
> > > This is when I use the Digikey search. If you type in comparator it will take you to a search page with many parameters to adjust. You can select the output type and many other features. 
> > Yeah, but... that's in-the-box searching, and it'll miss possibilities like using an op amp (or 
> > an unused section of a quad) or a discrete two-transistor circuit, or NE555, or TLV431, or power 
> > sequencer, all of which have comparator character. The "comparator" label isn't always 
> > the useful first-search-box criterion.
> Not sure what your point is. If you want a comparator, Digikey is a good way to search for one. If you want an alternative circuit, then search for that. What do the two have to do with one another? 
> 
> Interesting you mention transistor circuits. I needed to design a FF to remember power state when everything is shut down, including the MCU, which would only wake periodically. Because of the circuit, it worked better to power it from 12V, which could be as high as 15V, iirc. Digital logic was a bit too fast, making it subject to noise, etc and other concerns I don't recall at the moment (I think it had to do with undefined behavior with both set and reset asserted). So I opted for a transistor circuit, which included several more FETs for input and output drive, some in other than common source configurations. I checked it nine ways from Sunday, so I had confidence it would work fine. Since a handful of transistors would be easier to source than a couple of 4000 series CMOS, it wanted to use the transistors. Being a group project, some guy who never saw a Linear Tech device he didn't like (meaning it wasn't his money he was spending, why use something cheap when several things expensive would do?) kept criticizing it, not in engineering terms, but negging terms like kludge, etc. What he really meant, was, he didn't understand it. 
> 
> So how do you use two transistors to form a reasonable comparator? Is this something that only works in some special set of conditions? Some sort of positive feedback, perhaps? 
> 
> I can't think of any time when an opamp would be preferable to a comparator, unless the requirements were very lax and you simply wanted to reduce the BoM. 
> 
> -- 
> 
> Rick C. 
> 
> + Get 1,000 miles of free Supercharging 
> + Tesla referral code - https://ts.la/richard11209

and, FWIW, there is a 3 (?)-or-so BJT comparator circuit in
Tom Fredericksen's Intuitive IC Op Amps book.
(probably there to explain the core of a LM311)
So, Rick, you're in good company.

On Friday, November 18, 2022 at 12:55:40 AM UTC-5, whit3rd wrote:
> On Thursday, November 17, 2022 at 4:56:17 PM UTC-8, Ricky wrote: 
> > On Thursday, November 17, 2022 at 5:43:35 PM UTC-5, whit3rd wrote: 
> > > On Thursday, November 17, 2022 at 1:06:03 PM UTC-8, Ricky wrote: 
> > > > On Thursday, November 17, 2022 at 3:20:08 PM UTC-5, sonnic...@gmail.com wrote: 
> 
> > > > > An ordinary opamp does not fit my need, as my application could balance itself
> (I interpret this to mean he wants to exclude midrange parking of the output... which 
> generally just requires some positive feedback, creating hysteresis) 
> 
> > > > This is when I use the Digikey search. If you type in comparator it will take you to a search page...
> > > Yeah, but... that's in-the-box searching, and it'll miss possibilities like using an op amp (or 
> > > an unused section of a quad) or a discrete two-transistor circuit, or NE555, or TLV431, or power 
> > > sequencer, all of which have comparator character. The "comparator" label isn't always 
> > > the useful first-search-box criterion. 
> > Not sure what your point is. If you want a comparator, Digikey is a good way to search for one. If you want an alternative circuit, then search for that. What do the two have to do with one another?
> The application could be comparison to midrange of power supply, and benefit from hysteresis, and 
> require a high current output... so, an NE555 is a good off-the-shelf solution. 
> Or it could just need a repeatable ON threshold, and a CMOS power monitor would do that. 
> A two-transistor Schmitt trigger also has bistable output and an input threshold. 
> True comparators with input common-mode insensitivity are not the only choices 
> in reality, but are the only ones that will show up in a search.

Yeah, you've told me how to build a watch, but not answered the question, why would I care?  If I'm looking for a comparator, the Digikey search gets me what I want.  If I want something else, why would I search for a comparator? 


> > So how do you use two transistors to form a reasonable comparator? Is this something that only works in some special set of conditions? Some sort of positive feedback, perhaps?
> I often use resistors to add positive feedback to a comparator, and of course the Schmitt trigger is 
> a useful two-transistor circuit, might be preferable. Once you use positive feedback, an op amp 
> is interchangeable with a comparator (no need for stability as with negative feedback, just gain).

Ok, so special conditions.  Got it. 


> > I can't think of any time when an opamp would be preferable to a comparator, unless the requirements were very lax and you simply wanted to reduce the BoM.
> One time I used an op amp that way was an old design, LM301 (no compensation required) 
> when I wanted a near-positive-rail input; the old comparators didn't match that requirement. 
> On other occasions, I've simply had good stock of the op amps but not of comparators... 
> I was mainly doing one-off jobs, didn't want to delay for a trip to/shipping from a supplier. 
> Yeah, lax requirements is a major 'reason'.

Ok

-- 

Rick C.

-- Get 1,000 miles of free Supercharging
-- Tesla referral code - https://ts.la/richard11209 

On 2022-11-17, sonnic...@gmail.com <sonnichjensen@gmail.com> wrote:
> Hi all
>
> A standard comparator f.x LM393, LM2903 etc commonly have a open-collector output. 
> Does the group know about comparators with driver output, hence can pull up and down, and output like an ordinary opamp?
>
> How do I search for such ones?
>
> An ordinary opamp does not fit my need, as my application could balance itself
>
> WBR
> Sonnich

John Larkin likes using LVDS receivers as fast comparators. they will
have driver outputs.

If the input signal levels are compatible and you can wear the
typically larger offset votlage you may find that could be a solution.


-- 
  Jasen.

On Friday, November 18, 2022 at 6:00:51 PM UTC-5, Jasen Betts wrote:
> On 2022-11-17, sonnic...@gmail.com <sonnic...@gmail.com> wrote: 
> > Hi all 
> > 
> > A standard comparator f.x LM393, LM2903 etc commonly have a open-collector output. 
> > Does the group know about comparators with driver output, hence can pull up and down, and output like an ordinary opamp? 
> > 
> > How do I search for such ones? 
> > 
> > An ordinary opamp does not fit my need, as my application could balance itself 
> > 
> > WBR 
> > Sonnich
> John Larkin likes using LVDS receivers as fast comparators. they will 
> have driver outputs. 
> 
> If the input signal levels are compatible and you can wear the 
> typically larger offset votlage you may find that could be a solution. 

The trouble with using a digital differential input as an analog comparator, is they are not specified for the same sort of performance.  For example, the input offset can typically be rather large.  Maybe in experiments, they prove to work well with low offsets, but there's no good reason to use such a device in an application it is not specified for... or I should say, unspecified for. 

Also, LVDS receivers seldom have wide common mode ranges.  

-- 

Rick C.

-+ Get 1,000 miles of free Supercharging
-+ Tesla referral code - https://ts.la/richard11209

On Fri, 18 Nov 2022 22:57:17 -0000 (UTC), Jasen Betts
<usenet@revmaps.no-ip.org> wrote:

>On 2022-11-17, sonnic...@gmail.com <sonnichjensen@gmail.com> wrote:
>> Hi all
>>
>> A standard comparator f.x LM393, LM2903 etc commonly have a open-collector output. 
>> Does the group know about comparators with driver output, hence can pull up and down, and output like an ordinary opamp?
>>
>> How do I search for such ones?
>>
>> An ordinary opamp does not fit my need, as my application could balance itself
>>
>> WBR
>> Sonnich
>
>John Larkin likes using LVDS receivers as fast comparators. they will
>have driver outputs.
>
>If the input signal levels are compatible and you can wear the
>typically larger offset votlage you may find that could be a solution.

DS90LV012 is a good one. Input offset is 30 mV typ, input common mode
works almost rail to rail, prop delay is about 1.8 ns, output rise and
fall are under 400 ps, and it costs us 37 cents. We've used about 50K
of them so far.

Somebody makes essentially the same part but calls it a comparator,
and it costs about 5x as much.

On Thursday, November 17, 2022 at 3:20:08 PM UTC-5, sonnic...@gmail.com wrote:
> Hi all 
> 
> A standard comparator f.x LM393, LM2903 etc commonly have a open-collector output. 
> Does the group know about comparators with driver output, hence can pull up and down, and output like an ordinary opamp? 
> 
> How do I search for such ones? 
> 
> An ordinary opamp does not fit my need, as my application could balance itself 

You could use an op amp to buffer the LM393 OC output. Two cheap parts are still cheap. The super fast comparators mentioned are all very fussy about layout and have self-oscillation problems if you're not careful about every little thing.


> 
> WBR 
> Sonnich

On Friday, November 18, 2022 at 9:15:53 PM UTC-5, Fred Bloggs wrote:
> On Thursday, November 17, 2022 at 3:20:08 PM UTC-5, sonnic...@gmail.com wrote:
> > Hi all 
> > 
> > A standard comparator f.x LM393, LM2903 etc commonly have a open-collector output. 
> > Does the group know about comparators with driver output, hence can pull up and down, and output like an ordinary opamp? 
> > 
> > How do I search for such ones? 
> > 
> > An ordinary opamp does not fit my need, as my application could balance itself
> You could use an op amp to buffer the LM393 OC output. Two cheap parts are still cheap. The super fast comparators mentioned are all very fussy about layout and have self-oscillation problems if you're not careful about every little thing. 

No part is cheap if it takes up space on a crowded board.  

Many Greenpak devices have comparators in them, but few have opamps. 

-- 

Rick C.

+- Get 1,000 miles of free Supercharging
+- Tesla referral code - https://ts.la/richard11209