-0.7V rail creation for op amp

[snarkysparky]

So i want to use a LMC6482 op amp for a CPU A/D converter input buffer with only a single supply.  The amp will go close to the ground rail but not all the way ( 20mv  or 6 cnts with a 10bit 3.3Vref)  and near ground the sinking ability diminishes which would non linearly slow the settling time to ground.

My idea is to generate a 0.7V negative rail to supply just the op amp.  Plan to use one of the CPU outputs to toggle some kind of switching circuit with a few diodes sprinkled in.  I have an idea involving an inductor and some diodes but I would think you guys have some better ideas.  So let me have em please.

[ogden]

You may consider purpose-built negative bias IC like http://www.ti.com/product/LM7705 In case this one does not match your needs, there are many other Switched Capacitor Inverter IC's around.

[snarkysparky]

Wow,  whoda thought an IC for EXACTLY what I need.

Thanks all.  I think i will mess with blues solution in spice to see if I can get something to work.

[StillTrying]

The LMC6482 can do <20mV while sinking 0.15mA according to the data sheet. Page 10 Fig 8.

[David Hess]

Another way to generate roughly -0.5 volt at 10s of microamp is to use a diode or bipolar transistor optocoupler as a single junction photovoltaic generator.  4N25 style optocouplers are suitable and can be had for less than 25 cents each.

[ogden]

Another way to generate roughly -0.5 volt at 10s of microamp is to use a diode or bipolar transistor optocoupler as a single junction photovoltaic generator.  4N25 style optocouplers are suitable and can be had for less than 25 cents each.

That would work for input offset (compensation) circuit, but not as negative supply for opamp OP is looking for.

[David Hess]

Another way to generate roughly -0.5 volt at 10s of microamp is to use a diode or bipolar transistor optocoupler as a single junction photovoltaic generator.  4N25 style optocouplers are suitable and can be had for less than 25 cents each.

That would work for input offset (compensation) circuit, but not as negative supply for opamp OP is looking for.

Huh?  Why not?  I've gotten 200 microamps out of a 4N25 before. They are cheap enough to use several in parallel.

[ogden]

Another way to generate roughly -0.5 volt at 10s of microamp is to use a diode or bipolar transistor optocoupler as a single junction photovoltaic generator.  4N25 style optocouplers are suitable and can be had for less than 25 cents each.

That would work for input offset (compensation) circuit, but not as negative supply for opamp OP is looking for.

Huh?  Why not?  I've gotten 200 microamps out of a 4N25 before. They are cheap enough to use several in parallel.

Just quescent current of mentioned 1MHz GBW LMC6482 dual opamp is 1mA and it shall drive capacitive input of ADC. I would say that at least 10mA shall be available. Your price argument also is kinda shaky because for 0.8$ @100 you get 50mA NCP1729, 4n25 costs 0.42$ @100 (mouser.com)

[David Hess]

Just quescent current of mentioned 1MHz GBW LMC6482 dual opamp is 1mA and it shall drive capacitive input of ADC. I would say that at least 10mA shall be available. Your price argument also is kinda shaky because for 0.8$ @100 you get 50mA NCP1729, 4n25 costs 0.42$ @100 (mouser.com)

When I did it I used the low power LMC6462 variant (and later LMC6061 because the LMC rail-to-rail input parts have horrible precision) of the LMC6482 but there is a better way; use the negative bias current to directly pull the output down to and slightly below zero directly.  The common mode input range extends slightly below ground anyway and the output does not mind 10s to 100s of millivolts below the negative supply.  This of course applies even more so to the 358/324 which only provide an internal 50 microamp pulldown current with compliance limited by Vce saturation but I never liked them for single supply outputs anyway.

The only thing to watch out for is forcing the base-emitter junction into breakdown if you use it but this is not a problem in 5 volt circuits and the base-collector junction does not have this limitation.  There are some higher performance solutions but 4N25s are cheap.

There is another way to do the above using just a bipolar transistor to replace the optocoupler but the efficiency and performance is worse so it is not worth pursuing.  Many 2N3904s died to provide this information.

[duak]

Blueskull, is the 2N3055 trick the one where the base-emitter junction is reverse biased to breakdown, but a small negative voltage relative to the emitter appears on the collector?

If so, Pease had something on it in one of his columns.  I think it was from early 2009 but I can't find it now.  Here's a couple of links to more info:
 http://electronotes.netfirms.com/ENWN28.pdf
 http://www.planetanalog.com/author.asp?section_id=3330&doc_id=564080&_mc=NL_PA_EDT_PA_20151028&cid=NL_PA_EDT_PA_20151028&elq=1ec4b561d4844ba9a0dbc78607566156

If not, then what is the 2N3055 trick?

Cheers,

[exe]

Afaik, there was an IC that generated -0.2V across its leads. It was connected in between ground negative supply of the opamp. I can't find it. I saw it in old app notes.

But the more I think about it the less I'm sure it's possible to create such a two-terminal device (unless it's a battery).  May be I'm dreaming.

[ogden]

Afaik, there was an IC that generated -0.2V across its leads. It was connected in between ground negative supply of the opamp. I can't find it.

I did mention such already:

You may consider purpose-built negative bias IC like http://www.ti.com/product/LM7705

[David Hess]

Blueskull, is the 2N3055 trick the one where the base-emitter junction is reverse biased to breakdown, but a small negative voltage relative to the emitter appears on the collector?

I am sure it is.  I did it with 2N3904s and while it works, optocouplers provide at least 10 times the efficiency.

[TrickyNekro]

Does your ADC have a -Vref? Cause if yes you can bias the -Vref with a simple diode and resistor. Granted you lose a bit of voltage range but that can be accounted for.

[Hiemal]

As an alternative you can use a 555 timer as a charge pump to create a negative voltage.

[Zero999]

As an alternative you can use a 555 timer as a charge pump to create a negative voltage.

Yes, let's not forget the classic 555 timer circuit. It will need a lot of filtering though to get rid of the noise. Another option is to replace the 555 with a different oscillator. A Schmitt trigger oscillator using the 74HC1G14 would be perfect: it would use less power and space than the 555.