LM385 Low gain on small differential voltages

[nyaknyan]

I'm trying to make hi-side current sensor.
I used circuit from https://www.intersil.com/content/dam/Intersil/documents/an18/an1827.pdf and modified it by using separate power supply, so no need to use rail2rail opamp, because signal level is only about 1.4v greater than opamps's ground (this voltage is set by two diodes and resistor r3). Also I added zenner diode to protect opamp when there is short circuit in load, because i'll use it for lab power supply and output voltage can reach 60v

In ltspice it worked well, but in real life I used LM358 and i found out that it has low gain when differential voltage is low. My self-made shunt resistor has a voltage drop about 10mv per 1A and with 100r and 1k resistors circuit should have gain 10. But it only reaches normal gain (8 because shunt is not exactly 10mv/A, but i calculated like 10mv) when differential input voltage is greater than 6-7mv and can't sense 0-5mv, but I want to have current regulation in 0-1A range too, and can't increase shunt resistance, because power supply will be atx smps based i can get 50A from it on low voltages

Here is my circuit and results of measuring


What opamp i should use? And I haven't found gain on small differential signal in lm358 datasheet, which parameter I should look at? Can someone advise some cheap opamp with good characteristics for my case, which i could get on ebay?

 

[ronv]

Your schematic is a little hard to read, but it looks like tour input can be at a higher voltage than the op amp positive supply. They don't like that.

 

[nyaknyan]

No, it can't. Amplifier woks from separate power supply (separate winding only for it on separate transformer). It's v2 on schematics. Point (lets say P1 ) where v1 connected to shunt/two diodes/r4 always has a potential equal 2 diode voltage drop(i used tl431 instead of diodes in real circuit, so it is 2.5v) from the amplifier's point of view, and point after shunt where it connected to load has potential of (P1 - drop on shunt) and it can go as low as v1 voltage is, but zenner with 5k resistor protects amp's (-) input from getting voltage(which can be -120v) lower than amplifier's supply.

Also i have tested it with 9v battery as amplifier supply v2 and 5v power supply as v1, and there is no way to exceed amplifiers supply.
In ltspicie it works ok, but in real life lm358 has very little gain when non inverting input is 2.5v and non inverting is 2.498v

 

[MikeMl]

It is really hard to make a DIY high-side current monitor using a cheap opamp. To do it right requires opamps with very low offset and very tightly matched resistors. I have used the **broken link removed**and other similar high-side current-monitor ICs. They are extremely easy to use, cheap, and require very few external parts.

 

[nyaknyan]

As i see, **broken link removed**has maximum voltage only 20v. In my circuit output voltage is limited only by voltage of pnp transistor and isolation between v1 and v2 . And there is no need to have matched resistors in this circuit, there is no voltage dividers. Same current flow per r4 and r6, and drop on r4 is equal to drop on shunt resistor. Output voltage is on r6. And i plan to make two side power supply +60v(60v is max, 0-60 regulated) GND -60V so i have similar circuit with npn transistor for side below GND

 

[MikeMl]

If you need 60V offset, try **broken link removed**

 

[nyaknyan]

I figured out my problem - it's because of offset voltage. Difference between inverting and non inverting output is 1.8-2.5mv , lm358 doesn't make it 0v, there is offset voltage error. I was aware about offset voltage, but thought it is offset from gnd, and doesn't apply when one signal is for eg. 2.5v and other 2.499v. I had lack of understanding what offset voltage is, now i see that i just need low offset amplifier.

 

[MrAl]

Hi,

Your schematic is so hard to read you're lucky anyone can read it at all.
It looks like you also need a rail to rail input op amp but it's hard to tell from one of the worst drawings i've seen in a long time.
Maybe if you redraw it with nothing typed over anything else it will help too

Also, not a very good choice for the name of the file people will be downloading from your post.

 

[nyaknyan]

Sorry for the messy schematics, here is better one.
No, there no need to use rail to rail since i have separate power supply. I going to use OP07C opamp
Renamed file in the first post

 

[MrAl]

Hi,

Oh that's much better

I guess you want a gain of 0.1 volt per amp?
Do you need three more diodes anti parallel to the other three series diodes?

 

[nyaknyan]

>I guess you want a gain of 0.1 volt per amp?
Yes, but it doesn't matter, it can be easily changed by changing 100r and 1k resistors.

>Do you need three more diodes anti parallel to the other three series diodes?
No. Why i should need them? R1 and diodes are there to set potential of main power circuit part with v2/shunt/load to about 2v relative to opamp's ground. I actually used tl431 in real circuit instead of diodes, and zenner diode in 2-7v rage can be used here too. Because of this 2v voltage there is no need to have rail2rail amplifier like in original circuit from intersil appnote - voltages on shunt resistor sides are 2v and (2v - shunt voltage drop) . But my version requires separate power supply only for this opamp, in my case it's not a big problem, since i making lab power supply from atx pc power supply and just added extra windings on standby transformer.

Circuit itself works fine, so there is nothing to discus since i figured out that my problem was in LM358 offset voltage, which on small signals is as high as signal itself, causing large error. So at monday i'll go to electronic shop and buy OP07 opamp, it has more than 20 times less offset voltage and pins for connecting trimmer to set offset voltage to even lover level.

 

[MrAl]

Hi,

Ok great, i am happy to hear you got it working the way you need it to. Will be interesting to hear how the new op amp works with it in real life too.

 

[tomizett]

Nice circuit.
Just thought I'd mention that, in general, if you're interested in minimising offset it's a good idea to keep the input impedances equal on both inputs of the op-amp. That way, an equal input bias current on both inputs creates the same voltage drop on both inputs and doesn't upset the offset. If the input impedances are different, then even with identical bias currents (ie, zero input offset current) there will be a voltage difference between the two inputs. Hope I'm using the right terminology here, by the way...

In you're circuit that would mean making R3 and R4 equal - don't know if that's practical or not? Alternatively, you should be able to trim out this kind of offset quite easily.

 

[nyaknyan]

No, this is not practical. If i make 100r greater, current flowing through it and r7 will be very small and i worry about interferences from working transformer and other things, because this circuit will work inside smps. And if i'll make r4 smaller it will burn out with all circuitry, because if load is shortened then voltage on it will be -120V in worst case, actually i'll make it about 50k
So i'll use op07 trimming pins and multiturn trimmer resistor

 

[nyaknyan]

Tested it with OP07 - works great! Without offset trimmer it had offset less than 100uV and it was 1mV at circuit output with no load. And with trimmer i was able to get 0mv at output when no current flowing and 10ma = 1mv resolution

 

[MrAl]

Hi,

Sounds pretty good. Depending on bandwidth needed there are other op amps out there too that dont need adjustment because their offset is so low.

 

[tomizett]

Good work - glad to hear it's working.