-7

I'm trying to go over the problem of the rail to rail limitations of my op amp to get a full reading range over a current sensor and I thought offsetting the negative supply just of the op amp IC around 200mV is enough to solve the rail limitation problem..

Would love to hear any suggestions as to how it's being done in a standard way.

BTW here are specs of the AD4522-2 output rail offset to those who still believe everything in their lives needs to cost *1000 than actually is to function normally.

AD4522-2 specs

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208
MikeLemo
  • 155
  • 1
  • 1
  • 7
  • Comments are not for extended discussion; this conversation has been [moved to chat](https://chat.stackexchange.com/rooms/117884/discussion-on-question-by-michael-how-to-give-op-amps-negative-supply-a-slight). – Voltage Spike Dec 31 '20 at 16:11

2 Answers2

3

EDIT 2:

Given that OP's opamp comes from aliexpress at a fraction of the cost and the input offset seems huge, "1.7-2A" on a 20A scale, it is likely that the opamp is fake or that there are other design issues.

EDIT:

At very low voltages in the uV range, you have to consider thermocouple effects on the solder joints that can give you offsets.

A simple way is to use a capacitor pump voltage inverter like the MAX871.

Another way is to offset your input ground reference, but that can rapidly become tricky, all depending what you actually measure, whether it is floating. If you can afford the few extra cent of the voltage inverter, that's the way to go.

Another one is to find an op-amp that can take an input a bit beyond their rail values, some instrumentation amplifier can do so to some extent, but you also need to offset the output of the op-amp so AD8226 might be a contestant as it's input can be as low as -0.1V at 25°C (assuming V- is at GND) and its output can be offset.

enter image description here

As Spehro Pefhany mentioned in the comments make sure the opamp output remains within the ADC input range.

Damien
  • 7,827
  • 1
  • 12
  • 29
  • Nah I just need a way to slightly drop GND of my op amps supplys only with as little components and tradeoffs as possible if i'd want to make it more complex with this I'd just go for this IC https://www.aliexpress.com/item/4000194831369.html?spm=a2g0o.cart.0.0.64fa3c00dgPyNs&mp=1 and have negative voltage for the opamp altogether. – MikeLemo Dec 31 '20 at 09:45
  • I know a method to achieve this with 2 caps and 2 resistors though still looks like it could be simpler and I don't know how to calculate each value based on which parameters it might affect. – MikeLemo Dec 31 '20 at 09:47
  • 2
    @Michael Damien's +1 suggestion avoids the problem of having to clamp the op-amp output so it doesn't damage your ADC, which using a -5V supply does not. – Spehro Pefhany Dec 31 '20 at 09:48
  • 1
    If you put the board upside down maybe some electrons will fall off and drop the voltage ever so slightly. In my experience, using a trusted chip supplier is important if you don't want to sometimes waste hours of understanding why your circuit doesn't work over some fake chips. – Damien Dec 31 '20 at 09:48
  • @Damien Yeah wish flipping this PCB would make that work and you are totally right but there is still no source of e components that can suffice the need to buy components occasionally cheaply enough so aliexpress is your evil friend here and no Spehro the ADC will be fine with the + rail at 3.3V it's just that anoying offset that blocks the first 20% of measurement range – MikeLemo Dec 31 '20 at 10:27
  • just ordered the AD8226 for further investigation but it's atleast two weeks down the road and I need a solution now... – MikeLemo Dec 31 '20 at 10:28
  • 2
    There should be plenty of other opamps that can fill your need if this one is not available. You can use Digikey parametric search. Digikey is fairly cheap and quite quick. I'm located in China and I don't buy any components from China or any supplier in China, even resistors, as I had too many bad experiences of wasting hours of debugging stuff that wasn't design related. Buying an opamp on aliexpress you have 90% chance it is not the original part but some other cheap rebranded chip with somewhat similar specs and only an X-Ray machine will tell you the difference. – Damien Dec 31 '20 at 10:32
  • Sure thing but paying 100$ shipping for every time you need a few components and kits gets quite expensive over time for hobby projects. – MikeLemo Dec 31 '20 at 10:46
  • 1
    It depends on how you value your time. I usually have several projects in parallel and order once a week what I need. Shipping for me is around 50$ and I'm on the other side of the planet. Shipping cost, customs clearance, and all the delay is quite a pain, but you might spend a few months developing a board and when you try to make more of them it doesn't work anymore, you start again from point A, just because you had a fake component to start with. Is your AD4522-2 from aliexpress? :P – Damien Dec 31 '20 at 10:51
  • Yes but I'm a complete product developer of physical producs and I need to squize process sometimes to get things done and yeah start from point A is sometimes healthier but these particular solution could have a good long term impact over my future designs so I might as well optimize those elements as much as I can and yes that AD4522-2 is from aliex costs around 2$ per ic per 10Units it has the ADA logo on it and sharp silk screen so it looks original (Or a very good fake... :P) – MikeLemo Dec 31 '20 at 11:07
  • 3
    This chip cost 8$ a pop, if you think you can have it in aliexpress at 0.2$ a piece I can 100% guarantee you that they are fake and is most likely the reason of the performance problem you have in your design, and the reason of your post. – Damien Dec 31 '20 at 11:12
  • @Michael it's obviously not a very good fake if it misbehaves this much. You're wasting your time. Shipping isn't 100$. You just finding apologies to use bad electronics. – Marcus Müller Dec 31 '20 at 11:17
  • @MarcusMüller why misbehave this much? isn't it common that the output of an opamp doesn't reach the supply rail exactly, like ever? – MikeLemo Dec 31 '20 at 12:13
  • for a nominally rail-to-rail amp: yes, that is not according to specs. Specs! – Marcus Müller Dec 31 '20 at 12:18
  • @Damien You have clearly gone off the topic and the question.... I've just asked for a simple way to set a GND lower by a few milli volts I've asked for an opinion about where I source my components as I'm aware some may be fake but those that aren't work well enough and are cheap enough to make commercial... – MikeLemo Dec 31 '20 at 14:40
  • @MarcusMüller OK can you tell me what is the rail blocking offset suppose to be for the "Original" AD4522-2? there according to the datasheet I see it's 20mV and I've set the to 10 and my "measured output offset" is ~200mV you guys should be more clear on the subject or get and Electronics degree and come back... – MikeLemo Dec 31 '20 at 14:42
  • 3
    @Michael thank you for being so clear about how much respect you bring to the people trying to help you. Have a nice day. – Marcus Müller Dec 31 '20 at 14:50
  • Let us [continue this discussion in chat](https://chat.stackexchange.com/rooms/117882/discussion-between-michael-and-marcus-muller). – MikeLemo Dec 31 '20 at 14:53
  • 1
    No, thank you very much. Have a nice day. – Marcus Müller Dec 31 '20 at 14:53
1

Easiest solution is to pull up the + input enough get past the dead band. In your other question you say it doesn't respond until the 0.01 Ω shunt is passing ~1 A, which corresponds to 10 mV. You can add 10 mV to the input with this circuit:-

schematic

simulate this circuit – Schematic created using CircuitLab

Bruce Abbott
  • 55,540
  • 1
  • 47
  • 89
  • Thanks I currently looking into this solution until I come across an elegant way to offset the actual supply ground would you say the summing amplifier solution in this article be equivalent to what you are showing? https://www.electronicshub.org/summing-amplifier/ – MikeLemo Dec 31 '20 at 12:07
  • Also what prevents us from taking the value of R1 and change it to 330K instead of 3.3K 1mA of constant current draw for a little offset seem power wastive and a little not temperature stable. – MikeLemo Dec 31 '20 at 12:21
  • Nothing stops you from using larger resistor values. I only chose those values to make the principle more obvious (1k = 1V). However 1mA isn't much and temperature stability should be fine if metal film resistors are used. This bias creates an offset of 1A, so a 10% variation would equate to 100mA which is only 0.5% of 20A. More concerning is the use of a 5 W shunt that dissipates up to 4W. You might consider using a lower shunt value to reduce power loss and keep the temperature down. – Bruce Abbott Dec 31 '20 at 16:31