Electronics > Projects, Designs, and Technical Stuff

Current source feedback capacitor

<< < (3/9) > >>

mikerj:

--- Quote from: OM222O on May 16, 2019, 05:17:17 pm ---
--- Quote from: NiHaoMike on May 16, 2019, 05:11:28 pm ---Add a resistor between the shunt and inverting side of the opamp so that the opamp isn't trying to drive a heavily capacitive load.

--- End quote ---

wouldn't that defeat the purpose of the capacitor which is to provide quick feed back?
Also I think this is way too many extra components to stabilize an op amp and feels like wrong design.

--- End quote ---

It certainly isn't "too many components", I'm puzzled why you would think this?  Just look at the schematic and ask yourself how much of the signal being fed back by the 1uF capacitor is actually going to make it to the op-amp inverting input when you are shunting it to ground with a 1 ohm resistor?

If you are going to change that shunt resistor by 5 orders of magnitude then you have a problem since you are going to end up with a very large resistor value isolating the cap from the shunt (megohms) and a correspondingly small capacitor, making the circuit performance more dependant on parasitics.

Why don't you just simulate the the circuit and look at the open loop frequency response, so you can check gain and phase margin?

OM222O:

--- Quote from: Benta on May 16, 2019, 07:04:47 pm ---
--- Quote from: OM222O on May 16, 2019, 06:47:28 pm ---the main goal was to avoid fiddling around with components and just trying out random values  :-/O so far nobody has posted an equation or solution to check whether there will be loop stability problems or not. I did a bit of search but pretty much every video uses gain of the op amp which in this case is not easy to measure?

--- End quote ---

Your issue is really extremely odd. I've built more opamp/transistor current sources than I can count and have never had problems like this.
Could you say something about the oscillations? frequency, amplitude/measuring method/probe...

--- End quote ---

Here is a video using the TLV9002 which is the worst one by far:
https://streamable.com/ioakx

the scope is AC coupled and it's set to 20mV/div.
the 1A range was so bad that it would end up at about 700mA instead of 1A so I excluded it for this test.
the ranges tested in order are: 100uA,1mA,10mA,100mA and 10uA.
As you can see 100uA and 1mA are fine, 10mA has 40mV pk-pk oscillations, 100mA is fine again but 10uA has some weird triangular looking oscillation  :-// and as mentioned before 1A is excluded as it was a total failure! in the meanwhile replacing it with the MCP6002 (has the same footprint and I had a few on hand) fixed the issue all together, only the 1A range had slight oscillations, but nothing too bad! I made another board and tested the MAX4238 and MCP6V01 (again, pin compatible) and this time MCP6V01 had some issues on a few ranges and the MAX4238 was fine |O
the problem is not consistent at all and I'm not sure what I'm doing wrong here. In the original design I didn't have a base resistor for the darlington pair, so I cut the track and bodged a 1K resistor to the TLV9002. no improvements. I then bodged a 100nF cap from the output to the inverting terminal and again, almost no change.

here is the schematic if it helps:

Fets are digital N channel fets and are controlled by an MCU, but on the second board which I mentioned, I just used jumper pins to eliminate any effects they could have, but the results were the same, so they can't be to blame in this case. I thought maybe buffering the 1V input is causing some sort of delay / phase shift so I tried it with only one side of the TLV9002 but it was just as bad. it really seems op amp dependent rather than whatever else is on the circuit which is very frustrating!

OM222O:

--- Quote from: mikerj on May 16, 2019, 08:11:02 pm ---
--- Quote from: OM222O on May 16, 2019, 05:17:17 pm ---
--- Quote from: NiHaoMike on May 16, 2019, 05:11:28 pm ---Add a resistor between the shunt and inverting side of the opamp so that the opamp isn't trying to drive a heavily capacitive load.

--- End quote ---

wouldn't that defeat the purpose of the capacitor which is to provide quick feed back?
Also I think this is way too many extra components to stabilize an op amp and feels like wrong design.

--- End quote ---

It certainly isn't "too many components", I'm puzzled why you would think this?  Just look at the schematic and ask yourself how much of the signal being fed back by the 1uF capacitor is actually going to make it to the op-amp inverting input when you are shunting it to ground with a 1 ohm resistor?

If you are going to change that shunt resistor by 5 orders of magnitude then you have a problem since you are going to end up with a very large resistor value isolating the cap from the shunt (megohms) and a correspondingly small capacitor, making the circuit performance more dependant on parasitics.

Why don't you just simulate the the circuit and look at the open loop frequency response, so you can check gain and phase margin?

--- End quote ---

I tried using ltspice but the zip folder of spice parts provided by T.I didn't want to play nicely. I'm also not sure how to simulate for the open loop frequency response? if you can guide me a bit, I might be able to do it. thanks.

mikerj:
Linear Technology have a really useful video showing exactly how to do this step by step.

Benta:
After seeing your schematic, a lot of things become clear.
The opamps are working correctly (anything else would have surprised me), but your circuit is not in the least doing what you think.

You are doing a lot of no-nos.

The simple opamp/transistor current source (which you showed in your first post) is supposed to be used with the load placed between supply and collector of the output transistor. This always works!

Your circuit places the load at the emitter of the transistor (I have surmised that "Force" is your load). In series with your load, you have the current sense resistor of 1 ohm. And on top of that, you have a feedback resistor of ~200 kohms between load and sense resistor.

I can hardly imagine a setup that's more prone to instability than this, adding on top that the load is an unknown.

Re-design needed.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod