uSupply voltage follower ???

[ikrel]

I've discovered an issue while building up daves usupply for myself.

The lm358 voltage follower is not working correctly for me, it's output is always approx 2v lower then the input voltage (EDIT:this is when there is no current being drawn from the supply. did not mean it was like this at all times).

Is this because the lm358 can't get any closer to v+ then 1.5/2v as indicated by the datasheet?
If so would this mean that the supply would have to draw 2A before the follower will start to see the voltage drop?

the supply rails for the LM358 is the same as the input for the lt3080 right? whats is going on here and why does this work for Dave and not me?

[Kevin.D]

Your correct , that voltage follower  (U12A on the schematic) is going to do
nothing for the first 1.5 amps of current cause like you pionted out  max output for lm358 is Vsupply -1.5V .
I don't think Dave must have tested it  ?. I can't see how he got that to work with lm358 .

[paul_rautenbach]

Is your point the same as the comment made by Raphael (germany) on March 17th, 2012 at 00:35 in his comment under the EEVblog #259 – PSU Rev C Schematic – Part 12 video? If so, Dave acknowledged it in the next comment on this page.

I am also trying to build Rev C of Dave's uSupply so if you managed to fix this problem, please let me know what you had to do.

Thanks.

Uncle Paul

[ikrel]

i just removed the follower and used the differential amplifier without it.

[paul_rautenbach]

ikrel, thank you for your response.

The follower is in there to allow zeroing out of the current measurement. This is explained in the section of EEVblog #259 – PSU Rev C Schematic – Part 12 between 39:52 and 43:32.

Would it be possible to replace the LM358D with a pin-for-pin replacement which is rail-to-rail? Perhaps the OPA2170. I haven't been through the specs in detail but at first glance it looks like a possible rail-to-rail replacement that is not too much more expensive.

Let me know what you think.

Thanks.

Uncle Paul

[ikrel]

The chip you suggest would seem to work. the only issue I had was the LM358's inability to track the voltage below ~2v.

I would test it out if you have the opportunity, but I am not the best person to ask as I'm only a beginner myself. Although logic would dictate that your chip should work(I think).

[paul_rautenbach]

AcHmed99, thank you for your input.

I took the following snippets from the datasheet for the OPA2170 to mean the input could also be considered to be rail-to-rail for my purposes - a voltage follower in the high-side current sense part of a 0-20V power supply:

• Input Range Includes the Negative Supply
• Input Range Operates to Positive Supply
• Rail-to-Rail Output

The input can operate 100mV below the negative rail and within 2V of the positive rail for normal operation. Note that these devices can operate with full rail-to-rail input 100mV beyond the positive rail, but with reduced performance within 2V of the positive rail.

I'm not sure what "reduced performance" is referring to but if it is to do with speed of response, I assume I don't mind.

The original circuit was wrongly designed with an LM358D which is not rail-to-rail. You can look at the circuit in the schematic I referenced in my first post. The chip concerned is U12. The required resolution is only 8 bits.

I am looking for an 8-SOIC pin-for-pin replacement for the LM358D that is rail-to-rail. The chip is powered by a single supply from a pre-regulator that maintains the voltage at 2V above the output voltage of the power supply so I need a chip with a rail voltage up to about 25V.

I'm not sure what I should put into DigiKey to specify RRIO but entering "opamp RRIO" and then refining the parameters, the cheapest I can find is the LM7322. I will use this if it is necessary.

Thanks.

Paul

[paul_rautenbach]

AcHmed99, thanks again for your help.

In a previous post I said "You can look at the circuit in the schematic I referenced in my first post." Sorry, I didn't reference it there. This is the reference to it.

I think I've understood the reduced performance problems of the the OPA2170 now and see why I need a real RRIO opamp.

Do you already have a board made up?

Yes, I jumped the gun a bit with getting the PCB made. I thought I'd finished but found I had to make some changes.

I suppose Dave has his reasons for using opamps versus a dedicated current sense IC, price I guess but considering you will now need at least one RRIO amp plus 0.1% resistors you might as well just use a decent high-side current sense IC.

If I understand what "a dedicated current sense IC" is, I think Dave has used one. It's the INA219 (U5) on his schematic. This is the high-side current sensor read by the micro controller which is intended to measure the actual output current, which will be displayed. The OpAmp circuit (U12) is  a less accurate high-side current sensor, across the same shunt resistors. It feeds one side of the the current limit comparator U3B. The other side is set by the micro controller as the current limit. The follower OpAmp part of U12 is only there to avoid current which varies with the output flowing through R17, R18 and R33, as this would prevent being able to zero out the output current display.

I don't understand why Dave is not doing the comparison in software, using the current measured by the IN219, but I guess he has his reasons.

ADA4091

I've had a look at this. I can see from it's specs that it is a much better opamp than the others I've been looking at but it looks to me to be overkill and is twice the price of the LM7322. The current limiting circuit is not trying to be very accurate. It just needs to impose a rough limit. The output current measurement wants to be as accurate as possible across the whole current range (10µA to 1A). Hence the use of the INA219.

I would be grateful if you could take a look at the specs for the LM7322 in the light of what it is for and let me know if you think it is inadequate.

Thanks.

Paul

[amspire]

I don't think you need the voltage follower.

It is only there to eliminate the differential stage current from the current sense resistors, and at 18V, we are talking a 0.5mA error current.  The differential stage will have a offset error corresponding to at least 2mA, and also the LM334 across the output is designed for use as a temperature sensor, so the current changes with temperature anyway.

If you are going to use an opamp to zero out the differential stage offset, it is dead easy to precisely cancel the error (by calculation) of the differential stage. This means the main cost of omitting U12A will be if you are using the supply as a constant current source, the current will drop by 0.5mA, as you increase the output voltage from 0 to 18V.

If that is not a problem, leave U12A out, and perhaps use it for something else. A crowbar output protection circuit, or an over temperature shutdown?

By the way, don't forget to add the back-to-back diodes from the LT3083 SET to OUT pins or alternatively series 8V2 zeners. Under transient conditions (like output short circuit), the SET pin in the LT3083 must never differ from OUT by more then 10V.  If you use the LT3080, you need back to back schottky diodes instead as the LT3080 has less protection on the SET pin.

Richard.