LVC - Low Voltage Cutoff

[curious]

safe-
If you are talking about common mode voltage range requirement of the internal comparator then it slightly over Vref (typically 1.4V for these parts). Otherwise the part is specified up to 7V (absolute max rating) and common voltage range concept simply *does not* apply to it. A similar part in Bob/Gary LVC is used per cell and is *not referenced* to the battery pack ground, so there is no common mode voltage issues. The binary on-off signal is then translated from the local cell reference to the ground with an optocoupler.

 

[safe]

What is the CMRR?

http://en.wikipedia.org/wiki/Op-amp

Common mode gain — A perfect operational amplifier amplifies only the voltage difference between its two inputs, completely rejecting all voltages that are common to both. However, the differential input stage of an operational amplifier is never perfect, leading to the amplification of these identical voltages to some degree. The standard measure of this defect is called the common-mode rejection ratio (denoted, CMRR).

http://en.wikipedia.org/wiki/Common-mode_rejection_ratio

The common-mode rejection ratio (CMRR) of a differential amplifier (or other device) measures the tendency of the device to reject input signals common to both input leads. A high CMRR is important in applications where the signal of interest is represented by a small voltage fluctuation superimposed on a (possibly large) voltage offset, or when relevant information is contained in the voltage difference between two signals.

From the pdf for a "High Common-Mode Voltage DIFFERENCE AMPLIFIER" suitable for battery cell voltage monitoring (INA117):

The INA117 is a precision unity-gain difference amplifier with very high common-mode input voltage range. It is a single monolithic IC consisting of a precision op amp and integrated thin-film resistor network. It can accurately measure small differential voltages in the presence of common-mode signals up to ±200V. The INA117 inputs are protected from momentary common-mode or differential overloads up to ±500V.

So it appears that some comparators are designed specifically to counter the negative effects of having high common mode voltages. In the case of the INA117 they are built to withstand a full 500 (peak) volts. It's possible that the lower priced comparators can be used up to 36 volts or maybe even higher without problems, but.... I'm WONDERING if there are some things to be concerned about or is this only a very high voltage problem. 500 DC volts (peak) is a lot of power and 36 volts might be just fine for Voltage Detectors and weaker comparators. It's because I don't know WHY there is a difference in the "story" coming from different sources that I'd like to get some sort of authoritative knowledge about it.

Maybe 200 volt protection is just overkill? (it sure sounds like it)

This might be important information for people who have ideas about really high powered systems. What might work fine at 36 volts might not work okay at 100 volts or above.

 

[OneEye]

I'm not a EE, so my take may be just as wild as yours; regardless, if you look at the INA117 functional diagram, the V+ and V- are different than the voltage references being measured, so that is why there is some concern about comparing 200+ volt inputs. On the BMS designed by Bob McCree, the chip is being powered by the same voltage source it is measuring, so the chip will only ever see a voltage difference of a single cell, no matter where in the battery string it is located, or how high the voltage is when referenced to ground. The common mode rejection ratio is simply not applicable in this case. In order to screw up this one, you would need to exceed the optoisolator's ability to isolate voltages on either side of the led/photoresist pair in the package, which if I understand correctly is a phenomenally high voltage.

Curious points out the exact same thing in the post just prior to your latest.

 

[safe]

Curious points out the exact same thing in the post just prior to your latest.

Yeah, you guys both seem to not be understanding what "common mode voltage" is about.

The idea is that even though you are trying to measure a small relative difference in voltage (from one cell to the next) when the common mode voltage is very high it apparently distorts the behavior of the actual material that is in the comparator no matter what the local difference is. (the word "saturation" will pop out if you read about it) The high voltage threshold might be above where we are working (like maybe above 100 volts) so the actual problems might not crop up in the area we are working within.

"Common Mode Voltage" means the "baseline" voltage of the system and so you are not talking about the full gap of ground to top voltage. You are not talking about the 0-200 volt difference... but you ARE talking about how a baseline of 200 volts with a slight voltage difference on top of it will behave.

You got that?

That's how I understand it anyway. (maybe I'm wrong)

However, like I already said, the problem might not become serious when you are only in the 36 volt range. But it would be nice if someone with really expert level knowledge could comment.

 

[safe]

Product Review of the INA117 Difference Amplifier

First I just want to say that their SPICE model produces results in simulation that matches their product claims. Whether this is also going to match the real world I don't know. But at least "on paper" everything matches correctly.

The central topic is "common mode voltage" and how it might distort the results. I've found that from 15 volts to below their rated 200 volts the output is just about perfectly matching the expected. The oscillation is set up to run between 2.5 volts and 4.25 volts and that's pretty much what you see as output.

Above 200 volts things begin to fall apart (as the spec claims it should) and the output starts to get messed up and eventually the signal just gets saturated (off the chart) at about 6 volts.

Don't forget that this is a specialized chip (INA117) that was purposely built to deal with high common mode voltages. The distortion you should get with other lessor chips will be radically worse and so you ought to at least run a SPICE simulation on them to see if you are going to get what you think you are getting.

 

[curious]

Yeah, you guys both seem to not be understanding what "common mode voltage" is about.

OMG . One last attempt before I give up. Your circuit above does indeed have to deal with common mode voltage issues - both rejection ratio and absolute limits because your comparator generates output voltage referenced to ground and you power all the devices from the same V+/V-. This is why you need a special comparator that can handle the common mode voltage of entire pack.
In a Bob/Gary circuit the common mode voltage problem is simply not relevant to the voltage detector IC. Each voltage detector is referenced to it's own cell negative terminal and powered only by this cell. The detector IC is not aware where the system ground is at all. It is solely the role of the opto to re-reference the binary output signal to ground so it is the *opto* that deals with common voltage and not the voltage detector IC itself.

 

[OneEye]

Curious points out the exact same thing in the post just prior to your latest.

Yeah, you guys both seem to not be understanding what "common mode voltage" is about.

"Common Mode Voltage" means the "baseline" voltage of the system and so you are not talking about the full gap of ground to top voltage. You are not talking about the 0-200 volt difference... but you ARE talking about how a baseline of 200 volts with a slight voltage difference on top of it will behave.

You got that?

That's how I understand it anyway. (maybe I'm wrong)

However, like I already said, the problem might not become serious when you are only in the 36 volt range. But it would be nice if someone with really expert level knowledge could comment.

I think I see where the confusion is. By "baseline" voltage, you actually mean the V- end of the power supply to the part, not the ground reference of the entire "system". The part does not know, nor does it care what the ground reference of the entire system is, only what it is connected to. Voltage is always relative to something. Most of the time we reference "ground" or "earth" as 0V, but there is no universal law that requires you to tie in to a ground reference. The common mode rejection ratio you are discussing is when the part doing the measuring is tied to a voltage reference (the V- on the block diagram), and the voltages being measured are extremely high (200V+) relative to that reference. In Bob/Gary's design, the part is not tied to ground, nor is it tied to the negative terminal of the string of cells, V- is floating at the voltage of the single cell's (-) end, and the measured voltage is on the (+) side of that same cell. No matter where in the chain the part winds up, whether 1V relative to ground, or 10kV relative to ground, the only difference it sees between the voltage to measure and it's V- will always be less than 5 volts. If you exceed a 5v difference between each side of the cell you already fried it, so an ability to compare slight voltage differences up to 200V above V- is irrelevant to Bob/Gary's design. If you are going to use a part to compare voltages but tie the V- to the negative terminal of the string, you will need to contend with the common mode rejection ratio problem you have identified if you create a high voltage series of cells--which is why Bob lets the part float relative to ground and uses an optoisolator to give the undervoltage signal. Essentially the comparitor part of the package doesn't know that it is sitting sky high, it only sees that one side is less than 5V above the other. That is the beauty of optoisolators. They let you use one side of the part at one voltage potential relative to ground, and the other side of the part at a different voltage potential relative to ground.

 

[GGoodrum]

I am absolutely amazed about how far off on a tangent you will go. The TC54 chips only have to compare the voltage from one cell, against a calibrated reference that is generated within the chip. They don't know, or care about any other cells/TC54 chips. Each one works totally independent. Why can't you see this?

I've just been testing the new BMS board, which has the LVC circuits on it as well, and the TC54 stays high, even if the voltage is exactly at 2.100V. As soon as it hits 2.009V, the output goes low. It will stay low until the voltage is raised to 2.115V, and then it goes high again. All 16 behave exactly the same way, every time.

All this nonsense you keep bringing up is nothing more than a total waste of bandwidth.

-- Gary

 

[OneEye]

Note the line directly from V- to ground in your INA117 Diagram.gif (16.61 KB). That is why you have the CMRR problem. Tie V- to the cell being measured and this should clear up.

 

[safe]

Look Again :wink:

These INA117's as published in the company pdf (not mine which might have errors) CLEARLY NO DO NOT attach themselves to the same ground as the power. They are all relatively referenced. (just like a Voltage Detection circuit)

I'm right on this one... (hard to imagine huh? :lol: ) as long as the voltage is low enough and your comparators are well built they will work fine. But if you get up into the higher voltages (like 200 volts) the "common mode voltage" causes a saturation in the comparator. This saturation is caused by the high voltage and not the relative voltage. At no time is the 200 volt level compared to 0 volts... it just doesn't happen.

So in the end this might be a "non-issue" at low voltage, but it very well might start to show up at 72 volts. Some people might care about 72 volt systems and knowing if the distortion begins by then would be important to some people.

This could be resolved with simple tests, which you seem to have already done. Has anyone tested this on a 72 volt pack? A 16 cell pack is right around 48 volts...

 

[safe]

Note the line directly from V- to ground in your INA117 Diagram.gif (16.61 KB). That is why you have the CMRR problem. Tie V- to the cell being measured and this should clear up.

I made the disconnect and it has no effect. The same results exist at over 200 volts.

Let me add also that when the threshold is reached the comparator becomes saturated and the results RAPIDLY go bad. So you want to be well underneath the threshold, wherever it happens to be. (something that would be good to know)

 

[OneEye]

You are correct that I misstated the reference being V- on the company schematic, I failed to notice pin 1 5 and 8 on the company schematic are where the tie to ground occurs. This ground is the same ground reference being used to specify the +200V limit on the company schematic. Otherwise +200V max doesn't mean anything. Note how in the company schematic pins 2 & 1 and pins 3 & 5 form a voltage divider circuit with reference to ground.

That IC needs to be tied to ground because the signal line on pin 6 is not isolated from the rest of the circuit. Bob/Gary's signal pins are isolated from the input side.

When you adjusted your simulation you left V- tied directly to ground. All ground references in spice are the same ground reference. Put a 214.4V cell between V- and ground, make sure the V+ is still within spec relative to V- and let us know what happens.

 

[safe]

Put a 214.4V cell between V- and ground, make sure the V+ is still within spec relative to V- and let us know what happens.

It makes the output rise in proportion to the Elevated Ground reference. (assuming I made the test you wanted :wink: ) So that doesn't seem a valid avenue of inquiry... you are supposed to use ground to make the comparator work. The comparator always attempts to balance it's inputs (low and high) and at lower voltage it seems to do it pretty well. At high voltage the internal functioning of the comparator seems to malfunction.

There's a reason that Wikipedia has a whole section on this stuff... it's likely a real thing. Apparently if you stay low enough in the voltage you will have no problems.

The Nightmare Scenario

The Nightmare Scenario would be if you had two 36 volt packs that you decide to assemble in series. If the comparator inside the Voltage Detection circuit actually hits it's threshold at (say) 60 volts then it's possible that the last couple of cells in the upper voltage pack would malfunction in use.

By the same token if the threshold is ABOVE 72 volts (which it very well might be) then you never know that a potential problem might ever exist. Ignorance would be bliss.

So we just won't know for sure without an accurate SPICE model of the Voltage Detector being tested on a circuit with some higher voltage settings.

SPICE Models anyone? (for the Voltage Detector)

If you know of one just post it here as text... don't force me to hunt it down somewhere.

 

[OneEye]

Perfect. The output was supposed to climb with the elevated ground. An optoisolator is then use to take this elevated signal and translate it to a different ground reference.

You should be able to maintain stability as you ramp up the elevated ground reference along with the first cell in your series together to high values well above the +200V threshold. The output signal will increase with the increase in the "elevated" ground. It is the optoisolator in Bob/Gary's package that "translates" the signal between the "elevated" ground reference and a real ground reference (or in their case the negative end of the battery pack). This is one of the beauties of their design--the comparitor / opamp can operate at whatever voltage (even higher than +200V) and then the optoisolator translates that down to the level you want to use to signal a cell has dropped below its correct voltage. You no longer have to worry about CMRR, you just have to pay attention to the isolation limits of the opto.

 

[OneEye]

Oh, don't forget to remove the ground you put in on the end of the 10K resistor you are calling "true voltage". That should be tied to the elevated ground as well.

ElevatedGround should equal Cell1 + Cell2 + Cell3

The (-) of Source should be tied to the (+) of ElevatedGround

 

[OneEye]

Analysis of Bob Mcree's Circuit

<SNIP...>

Part Two:
The second part uses an opto device that reads the current flow in the first voltage comparison loop and creates an actual output signal.

My GUESS as to why you do this is to prevent leakage.

Is this correct?

Is the main reason for setting things up this way (with the opto half of the circuit) is so that you can cut leakage to zero?[/color]

It makes the output rise in proportion to the Elevated Ground reference. (assuming I made the test you wanted :wink: ) So that doesn't seem a valid avenue of inquiry... you are supposed to use ground to make the comparator work. The comparator always attempts to balance it's inputs (low and high) and at lower voltage it seems to do it pretty well. At high voltage the internal functioning of the comparator seems to malfunction.

The answer to your question in the first quote is the solution for the problem in the second quote. The optoisolator translates between electrical systems that use different ground references. So a system on one side of the isolator can operate at an elevated ground, and the system on the other side can operate at a real-world ground reference.

 

[safe]

The optoisolator translates between electrical systems that use different ground references. So a system on one side of the isolator can operate at an elevated ground, and the system on the other side can operate at a real-world ground reference.

Okay, that seems to make sense and resolve the issue about high common mode voltage problems. The brute force way to deal with it is to just buy more expensive comparators that can handle the full intensity of the higher voltage. With the brute force approach you also get an actual voltage result which could be fed to a display somewhere. Using the Voltage Detector combined with a Optocoupler allows the opto's to all ground themselves together. When what amounts to a signal is passes through the LED it jumps out of the reference frame for ground and avoids all the issues for that. So it does make sense... you really need all the parts (Voltage Detector and Optocoupler) to make the overall system function, in other words you can't use just the Voltage Detector and then try to read that voltage because that forces you back to ground... which defeats the whole point of the design.

Okay... I get it... a very slick solution. 8)

So I'm sufficiently convinced that it can scale up without need for testing.