Accuracy of division using MPY634

Hi all,

I'm doing a simple project which will do the voltage division, Vo = 10(V1/V2). I do this using MPY634.

My connections are as below:

_____________________ V2 --- | X1(1) (14)+Vs|--- +15V GND--- | X2(2) (12)Out| ----------- |SF(4) (11)Z1 |--- GND | GND--- | Y1(6) (10)Z2 |--- V1 | __|Y2(7) (8)-Vs |--- -15V | | --------------------- | |_________________________________|

The output that i got will be very inaccurate. For eg, when i pump in V1=0.406V and V2=0.008V, the answer =0.0433V. The answer should be

My circuit's connections are based on the data sheet of MPY634 at page

Does anybody know why this inaccuracy happen? Please advise.

Thanks, Albert

Hi, one.

Your results are consistent with the input offset voltage specs for that part. The larger reason is that analog circuits are always inaccurate to some degree. It is up to the designer to make sure that the accuracy is appropriate for the application.

You may want to add a trimmer or two, or determine what accuracy you really need and maybe find another technique. Be sure to consider temperature (in)stability before deciding that trimmers will do the job.

You're welcome.

Thanks, Larry.

What do you mean by "consistent with the input offset voltage specs" and "add a trimmer"? Can you explain by giving an example?

So, my connections are correct right?

sorry if my questions are stupid.

You are welcome.

Go to the spec sheet and find the guarantees placed upon input offset voltage by the part manufacturer. Measure some actual (Xin, Yin, Vout) triplets. You will find that there is a pair (Xio, Yio) such that for each triplet, Xin+Xio times Yin+Yio is very close to Vout, (much closer than the errors you posted), and both Xio and Yio are within the input offset specification. You can think of the input offset as a secret voltage that is added to each input with no extra parts or effort. Your trimmer would be an explicit circuit to add a seperate constant voltage to each inputs. That can be easily done (for your circuit) by putting a high value pot between the supply rails, dividing the wiper with a resistor divider that reduces voltage by a factor of 500 relative to ground, and using the result in place of the grounded side of each differential input. (See figure 2 in the spec for an almost usable example. Just don't ground the divider output as appears in the diagram.)

I did not check them, but your results were close enough to right that I had no reason to doubt the connection. The input offset issue is adequate to explain what you see and the probability of getting there with wrong connections is very low. And since there is a simple divider hookup in the datasheet, and your schematic resembles it closely, I figured you had not botched it.

They are not stupid. They are evidence of some learning opportunities.

The classic answer to this is, "The only stupid question is the one you don't ask," but recently I saw a beer commercial that started out looking like a laxative or hardon pill ad, but about "Taste Loss". One of the characters asked something stupid, and the voice-over says, "There are no stupid questions when it comes to Taste Loss..." and the other character says, "Can you get Taste Loss from a Foosball table?" And the voice-over says, "Let me rephrase that..."

Cheers! Rich

I beg to differ, since it looks like in Albert's circuit, assuming Y1 and Y2 are a differential input, that Y1 is equal to -Out, so there's some kind of a feedback term you're neglecting, which I'm way too lazy to get into any deeper here. And don't you have Z2 and Z1 backwards?

Thanks, Rich

I guess I'll do some diligence here, overlooking whether it is due or not.

The OP's schematic matches that shown in the datasheet he cited over the caption "FIGURE 4. Basic Divider Connection". I had noted before in this thread that his "schematic resembles it closely", so I was surprised to see your allegation of an incorrect hookup.

Getting to the meat of your critic, (and ignoring the hookup issue):

The function claimed in the spec sheet, after simplifying for the absence of the optional summing input, is: Vout = 10V * (Z2 - Z1) / (X1 - X2) The OP's connection substitutes V2 for (X1-X2) and V1 for (Z2-Z1), so the expectable result is Vout = 10V * V1 / V2. Using his numbers, (V1=0.406V, V2=0.008V, gotVout=0.0433V, hopedVout=0.019V), he hoped for 0.019V = 10V * 0.406 / 0.008 == 507.5, as you claim. So, it appears that either the part manufacturer's datasheet is somewhat fanciful, or the OP's expectation is wrong, or the hookup is wrong. Having excluded the last, and much doubting the first, the middle way seems worth a look.

If we assume the OP got his input numbers swapped and forgot to apply the 10V scale factor, we get 0.197V = 10V * 0.008 / 0.406 . His expected 0.019 V figure is close to the improperly scaled result. And as I was saying, (having only looked at ratios without actually evaluating the above related mappings), simply assuming some input offset within the spec would explain the remaining discrepency.

At this point, I think we need the OP's further input to resolve this.

....

If you were paying attention, you would see that I never went away. So, no, I am not "back".

No, because, as I stated, his schematic already looked right to me. The topology matched the diagram, no inputs were left floating, and the output was not saturated. So my surprise was based on information, not ignorance. Why do you need to assume ignorance when there are much simpler and more evident explanations?

Gawd (sic), you can be obnoxious at times.

But you are quite the speller, I'll give you that.

I excluded it because a close study of the OP's schematic and the datasheet led me to believe his schematic was good. And the nonsaturated output led me to believe the inputs and power were all actually connected to levels not far from what correct wiring would have provided. (If you doubt this, I can explain it in more detail.)

How could I not have thought of the very possibility that I listed? If that is intended as an insult, you need to brush up on that skill. Otherwise, it looks much more like hastily composed nonsense.

The following quote was my effort to make sense of what the OP wrote. If you cannot make sense of it, I am not inclined to improve upon it for you.

Do you see me claiming "This is the one and only correct explanation for the OP's results."? I offered it as one hypothesis. That is why I invited the OP to clarify.

To me, it looks like you are the one with some unpleasant load to dump. It seems to occur almost daily, whether I am the dumpee or not.

Here, you engage in projection. You imagine my motive based only on why you might act as you incorrectly perceive I have acted. My mistake, if that's what it was, lies simply in not applying the effort earlier. Perhaps your notion of "due dilligence" makes that a mistake.

Fine. It still could go either way. However, if the OP resolves this my way, I expect you will show the intellectual honesty to retract that comment.

Generous of you.

There is some technical content here, about 100 lines down, where John's "flight of fancy" is controverted.

"John Fields" wrote in message news: snipped-for-privacy@4ax.com...

I fear we are miscommunicating here. Your "simplest" is little more than "whatever comes most readily to mind", motivated by a need to feel superior. As we shall see later, there is a simpler explanation in the usual "Law of Parsimony" sense.

As for the "more evident" part, I am forced to admit error on that point since it is so dependent upon the observer's perspicacity.

Oh! Now I get it! You mean to say that not every hand built circuit conforms to the intended schematic at first? How could I have missed this obvious possibility after building engineering prototypes for years? How wonderful to find a place where one's betters are so willing to educate! I can hardly express my appreciation adequately.

Let me try that again. You have earned my barely finite admiration as someone capable of discovering spelling and grammar errors.

Having a hard time with authority, I must disobey. Even if your education is complete, I am inclined to make what I thought to be obvious enough more so that you might get a glimmer of how mistaken you have been since you elected to dump here.

From the datasheet, (eliminating the unused summing input), the divider function is: Vout = 10V * (Z2 - Z1) / (X1 - X2) Substituting the signals from the OP's circuit: Vout = 10V * V1 / V2 Now, rewriting to incorporate the input offset that exists for each of the two inputs: Vout = 10V * (V1 + Ezios) / (V2 + Exios) Substituting claimed inputs and observed output, but swapping the inputs (as discussed elsewhere): 0.0433V = 10V * (0.008V + Ezios) / (0.406V + Exios) Solving for Ezios as a function of Exios: Ezios = (0.0433 * Exios - 0.0624202) / 10 (If you want to see more algebra, just say so. I am torn between your "bellyfull" lament and your complaint about seeing too little math.) Now, (to make this easier), suppose Exios = 0, a value well within the typical +/- 25mV spec of the low-grade part or the maximum +/- 20mV to 30mV spec of the mid-grade part. This Exios value then yields Ezios = -6.24202mV, which is also well within the input offset spec. There are numerous other pairs of Exios,Ezios values that could also result in the observed Vout.

It was by means of my initial calculation, very similar to the above, that I concluded and told the OP: "Your results are consistent with the input offset voltage specs for that part."

I will be interested to see if you insist upon continuing to label this hypothesis my "'offset error' flight of fancy" or will be willing to acknowledge that it merits less denigration. (My bet is placed; please do not disappoint me.)

That's clever, I guess, John. The list item at issue was: "or the hookup is wrong". After you suggested I did not think of "it", (referring to that list item), I claimed that I had to think of it to list it. Now, you are trying to change "it"'s referrant from "the hookup is wrong" to "the signal inputs could have been interchanged".

What point are you trying to make here? I guess it's that it is news to you that I did something I have neither done nor claimed to have done. I have more news: It would be news to me as well.

My first post on this thread reflected my belief that no wiring error was *necessary* to explain what the OP was seeing. Contrary to what dolts might conjecture, I have long been familiar with the possiblity of construction errors. [1] My position that something other than wiring error can explain the OP's observations should not be construed to mean that I deny the possiblity of wiring error.

Your persistence in misconstruing my position as "mine is the only explanation" requires you to ignore my clear statements such as "At this point, I think we need the OP's further input to resolve this.", "It still could go either way.", and "I offered it as one hypothesis."

Why do you persist, in the face of all my words to the contrary, at leveling your scurrilous allegation? It really makes me wonder if your motives are so pure and admirable as you would prefer them to be seen.

In other words, you neglected a hypothesis. As truly competent folks can see, it was/is in fact a viable hypothesis. Your notion of "having to resort" to a hypothesis is laughable. Unless a hypothesis is impossible or too improbable to bother considering further, there is no reason to treat it as something one must "resort to".

Your quote is misleading. I never wrote "some offset in the spec" and your apparent attribution of such silliness to me is either dishonest or a very sloppy mistake for someone so familiar with spelling and grammar.

There are some intriguing inferences to be made from your misquote. If you intentionally converted my repeated mention of "input offset voltage" to "some offset in the spec", it means either that you are disgustingly dishonest or simply do not comprehend the importance of an input offset in the 10's of mV when an input of 8 mV is being presented. If you had actually looked up the input offset to check my assertion, I doubt you could honestly attempt to slight my attention to input offset in that way. So, I am led to conclude that you never did look at the input offset spec since I am not convinced that you are as dishonest as the alternative indicates. That failure strikes me as exceedingly arrogant and moderately comical in light of your incessant yammering about "due diligence".

Maybe you can explain what really happened if my above analysis omits more favorable possibilities.

That is very noble of you, in a way. You come along and cannot understand something before it is explained in minute detail, so you presume it must be bullshit and spend more effort upon your dumping than you put toward actually understanding anything.

You are way too sure that you are right as compared to your demonstrated talent. I suspect this is why you exhibit hateful reactions when you perceive somebody as putting on airs, reminding you of a trait you dislike and would disown in yourself.

But maybe I am mistaken.

Well, see the above starting at "From the datasheet". Funny how your willingness to let me explain the role of input offset has changed so soon from "Spare me, I've already had a bellyfull". What happened?

You've got me confused here. I do not know what "magic trick" you are going on about. What I have come up with is my own hypothesis as to how the OP came to make his observations and why the output he saw differed from what he expected, (not to mention differing from anything your ever predicted). I cannot imagine who you are suggesting handed that idea to me. It certainly was not you.

Perhaps you have confused my mention of swapped inputs for an attempt to "take credit" for "your" idea. What you seem to not yet grasp is that my earliest calculation, done before you ever piped in, simply assumed the input arrangement that would be needed to get what the OP saw output rather than some 500V (which would surely have saturated the output, contrary to what was observed). I did not even notice that the OP's description had swapped (or may have swapped) those inputs until after you had posted.

If that is the "magic trick" you accuse me of "pretending [I] would have come up with", I challenge you to quote my words to that effect. I have never thought to attempt such a pretense, so I doubt you can justify the notion that I did so.

I confess to laying it on a little thickly there. The fact is I do not expect intellectual honesty. Thank goodness for your diligent scrutiny.

I was jesting! I earned it! Wah!

"fluff" == "Cannot intelligently refute".

If your inane dumps are "heat", I am not worrying, believe me.

Thanks, buddy. You have acted true to form, allowing me to win that bet I mentioned.

Now that's pretentious.

--- Hardly. What you mean is that you've posted some fluff designed to extricate you from the position you've gotten yourself into and are determined to try to change the focus of the argument in order to "get the heat" off of yourself.

Good luck, but you'll have to carry on without me, since I have better things to do with my life than proving over and over again that you're full of s*it. You're free to go.

-- John Fields