Datalogging Vbat....editting a DBF file

[92sho16]

Since the tweecer cannot datalog vbat accurately Ive been trying to datalog through my dataq. I have cut the voltage down with a simple two 10k ohm voltage divider and know need to undo that with software. I cannot seem to do this part successfully, has anybody? There are no predefined settings but you can edit the dbf files for any of the sensor manufacturer but when i edit them tweecer does not like this.

 

[92sho16]

nobody has done this? I know yahamaSHO maybe he'll chime in.

 

[sho_sc]

nobody has done this? I know yahamaSHO maybe he'll chime in.

I haven't done it ... but would like to....

What's the specs on the voltage divider?

 

[92sho16]

I haven't done it ... but would like to....

What's the specs on the voltage divider?

This is what i did. I just can't double the incoming value in tweecer.

 

[Lupo]

Since the tweecer cannot datalog vbat accurately Ive been trying to datalog through my dataq. I have cut the voltage down with a simple two 10k ohm voltage divider and know need to undo that with software. I cannot seem to do this part successfully, has anybody? There are no predefined settings but you can edit the dbf files for any of the sensor manufacturer but when i edit them tweecer does not like this.

How is cutting the voltage going to help the tweecer log more accurately? Is 12-14v too high, but it logs lower voltages correctly?

 

[gmorrell]

He isn't datalogging Vbatt with the TwEECer, he's datalogging Vbatt with his DataQ DI-148 (I suspect...), and the maximum input voltage is +/-10 Volts on any analog input channel.

Based on what I know about the DI-148 input impedance and it's input divider, you could also just put 100K in series between Vbatt and a DI-148 analog input, this will keep the input from going over 10V, but you'll need to input a known voltage close to 12V and measure it once with a good DVM so you'll know what that voltage reads back as in the DataQ.

 

[92sho16]

He isn't datalogging Vbatt with the TwEECer, he's datalogging Vbatt with his DataQ DI-148 (I suspect...), and the maximum input voltage is +/-10 Volts on any analog input channel.

Based on what I know about the DI-148 input impedance and it's input divider, you could also just put 100K in series between Vbatt and a DI-148 analog input, this will keep the input from going over 10V, but you'll need to input a known voltage close to 12V and measure it once with a good DVM so you'll know what that voltage reads back as in the DataQ.

This is the exact scenario.I using a supply right from the underhood fuse block. My problem is that i cannot get the value double to reproduce what it was before the divider, just doubling it probably wont get the exact value it was but it will be a **** of alot better than the tweecer's value of 4-8v typically. If i cannot get it doubled in software im thinking about getting a different dataq.

 

[gmorrell]

This is the exact scenario.I using a supply right from the underhood fuse block. My problem is that i cannot get the value double to reproduce what it was before the divider, just doubling it probably wont get the exact value it was but it will be a **** of alot better than the tweecer's value of 4-8v typically. If i cannot get it doubled in software im thinking about getting a different dataq.

You don't necessarily need to double or normalize the DataQ output, you can post-process the data in Excel after you collect it.

If you do what I suggested, which is inputting about 12V into the divided DataQ channel, and measure that voltage with a good DVM, then you know that X Volts (DVM measured, into the divider) is equal to the voltage you read from the DataQ. Once you know that, and if you want to get **** about it, ground the input and measure the zero offset, it's just good old Y = mX + B Basic slope-intercept math.

You can drop the DataQ output into Excel, apply the scaling factor, and get it normalized back to the actual voltage before the divider.

 

[Lupo]

You don't necessarily need to double or normalize the DataQ output.

If you do what I suggested, which is inputting about 12V into the divided dataQ channel, and measure that voltage with a good DVM, then you know that X Volts (DVM measured) is equal to the voltage you read from the DataQ. Once you know that, and if you want to get **** about it, ground the input and measure the zero offset, it's just good old Y = mX + B

You can drop the DataQ output into Excel, apply the scaling factor, and get it normalized back to the actual voltage before the divider.

Assuming is scales at a constant, correct?

 

[gmorrell]

Assuming is scales at a constant, correct?

Correct, the system linearity is assumed to be a constant, assuming that DataQ put a decent analog to digital converter in their box, which they did.

All we're doing here is re-scaling the input of a fixed input range A to D converter. If you know how much it's been scaled, and the system is linear, which it is for DC, it's just math.

Trust me, I do high-speed, high-accuracy (up to 20 bits) data acquisition for a living, and I make a very good living at it.

 

[92sho16]

You don't necessarily need to double or normalize the DataQ output, you can post-process the data in Excel after you collect it.

If you do what I suggested, which is inputting about 12V into the divided DataQ channel, and measure that voltage with a good DVM, then you know that X Volts (DVM measured, into the divider) is equal to the voltage you read from the DataQ. Once you know that, and if you want to get **** about it, ground the input and measure the zero offset, it's just good old Y = mX + B Basic slope-intercept math.

You can drop the DataQ output into Excel, apply the scaling factor, and get it normalized back to the actual voltage before the divider.

I see what your saying. So its possible then to process vbat in excel then put it back into the log?

 

[gmorrell]

Simplest of examples.

Let's say you have an analog to digital converter with a range from 0 to +5 volts, but you want to measure up to +15 volts. What do you do?

First you build a voltage divider from two resistors, we want the divider to divide the incoming voltage by 3, so we'll use 2000 ohms and 1000 ohms. If we put these in series with the 2000 ohm connected to 15V and the 1000 ohm connected to ground, we would measure 5V across the 1K resistor, so we connect this point to the input of our A to D.

Now if we put 15V into our divider, our A to D sees 5V it tells us it's at full scale. This is one of the points we need to define a straight line.

We get the other point by grounding the input to the divider, and if all is working properly, the A to D should tell us it's reading 0V. It's safe to assume the system is linear for DC voltages, so we now know that whatever the A to D tells us, the voltage at the input to our divider is 3X the A to D reading.

The above is an ideal example, there are some complications using the DataQ DI-148 because the input already has a divider with a weird-ish divide ratio, and it has some impedance that will load your added divider. If you want some off-line help, PM me.

The divider circuit you proposed a few posts up will work just fine as a divide by 2, however, the loading effects of the DI-148's input will introduce about a -2.2% error. If you want to reduce that error to -0.2%, change the divider resistors to 1K and 1K.

 

[gmorrell]

I see what your saying. So its possible then to process vbat in excel then put it back into the log?

I can't answer that as I don't fully know what you're up to, but you can get the voltage measurement from the DataQ, you can do a simple off-line mathematical manipulation to get the correct voltage, and then perhaps put it back into your data file, assuming it's something simple like comma delimited text or csv. Excel supports formulas in cells, it's easy to manipulate data. Lord, I do quarter-million point FFT's and non-linear curve fits in Excel.

I've had this discussion with others outside of SHOForum, and I can't understand why somebody doesn't just fix the damn TwEECer RT so it datalogs Vbatt. Whose product is this anyway?