Re-calibrating Xantrex Link 1000

[JDM2007]

I'm looking for the low-level calibration procedure for a Xantrex Link 1000.

By "calibration" I mean adjustments not usually performed by an installer/user, and not in the manual. I couldn't find them at xantrex.com or elsewhere.

This is for a unit that I have repaired for educational / self-reliance / entertainment purposes... (I'm sure sending a unit back to Xantrex is more cost-effective in most cases).

I've got as far as figuring out how to get into calibration mode and set the current calibration for the high and low ranges, but it's not yet clear to me what the optimal inputs are for greatest accuracy. Voltage calibration info would be helpful too.

If the calibration procedure is considered proprietary, I would be happy to keep it in confidence. Otherwise, anyone else that is interested should let me know if I should post/email what I figure out myself.

Thanks,

James

[Rick]

There have been several times when I've done a current "offset" recalibration in the field, especially after Xantrex took over the production from Cruising Equipment and Heart Interface. Because there are some differences in the procedures amongst the various monitors I will have to consult my notes before outlining it here otherwise I might mislead you into confusion.

The only real caveat to making a valid calibration in the field is to ensure that no load current or ground-loop current flows through the shunt AND that the shunt is correctly wired as instructed.

The various Link and E-meter battery monitors have been designed to have current stability over a wide temperature range so that when a true "zero" current passes through the shunt the internal electronics vary about 1/3 of a least count so that not much "racking" occurs in the display. Keeping in mind that even with a so-called perfect stability an A/D conversion will result in a statistical racking back and forth between a "plus zero" and a "minus zero" corresponding to one count of error with each racking of value. When the instrument is calibrated correctly the average racking will result in a slight net "positive zero" value.

I'll have to get back to you later when I get the detailed procedure, O.K?
Regards,
Rick

[JDM2007]

Originally Posted by Rick

I'll have to get back to you later...

Thanks, whatever info you can provide will be greatly appreciated (and I'm not in a big hurry, so at your convenience).

I understand that +/- one lsb is the best one can achieve.

FWIW, I'm doing this calibration on an electronics bench (rather than installed). So, for example, I'm powering the Link from a separate supply and applying millivolts directly rather than measuring current through a distant shunt. I can also probe the ADC inputs on the MPU if that makes it easier to attain high precision.

I'm quite impressed with the design and features, and wouldn't hesitate to recommend this unit. (I have 3 myself.)

Thanks for your assistance,

James

[Rick]

Probing the ADC input is likely to negatively affect the calibration. At the least it offers opportunity for noise injection. Probing is O.K. for measuring and troubleshooting but I don't recommend that during a cal.

There are only a few nanovolt sources which exhibit a sufficiently low source impedance to use when calibrating the meters. In addition, when I used to review the tech bench setups for ground loops I would often find some wire added for convenience which would affect the calibration negatively. A nanovolt source impedance needs to be on the order of that of the shunt when simulating current measurments. The shunt conductance is 10k Siemens.

For a relatively accurate zero calibration the conductance does not have to be that low unless in a noisy environment. Otherwise it should have a conductance of at least 100 Siemens

[JDM2007]

Originally Posted by Rick

There are only a few nanovolt sources which exhibit a sufficiently low source impedance to use when calibrating the meters.

The shunt conductance is 10k Siemens.

For a relatively accurate zero calibration the conductance does not have to be that low unless in a noisy environment. Otherwise it should have a conductance of at least 100 Siemens

Ok, that's helpful.

To obtain the millivolt input, I was putting current through a 10 milliohm shunt resistor (so 5A would simulate the 50 mV resulting from 500A through the typical shunt), but I have a 1 milliohm resistor I can use instead (at least for moderate currents, I'm not eager to put 50A through it unless I need to).

The discussion of noise has got me thinking about time constants... do you know offhand how the Link processes the measurements during normal operation? I'm supposing it averages some number of conversions over some time period, and uses that for display/calculations.

Thanks,

James

[Rick]

Link1000 Current Calibration Procedure

• Hold SEL & SET simultaneously before and during power up.

Notice that the unit indicates a calibration mode by alternating the “V” and “A” indicators. The display alternates between BAT1 Volts and BAT2 Volts. Consult the manual for setting the proper mode to indicate and auxiliary battery voltage when the input is wired for that purpose. The display will not change unless the two inputs differ in voltage as the two indicators alternate.

• Press SEL to start the calibration when ready. The first calibration is for the low current range. Use INV to change offset + and press CHG to change – until a “plus” zero indication persists. The range of offset is +/- 6 counts.
• Press SET to move on to the high current range calibration and repeat the process.
• Press SET again to end the calibration. The display alternates between V1 and V2 indicated by “V” and “A” function LEDs.
• Push SET again to enter the display test.
• Push SET to exit the display test and calibration. In order to repeat any test or calibration the entire process must be repeated.

In addition to passive anti-aliasing filters there are between 300 and 1000 running averages made along with a 1 second (approx.) display update.

[JDM2007]

Originally Posted by Rick

... until a “plus” zero indication persists.

Ok, good... I can make some sense of that.

However, on my Link 1000, there is a further condition that seems to be required to get into current calibration mode, is it ok to post it (perhaps it was omitted on purpose)?

For this calibration for zero current, I presume the ideal input is shorting the two current sense inputs together, then to ground (and all close to the Link), is that correct, or should the connection between current sense and ground be through some resistance?

I apologize for being a bit dense, but I suspect I'm still missing a step (or a few):

I would have expected that there would also be a (nearly) full scale calibration for each of the high and low current ranges, perhaps for each of + and - current inputs.

1) If not, do you suppose the procedure is to adjust (if necessary after a repair) the full scale range via changes to the op-amp gain setting resistors?

2) In addition, perhaps one should first get the analog reference voltage to a precise value? (It looks rather close to 5V, but my one example is not very good statistics.)

3) Finally, I'm seeing slight differences in readings from the Bat1 and Bat2 inputs (when shorted together and connected to a variable supply). Originally I attributed this to the digital conversion, but if hundreds of readings are being averaged, I think my Link must have a persistent error. I wonder if this can also be calibrated, or if resorting to resistor changes is required for trimming the voltage sense inputs.

Perhaps this is going beyond this forum's level of interest or asking for too much help. If so, my apologies, and many thanks for the assistance so far.

James

[Rick]

I did not include voltage calibration in the procedure. In order to do that connect the two voltage inputs together and to a 40.0V stable source. After entering the calibration mode after power-up pushing SEL actually enters an auto-calibration mode for Amps and, when 40V is present on the voltage inputs, Volts. If the 40V is not present the voltage calibration is skipped. There is no manual adjustment for the voltage channels.

If the 5V reference is not within 5.00V +/- 15mV then either the reference is overloaded or is suspect. I have never seen the input voltage divider resistors (0.1% tol.) require replacement short of a lightning discharge or overvoltage application in which case there will be other collateral damage. The voltage readings @ 12.X Volts should be within 1 count of each other after the autoCAL. I am not recalling any other condition required to enter calibration..what are you observing?

The zero current calibration can be made with the two shunt wires connected to the negative reference input. In the field it suffices to merely have no current passing through the shunt or to disconnect all loads from the load end of the shunt while making the calibration. No discrete restance need be added to any wires.

There is no need or provision for an upper end calibration on current. The gain and stability is fixed and verified for each unit by design. I have not seen significant changes over time once a particular unit is checked before shipping. Changes of current accuracy are usually due to damage or to some influence on the voltage reference.

[JDM2007]

Originally Posted by Rick

I am not recalling any other condition required to enter calibration..what are you observing?

There is no need or provision for an upper end calibration on current. The gain and stability is fixed and verified for each unit by design.

I find that I can't get past the alternating voltage display to perform the zero-current calibration unless the Bat1 voltage displays between 40.0 and 40.4. Otherwise, the alternating voltages remain regardless of what buttons I press. Perhaps this restriction was added at some point to avoid inadvertent recalibration?

Since the gain for the current sense is likely to affect positive and negative currents equally, I can see that variations in the absolute gain are probably not usually an issue. However, since there are two gain modes, I wonder if a problem could arise when charging (positive current) generally occurs on the low range, but discharging (negative current) occurs on the high range. In this case, deviations from the design gains would seem to trend the calculated net charge away from the actual. Perhaps any error here is handled by the way the Link resets the AmpHours to zero under certain circumstances?

At this point, it seems my Link is now ready for action, many thanks for your time and generous assistance!

James

[muskoka]

I'm in the process of installing a Link 1000. It comes supplied with a 500 amp precision shunt. My battery bank is 650ah.

I presume the 500 amp shunt refers to the maximum charging current you can pass through it and not the AH size of the battery bank?

Thanks!

[GordMay]

For continuous operation, it is recommended that Shunts are not run at more than 66% of the rated current under normal conditions of use, so a 500A shunt will be suitable up to about 330 Amps continuous (500 x 0.666 = 330).

A shunt is a very accurate, very low resistance, temperature stable (up to 145̊ C*) resistor. When current flows through the shunt, a small voltage is developed across the shunt , which is proportional to the current flow. The ammeter (or battery monitor) accurately measures this very low voltage, and converts it to the "amps" reading on the meter/monitor.

Commercially manufactured shunts are typically available in either 50 or 100 millivolts.
Shunts, labeled 500 Amp, are actually specifying either 500 Amp/50 milli-Volt (500A/50mV), or 500A/100mV.
Ie: The Link 1000 and the Blue Sea Systems #8255 are both 500A / 50mV shunts. DC Shunts - Blue Sea Systems

* Shunt current ratings are usually based upon a 105̊ C rise. A shunt may operate safely up to about 145̊ C. If the ambient temperature is 40̊ C (103̊ F), a rise of 105̊C will yield a final temperature of 145̊C. This is the maximum temperature at which the shunt material (usually “Manganin”) maintains a stable resistance.

[muskoka]

Gord,

Does that mean a 500 Amp shunt is OK for a 650 amp-hour battery bank?

The charging current from the battery charger is 100 Amps max.

Thanks!

[GordMay]

Your 500A shunt will be suitable up to about 330 Amps continuous, either Charge or Discharge - which is more than adequate for your application.

The current necessary to nominally charge a 650 A/H (ampere-hour) battery from zero to full charge in 5 hours (130 amperes) is referred to as the battery's C/5 rate.
Conversely, a 100 A charge will replenish the 650 A/H battery in 6.5 hours, and would be referred to as a C/6.5 rate.

[muskoka]

Thanks Gord, That's what I'd thought but it's always best to double check when dabbling in the dark arts (electrics)!!!

[DanB444]

Would anyone have a similar calibration procedure for Xantrex Link 20? I've recently installed one and have been getting erroneous readings of -0.2 A on both battery banks even when the positive posts of the batteries are disconnected. It's not much, but it does add up to about -5 Ah every day.
Many thanks.
Dan