Axpert Voltage overshoots

[Chris Hobson]

I am not about to become a Victron groupie (sorry TTT - not yet anyway) but I am well aware of some folk (Dockarl immediately springs to mind) not being happy with their Axpert after various failures. It would be naive to ignore Axpert issues. I have in the past mentioned that I did not think Axpert's and Trojan batteries were a good mix and I now have some evidence. I am not a big fiddler so my settings remain constant. The most I do is to change my bulk voltage to suit battery temperature which has been variable in the last couple of weeks.

Having recently been logging data to Emoncms I have noticed some very big overshoots which don't last more than 10 seconds or so but well over the rating for the capacitors in the Axpert. I will try an explain under what conditions it occurs and propose ways to avoid it.

Today we did the ironing which is probably one of the worst culprits for setting off overshoots in that it is an intermittent load. The second factor that encourages overshoots is broken cloud with PV production being highly variable.

  

This is today's PV production (W) graph, not the best looking bell curve I have every seen. Part of the variation can be ascribed to the intermittent load but part of it is cloud cover.

This is the battery voltage feed covering the period around midday. One can clearly see the transition from bulk (constant current) to absorb (constant voltage) with the ironing causing perturbations. Right near the end of ironing when the SOC is about 94% and the batteries are accepting about 4-5% of capacity there is a massive overshoot to 63.41V. That voltage sustained for any length of time will pop the capacitors which are rated at 63V. This situation may be considerably worse if bulk was at 58.8V rather than the more moderate 56.4V that my batteries are currently set at.

How does one minimise the occurrence of these overshoots and perhaps negate the negative consequences?

1. Do not have intermittent loads. I would rather just take it on the chin. My missus would have my guts for garters. Folk who have access to Eskom perhaps run the iron off the grid.
2. Use batteries that have a moderate bulk voltage. This is a 7V overshoot which would take us into scary territory if  my bulk voltage was set for a Trojan.
3. Have a large battery bank. This I think is important. My 260Ah bank is just above the minimum recommended battery bank size. A large bank would be better able to absorb this overshoot. That minimum size is definitely there for a reason.

It is not all doom and gloom. My inverter has been in constant use for 21 months and both it and my battery bank are still here. There is no doubt I work both of them hard (@Don told me so) as being off grid they are my primary energy source. In the roughly 500 days that my BMV has logged data we have had 560 overshoots over 57V. Don's battery bank is twice mine's size and when we discussed these overshoots said he had not recorded the big overshoots I was seeing. Now I am recording to Emoncms and not relying on the BMV I see some of these overshoots are large indeed.

(I can hear TTT whistling  "We are the champions" to himself).

[___]

62v zener that activates some kind of circuitry that sinks the energy into a dump load?

Sent from my GT-I9195 using Tapatalk

[Mark]

"This situation may be considerably worse if bulk was at 58.8V rather than the more moderate 56.4V that my batteries are currently set at."

 

I have immediately set mine to 56.4v

@Chris Hobson How do I tell the bmv to tell me how many overshoots I had...or do look at solmon... todays project. My draadkar battery bank is only 200ah.

 

 

Sent from my SM-G800F using Tapatalk

 

 

 

[Chris Hobson]

2 hours ago, Mark said:

I have immediately set mine to 56.4v

You should set according to your battery datasheet. I also think my  3000W of panels exacerbates the situation. 1800W and you have ½ the problem. With fewer panels the Axpert has a greater chance of controlling the overshoots.

2 hours ago, Mark said:

How do I tell the bmv to tell me how many overshoots I had...

First you need to set  Program 37's threshold for high voltage alarm. Mine is set to a low 57V (0.6V above bulk), hence the high number of overshoots recorded. Then you can look in the BMV reader. The low voltage is set to 47V and sometimes in the morning it is triggered by the missus's morning cuppa but much less frequent as compared to the overvoltage.

 

 

 

[Mark]

36 minutes ago, Chris Hobson said:

You should set according to your battery datasheet. I also think my  3000W of panels exacerbates the situation. 1800W and you have ½ the problem. With fewer panels the Axpert has a greater chance of controlling the overshoots.

First you need to set  Program 37's threshold for high voltage alarm. Mine is set to a low 57V (0.6V above bulk), hence the high number of overshoots recorded. Then you can look in the BMV reader. The low voltage is set to 47V and sometimes in the morning it is triggered by the missus's morning cuppa but much less frequent as compared to the overvoltage.

 

Hi Chris (and thanks)

I think I have been lucky... also 1800W will make a difference, yes.

I feel better now...

[Guest]

 

On 3/25/2017 at 8:09 PM, Chris Hobson said:

(I can hear TTT whistling  "We are the champions" to himself).

Naaa, am a wee bit miffed at the moment..

My "cheap" Victron inverter blew a router when it switched off at low voltage, the big inverter is powering a online UPS, so that problem cannot surface there.

So ja, in my case the inverter was cheap but now I add the replacement cost of router and probably a small UPS. So inverter is not "cheap" anymore.

 

Am not sure the quick spikes, and I am guessing here, is a problem for the batts IF it is very quick but 10sec is not quick, or is it? 

Fact that the BMV records it means inverter could not use the power. So what is the effect on things connected direct to the batts over time like 48v to 12v converters, Pi's running direct off the batts or circuitry of battery balancers, is that not maybe a point to ponder?

[___]

4 hours ago, The Terrible Triplett said:

blew a router when it switched off at low voltage,

It could also be a complete coincidence. I'm not just saying that because I'm trying to defend "our champion", just in my experience, consumer grade networking stuff tend to fail every few years and a failure mode like this is common in my experience. It still works as long as it's switched on... and then it fails the moment there is a power failure and fails to come back on. Because the components are slowly failing due to heat soak.

I had an ADSL modem like that. Put it in the freezer for a few moments, it would boot right up... until the next power failure :-)

[Guest]

1 hour ago, plonkster said:

It could also be a complete coincidence.

Good point. It could have been, most definitely.

Keeping in consideration that the router in question was not cheap, barely over a year old and they are, as I have 2 routers and a Mikrotik, running in a very cool place. 

Replaced the expensive one with a cheapy this time round.

Will not throw "our champion" out yet, as it is only on maybe strike 2 - strike 1 was when the software they supply with great fanfare does not read the amps if the inverter is operating below 110w-120w. 

If strike 3 happens, they can keep this CHEAP range, will revert back to the older type. Was more robust in my opinion.

[Chris Hobson]

On 25/03/2017 at 8:39 PM, plonkster said:

62v zener that activates some kind of circuitry that sinks the energy into a dump load?

 

Hi Plonky

Am I going to need a 100W Zener and some sort of amplification circuit? Where do I get 100W Zeners?

 

[superdiy]

31 minutes ago, Chris Hobson said:

Hi Plonky

Am I going to need a 100W Zener and some sort of amplification circuit? Where do I get 100W Zeners?

 

I don't think zeners are the best way to go.  Rather a comparator and a resistive dump load.

[Chris Hobson]

14 minutes ago, superdiy said:

I don't think zeners are the best way to go.  Rather a comparator and a resistive dump load.

Morning SuperDIY

Using a comparator what are you going to use as V_ref?

[___]

1 hour ago, Chris Hobson said:

Using a comparator what are you going to use as V_ref?

A zener!

I didn't mean use the Zener itself. I meant use the Zener to trigger some other switching device that activates a dump load. Then you can get away with a 0.25W thingie, it just has to pass enough current to switch a transistor or a comparator.

Then again, these days Arduinos are so cheap, might as well use one of those (it already has a 5V supply and a reference), just add a nice big divider chain to get your 70V(ish) down to 5V(ish), and just sit in a loop reading a value on a port. Doesn't matter if the accuracy is somewhat off, you just want to chop off the worst of it. Using an arduino solution also means you can log stuff.

Then use a "big array of inexpensive lamps" (ie automotive bulbs) to sink the current. Or whatever else you may have... :-)

[Chris Hobson]

I had though of old 32V incadescent bulbs. Perhaps Automotive bulbs will be neater.

 

[superdiy]

23 hours ago, Chris Hobson said:

Using a comparator what are you going to use as V_ref?

Not a zener, they are too inaccurate, rather something like a TL431 / TS431 (Precision Shunt Regulator) or LM385-1.2 / LM385-2.5 (Micropower Voltage Reference Diode)

[___]

1 hour ago, superdiy said:

Not a zener, they are too inaccurate,

My thinking was that for the application at hand precision isn't too big an issue. Then I decided to actually look into the matter properly, discovered that Zeners do 2% to 5%, and at 5% it could be off a whole 3 volts, meaning your 62V protection device might well only wake up at 65V. Point taken, forget about the Zener.

Thinking back to the old battery regulator I built when I was in high school (for a 36V bank), I actually used an 18V Zener in series with a 7812 linear regulator. That was because the 7812 can only do inputs up to 35V, so adding the Zener would quickly drop some of it and bring it in range for the more accurate linear regulator. I then used a voltage divider to create a 6V reference for the comparator. This worked very well. Of course it was horribly inefficient by modern day terms (even driving a 12V relay at around 50mA from 36V, which means we burned like a whole 2 watts just on holding a relay closed... but it worked very well. Those were the days when PWM was just coming into vogue and MPPT have not been invented yet.