Epever charge controller dropping out of MPPT mode

[venquessa]

I will re-read the thread, but is this the behaviour you are referring to?

Trace 4210N 330W panel. 24V Lithium.

From dawn the panel voltage rises to 28V and basically tracks the battery voltage. Note, I used to think this was "buck converter 100% duty cycle" and a direct connection from panel to battery in low light. However, as you can see, when I upgraded to a 41VoC panel and a 24-28V battery the panel does not directly track the battery voltage, but is offset. So some DC/DC conversion is happening.

Each hour, seemingly around ten to the hour, each hour, it resets the MPPT tracking algorithm, forcing it to go "open circuit" and start hunting from there. Sometimes this initiates the MPPT, sometime it initiates on it's own anyway. MPPT is easy to see working on this fairly dull morning.

It's so dull right now (pouring rain) it gone back into "battery voltage tracking".

Personally, as a professional software engineer, I don't like that hourly reset. It sounds like a hack or a work around for some corner of "double knee" condition which causes the algorithm to get stuck. It's a cheap fix though and when I was designing my own algorithm I had even intended "recon scans" which probably would have ended up looking just like the EPEver approach!

Also note, the spikes are actually amplified as the system beyond the MPPT is under about 150W of load.

 

[lightbulb]

I will re-read the thread, but is this the behaviour you are referring to?

Trace 4210N 330W panel. 24V Lithium.

From dawn the panel voltage rises to 28V and basically tracks the battery voltage. Note, I used to think this was "buck converter 100% duty cycle" and a direct connection from panel to battery in low light. However, as you can see, when I upgraded to a 41VoC panel and a 24-28V battery the panel does not directly track the battery voltage, but is offset. So some DC/DC conversion is happening.

Each hour, seemingly around ten to the hour, each hour, it resets the MPPT tracking algorithm, forcing it to go "open circuit" and start hunting from there. Sometimes this initiates the MPPT, sometime it initiates on it's own anyway. MPPT is easy to see working on this fairly dull morning.

It's so dull right now (pouring rain) it gone back into "battery voltage tracking".

Personally, as a professional software engineer, I don't like that hourly reset. It sounds like a hack or a work around for some corner of "double knee" condition which causes the algorithm to get stuck. It's a cheap fix though and when I was designing my own algorithm I had even intended "recon scans" which probably would have ended up looking just like the EPEver approach!

Also note, the spikes are actually amplified as the system beyond the MPPT is under about 150W of load.

View attachment 149149

The poor behavior I am referencing takes place at ~1115 on your plot. The panel voltage drops to 28V. Based on your chart the Vmp is higher, maybe 32V. It would be interesting to know what happened there. Maybe it transitioned from boost to float? My opinion is it should not do that. It is not as drastic when Vmp is close to the battery voltage, but a 100V Vmp on a 12 or 24 volt system is a real problem.

I have not noticed those spikes before, your sample rate is probably pretty good. I imagine it is designed to validate Vmp every hour.

 

[lexio]

i have now test for a while still dropping to 14.6 v so i would say it is a solid NO BUY . brand Epever model 4210 an
see what a victron can do planning to buy a victron smartsolar mppt 100/50

 

[venquessa]

For what it's worth I now have a Trace 4210AN in parallel with a Victron SmartSolar 75/15.

Tomorrow I will have 2 Victron SmartSolars and can do a direct test between the two with identical side by side panels.

At the moment the Victron is handling two panels and the EPEver 1.

Observations. The victron MPPT is far faster and much more explorative. It moves it's voltage constantly over about 1 volt. It also do the "spikes" where it resets the MPPT tracking to open circuit and refinds the MPP... except the victron does this every 10 minutes rather than every hour.

As to efficiencies between the two I can't give a like for like comparision via data just yet. Bear with me if you can and I should be able to produce side by side direct tracking graphs for both.

 

[venquessa]

A word of caution.

The tracer 4210AN is about £120. It's 40A 100V.

A BlueSolar MPPT 75/10 the smallest Victron is about £80.

Do not be fooled by the price of those low end units. For an equivalent 40A and 100V you would need the MPPT 150/50, which is more in the order of £250.

When you have priced that up, consider you can't configure it unless you either buy the Bluetooth (SmartSolar) version (usually +20%) AND/OR you invest in the appropriate cables and admin hardware. For many things, not all, you can use a raspberry PI, but you still have to buy or provide the fully DC isolatied interface cables.

That all adds up. My £80 VE MPPTs, rose to £110 to get Bluetooth, then I found out I also need the VEDirect cables £35. So they quickly became £145 controllers!

 

[venquessa]

This is just the VE MPPT75/15 in action on a near cloudless but slight hazy day (single 330W panel).

EDIT: Of note, the two periods with elevated power point are periods of "situational shading". The house in the morning and the hedge in the evening interfere.

 

[lexio]

just orderd this one https://www.obelink.nl/victron-smartsolar-mppt-100-50-laadregelaar.html
Help my bank account is braking ?

i have observed this problem already a few summers
hopefully this will solve the problem will report back

 

[venquessa]

The reason I can't quite show you the comparison, even for 2 vs. 1 panels... is this is the mess it made today.

There are all manor of different things going on here from top of charge, straight to load, float, boost, HVCs on the BMS, everything! 1000W panels into a 100Ah battery.

 

[venquessa]

Help my bank account is braking ?

It has Bluetooth config. It works. If you want to integrate into VEBus (GX or Raspberry Pi) you need the victron VEDirect cable.

Note. It doesn't need to be a victron cable. Any UART to USB, Isolated interface/cable will do. As those are not easy to find and thus rarely cheap the victron cables aren't "that" bad.

 

[lexio]

forbidden chart to read when you are drunk

 

[lexio]

It has Bluetooth config. It works. If you want to integrate into VEBus (GX or Raspberry Pi) you need the victron VEDirect cable.

Note. It doesn't need to be a victron cable. Any UART to USB, Isolated interface/cable will do. As those are not easy to find and thus rarely cheap the victron cables aren't "that" bad.

do you have a web adress where to buy one and thanks for the info

 

[venquessa]

https://www.amazon.co.uk/dp/B01LZ6WTLW

I assume Amazon will redirect you to a local seller.

Actually. I think mine came FROM the Netherlands

 

[venquessa]

Disclaimer. Please check this IS indeed the required cable for your MPPT. The bigger units switch over to VEBus I believe. DIfferent cable!

 

[venquessa]

forbidden chart to read when you are drunk

Yep! You know what impressed me. Through out all of that chaos on the MPPT DC bus. The Multiplus never once so much as glitched the AC power. At one point I believe, unconfirmed, but the SmartSolar spun up and produced DC output direct to the invert while the BMS charge was disconnected. It should be able to do so, although, it would be difficult to regulate the DC voltage... so I'm not convinced.

EDIT: Maybe overthinking, but..., if the BMS cuts the charge circuit to protect the battery, the sudden removal of a current sink seems to reset both MPPTs to zero. However, they both still have the ability to take power from the BMS and so they still know the voltage. They also have their "MAX Output voltage". In theory, the MPPTs should be able to compete (and win) over the battery on the DC bus, supplying the inverter. They may only really be competing to keep the input caps on the inverter above battery voltage, but it "should" work.

I shall have a think.

 

[lexio]

the most imporant thing for me is when i drive my rv and going in and out of sunshine lets say tunnels tree lanes cose the epever was realy bad
always dropping down to 14.2v -- charge 4 amp until i pulled the fuse of my solar then it start working a gain 35 v 25 amp charge in a 12v system

sorry no spell check

 

[venquessa]

I can't say really. Solar is fussy. If I put a single finger across one cell of a panel, I limit not just the current through that cell, but all the cells on the panel. I don't just drop the output of "that" cell to 80%, I drop the whole panel to 80%. Worse. It drops all panels in parallel to that current limit. A single cell.

I think what you will benefit from is the victron MPPT seems to reset itself on a 10min cycle rather than an hour long cycle. That should mean the MPPT in the Victron resets itself before you intervene on the fuse

 

[venquessa]

The MPPT (Maximum power point tracking) algorithm seems very basic and it is. However it's a very, very precise and sensitive kind of basic. Given an ideal panel and sunny day the "high school" implementation will work fine. Given real world conditions where clouds arrive and reflect more sunlight for a few minutes, then cut most of it, then reverse every few minutes.... the MPPT process has what we could call "false knees" and "false elbows" which cause "algorthimic stagnation". Basically it backs itself into an oscillation around a "false knee" point far away from the actual power point. The "RESET" is a quick way to stop this persisting.

It's also a dirty hack. It's not elegant and ... well, it's what most do. The gains from more advanced MPPT will not offset well against the cost of more micro-controller power and... software engineering shows us that the more complex you make something the less reliable it is.

 

[venquessa]

On YouTube Jullian Ilett experienced this beautifully on his first attempt to implement an MPPT "bare bones" in a series. His got caught in a false knee. I can't recall if it was for the usual or different reasons.

 

[venquessa]

If you dig deeper into the equations, you have to include the mirror equation for efficiency.

If "Direct panel to battery" 100% duty cycle results in 100W of power flow.... and at maximum power power the system produces 110W at 80% efficiency.... a "good" MPPT will default back to 100% duty and direct to battery.

MPPT involves DC to DC conversion which is lossful.

Avoiding it can actually produce higher output. Do not be fooled by voltage OR current. You MUST consider both together on the input and the output of the charge controller.

It could turn out it was right and you were wrong. Most likely given the symptoms it was wrong because it was too slow to realise it got stuck in a "shadow" conditon at battery direct when the sun was indeed shining and if it had reset faster it would have realised that.

 

[venquessa]

MPPT in a it's high school form.

Draw more current.
Does the "power" increase?
Yes - Loop.
Other wise:
Does the "power" decrease?
Draw less current and loop.

The "battery coupling" with a "charge" controller wraps that in an outer loop which gives the battery charge current and voltage limits priority over the results. The sun might say "MORE!", the battery can say, "No thanks".

As with all software the concept and core is simple. It's when you address all the corner cases and complete the thing to be fool and full proof... that is the 80% of the 80/20 rule.