Arco Zeus regulator - first impressions and a look inside

[MtB]

9 minutes ago, dmr said:

(but adding a length of thin wire as a current limiter is much worse 😀)

 

But but, there is an elegance in its simplicity, shirley. Ok, an elegance in it being understandable to muppets like moi. 

 

I've just bought a ten year old Audi, its like driving a computer! 

 

 

If this is the future, I'm inclined to get myself a restored Ford Zodiac.

 

 

 

[dmr]

18 minutes ago, MtB said:

 

But but, there is an elegance in its simplicity, shirley. Ok, an elegance in it being understandable to muppets like moi. 

 

I've just bought a ten year old Audi, its like driving a computer! 

 

 

If this is the future, I'm inclined to get myself a restored Ford Zodiac.

 

 

 

 

But, an extra hot day, a bit of belt slip, or a few dog hairs inside the alternator, and you resistance wire will no longer be correct. Measuring the alternator case temperature is a big step closer to doing it right.

 

The new car is very comfotable and sorts of floats along nicely, and the steering wheel and pedals sort of tell it how to float along, but its not like actually driving it 😀. Our old mk1 Focus was a lovely car to drive.

At least the boat is nicely mechanical and free of processors........oh bugger, Ive just installed the Zeus.

[MtB]

22 minutes ago, dmr said:

But, an extra hot day, a bit of belt slip, or a few dog hairs inside the alternator, and you resistance wire will no longer be correct. Measuring the alternator case temperature is a big step closer to doing it right.

 

 

Dog hairs? What dog?! 

 

I tend to notice if the alt is overheating, the smell and the smoke draw it to my attention. 

 

But even if I miss it another A127 is only £60 in the Les Smith car accessory shop.... 

 

Oh the joys of Basic Boating....

 

 

 

[nicknorman]

7 hours ago, dmr said:

There is a bug where the Zeus occassionally looses control of voltage regulation, I see 15.2volt transients on my 14.8volt target. Could be an issue with Lithiums?????

This happens because the Zeus cycles its duty cycle up and down to control alternator temperature, so switching between a "constant temperature" and "constant voltage" regime. At the transition it sort of takes its eye off the ball and continues to ramp the duty cycle upwards before it gets it back in control and this gives an over-voltage pulse. I am hoping for a good response from Arco. 

 

This does seem a bit piss poor from a coding point of view. My code works out the max field current limit by looking at various issues, and then chooses the lowest one. The code for the max field current (max excite current = IEM) as a function of alternator temperature is pretty simple really (the max field current due to circuit resistance is 4.5A):

 

IEM_temp = 4.5 -(((float)(Alternator_Data.Alternator_Temperature - 75))/8);

 

So for every degree over 75, the max field current is reduced by 0.125A. It tends to stabilise at around 85 to 90 degrees (depending on ambient). There is no sudden switching between constant temperature and constant voltage regimes. Not rocket science!

 

   

3 hours ago, dmr said:

I think its reasonable that the Zeus needs its own voltage sense. This is a critical bit of info, in fact its the main objective of the Zeus, so it would be risky to rely on some other device providing data over a bus. In addition I suspect that bus issues and delays/slow update rates could play havoc with the control loops (see my post above 😀 ) .

It might be good to have an option to take temperature, and maybe current, from CANbus.

 

I have shared the Victron BMV shunt with the Zeus. Arco do not provide a shunt so I assume that this is their intention, and shunts in series is not a great idea.

There is a post somewhere on a Victron forum saying that sharing shunts is not good, but I feel that even though the signal is small the source impedance of a shunt is so incredibly low that it must be ok.

 

I don't see a problem connecting the Zeus to the BMV shunt. As you say, the impedances are massively different. But I would say that there is no particular issue with having shunts in series, I have the Mastershunt in series with the BMV. The max voltage drop is 50mV at 500A so it is not a big deal especially if you are careful with alternative ground paths (but don't tell IanD!)

3 hours ago, dmr said:

How quickly does stuff update on a typical CANbus system?.

I suspect the problem is that a sudden increase in engine speed, or more likely a reduction in load current as something turns off, will give an increase in voltage and the Zeus needs to respond to this as fast as possible.

I was trying to look at some transient responses today, and will look again next time I need to run the engine (Tuesday).

 

CANBUS can be quite fast  (1mB/sec) but of course you need to take into account the processing speed at each end. And often it is implemented at 250kB/s which is still quite fast bearing in mind you only need a few bytes of data to give (eg) current. I would agree though, that voltage needs to be internal not dependant on some databus which might or might not be plugged in. Ditto the entire regulation loop. But slowly changing or less imporant things like battery & alternator temperature, charge current (used for the purposes of deciding when to go to float) can all be sent over a databus providing the code notices and makes safe allowances for when the data is lost (due to you unplugging the cables etc)

3 hours ago, MtB said:

 

But but, there is an elegance in its simplicity, shirley. Ok, an elegance in it being understandable to muppets like moi. 

 

No I strongly disagree. There is no elegance in throttling a power source by disspating lots of power. The elegance comes in working out how much power should be generated in the first place and limiting it to that. Think steam engines - boiler is fired at full power all the time and excess steam is blown off, vs boiler is fired at a rate appropriate for the present load. This almost sounds like a modulating gas boiler!

[dmr]

16 minutes ago, nicknorman said:

 

This does seem a bit piss poor from a coding point of view. My code works out the max field current limit by looking at various issues, and then chooses the lowest one. The code for the max field current (max excite current = IEM) as a function of alternator temperature is pretty simple really (the max field current due to circuit resistance is 4.5A):

 

IEM_temp = 4.5 -(((float)(Alternator_Data.Alternator_Temperature - 75))/8);

 

So for every degree over 75, the max field current is reduced by 0.125A. It tends to stabilise at around 85 to 90 degrees (depending on ambient). There is no sudden switching between constant temperature and constant voltage regimes. Not rocket science!

 

   

 

I don't see a problem connecting the Zeus to the BMV shunt. As you say, the impedances are massively different. But I would say that there is no particular issue with having shunts in series, I have the Mastershunt in series with the BMV. The max voltage drop is 50mV at 500A so it is not a big deal especially if you are careful with alternative ground paths (but don't tell IanD!)

 

 

 

The Zeus has a very nice feature where it logs some parameters such as volts, amps and temperature, though only 600 samples which is maybe the most recent few minutes. These time history graphs showed the Zeus ramping duty cycle/current up and down to try to control temperature, so a bit like an unstable loop. Maybe it will "learn" and get better, or maybe my alternator is thermally very different to whatever models it might have built in.

Sadly I discovered that the screen grab feature on my iphone does not work so I could not capture this interesting stuff. Will try again another day.

The overshoot does look like a bit of bad programming, but then Zeus was done in a rush. As long as Arco fix it I will have no complaint.

 

Shunts in series are a bit like the length of thin cable, rather wasteful of otherwise low resistance cables.

As you say, its fine if each device is only using the shunt "signals" to derive current, but if it is using one terminal as a system ground then it gets a bit more worrying.

[cheesegas]

7 hours ago, dmr said:

Shunts in series are a bit like the length of thin cable, rather wasteful of otherwise low resistance cables.

The resistance of a shunt isn't ideal, but it's tiny compared with a hybrid cable. At 500a - which in most systems is very unlikely, as with a 12v nominal system that's the same as a 6kw AC load - that's a voltage drop of 50mv, or equivalent to a 0.0001 ohm resistor which is 25w power dissapation at full load.

 

A more likely load is half that, 250a which is equal to a four slice toaster or a double ring induction hob on a 3kva inverter. That's 12.5w dissapation, which isn't great as it's wasted power, but in the real world if you're running a 3kw load, 0.0125kw isn't of concern.

 

Now the same also happens in reverse, which is when power loss is more important...when charging. A likely high charge current is 120 amps, the most a JBD BMS can take. That works out at just over 6w of dissapation which is again insignificant. A 0.05v drop isn't much.

 

A hybrid cable is designed to drop significantly more than 0.05v..

[dmr]

1 hour ago, cheesegas said:

The resistance of a shunt isn't ideal, but it's tiny compared with a hybrid cable. At 500a - which in most systems is very unlikely, as with a 12v nominal system that's the same as a 6kw AC load - that's a voltage drop of 50mv, or equivalent to a 0.0001 ohm resistor which is 25w power dissapation at full load.

 

A more likely load is half that, 250a which is equal to a four slice toaster or a double ring induction hob on a 3kva inverter. That's 12.5w dissapation, which isn't great as it's wasted power, but in the real world if you're running a 3kw load, 0.0125kw isn't of concern.

 

Now the same also happens in reverse, which is when power loss is more important...when charging. A likely high charge current is 120 amps, the most a JBD BMS can take. That works out at just over 6w of dissapation which is again insignificant. A 0.05v drop isn't much.

 

A hybrid cable is designed to drop significantly more than 0.05v..

 

Before the Victron we had the old Adverc BMS/monitor, an old design so they used a higher resistance shunt, 100mV at 100amp I think, and I melted that one 😀.

 

The Victron shunt system is impressive, it resolves 10mA and that corresponds to just 1 microvolt (if my calculation is correct)

[cheesegas]

15 minutes ago, dmr said:

Before the Victron we had the old Adverc BMS/monitor, an old design so they used a higher resistance shunt, 100mV at 100amp I think, and I melted that one 😀.

 

The Victron shunt system is impressive, it resolves 10mA and that corresponds to just 1 microvolt (if my calculation is correct)

100mv is 50w, fair amount of power to shed for a small thing! Same as a big ish soldering iron.

 

 

 

On the DVCC issue, I've got somewhere. The firmware changelog on the Victron site mentions adding support for the Zeus and nothing more. However, the changelog doc on the Victron installer portal are more detailed, it says...

Quote

Add support for the ARCO Zeus alternator controller. Includes reading out locally on the display of the GX device as well as monitoring on the VRM Portal. Note that there is no DVCC integration. See Arco ZEUS their documentation on Victron integration for further information.

There's no documentation relating to DVCC on the Arco site either. It's a new product with over-the-air firmware updates via the app, so let's hope that adding DVCC in a later release is coming. I've emailed them to ask...not being able to use voltage/temp/current data from a shunt already in the system is a real negative point.

[TheBiscuits]

9 hours ago, dmr said:

Shunts in series

 

If you don't like them in series you could try wiring them in parallel ... 

 

 

 

 

 

 

 

 

 

 

(Disclaimer: This is a geeky joke.  It's not a technical recommendation! )

[IanD]

1 minute ago, TheBiscuits said:

 

If you don't like them in series you could try wiring them in parallel ... 

 

(Disclaimer: This is a geeky joke.  It's not a technical recommendation! )

 

If you did this, the reduced GND voltage drop would obviously reduce any risk of melting your CANbus cables... 😉 

[nicknorman]

1 hour ago, dmr said:

 

The Victron shunt system is impressive, it resolves 10mA and that corresponds to just 1 microvolt (if my calculation is correct)

Yes it does seem very good for a cheap product. There is a little bit of calibration drift but it is very small. 10mA in 500A is 16 bit resolution, it would be interesting to know how they did 16 bit AtoD but presumably there is some range switching in the AtoD converter input.

[IanD]

50 minutes ago, nicknorman said:

Yes it does seem very good for a cheap product. There is a little bit of calibration drift but it is very small. 10mA in 500A is 16 bit resolution, it would be interesting to know how they did 16 bit AtoD but presumably there is some range switching in the AtoD converter input.

Instrumentation ADCs with 20bit (or more!) resolution without range-switching and with low Vin are easily and cheaply available off-the-shelf nowadays... 😉 

 

https://www.icbase.com/psen/AdvProPage.aspx?Id=49

Edited March 4 by IanD

[nicknorman]

48 minutes ago, IanD said:

Instrumentation ADCs with 20bit (or more!) resolution without range-switching and with low Vin are easily and cheaply available off-the-shelf nowadays... 😉 

 

https://www.icbase.com/psen/AdvProPage.aspx?Id=49

 

Not going to work. From the spec sheet: Common-Mode Input Range AGND + 1.5V to AVDD – 1.5V 

And it would  be unidirectional whereas a shunt has to work with both + and - currents. Next...

[IanD]

Just now, nicknorman said:

 

Not going to work. From the spec sheet: Common-Mode Input Range AGND + 1.5V to AVDD – 1.5V 

And it would  be unidirectional whereas a shunt has to work with both + and - currents. Next...

 

It will work just fine with a shunt in the GND -- and if you actually read the spec it's bidirectional.

 

For a "smart shunt" like Victron used in the PWR line, the internal supplies to the ADC are also internally referred to PWR so it still works.

 

And if in spite af all this you still don't like this particular part, there are many others available, some with massive common-mode input voltage ranges... 😉 

[nicknorman]

Just now, IanD said:

 

It will work just fine with a shunt in the GND -- and if you actually read the spec it's bidirectional.

 

For a "smart shunt" like Victron used in the PWR line, the internal supplies to the ADC are also internally referred to PWR so it still works.

 

And if in spite af all this you still don't like this particular part, there are many others available, some with massive common-mode input voltage ranges... 😉 

 

A ground shunt will have one input at ground. But the spec says minimum input voltage is GND + 1.5v Not going to work.

But oh sorry, my mistake, of course it will work because you said it would and obviously you designed the chip.

[IanD]

5 minutes ago, nicknorman said:

 

A ground shunt will have one input at ground. But the spec says minimum input voltage is GND + 1.5v Not going to work.

But oh sorry, my mistake, of course it will work because you said it would and obviously you designed the chip.

Oops, misread the spec, my bad 😞 

 

In practice any circuit like this used to measure a voltage across a shunt will either generate suitable internal supplies referred to it, or will use an ADC with a common-mode range which extends to either above VDD, below GND, or both.

 

And that was a good guess on your part, because I *have* designed similar ADCs which did exactly that -- though not this particular one, because I didn't work for TI 🙂 

Edited March 4 by IanD

[dmr]

2 hours ago, TheBiscuits said:

 

If you don't like them in series you could try wiring them in parallel ... 

 

 

 

 

 

 

 

 

 

 

(Disclaimer: This is a geeky joke.  It's not a technical recommendation! )

Thats ok but I think  they are llike batteries and must be from the same manufacturer and have the same capacity.

and you get an RPi and write some software to add the two currents together 😀

[IanD]

20 minutes ago, dmr said:

Thats ok but I think  they are llike batteries and must be from the same manufacturer and have the same capacity.

and you get an RPi and write some software to add the two currents together 😀

 

They don't *need* to be the same, but if they're not (different current ratings or rated voltage drop) the current capacity won't necessarily be the sum of the two individual ones -- but the current reading when the RPi adds the numbers up may still be correct. They're just very low-value resistors with voltage sensing terminals... 😉 

 

For example, a 500A 50mV (0.1mohm) shunt in parallel with a 250A 50mV (0.2mohm) shunt will give a 750A 50mV shunt when you add the readings together. But a 500A 50mV (0.1mohm) shunt in parallel with a 500A 100mV (0.2mohm) shunt won't work as expected (it'll be a 750A 50mV shunt) unless you scale the readings in the Rpi before adding them together...

 

None of which is relevant in practice, paralleling shunts is a really bad idea in practice (think contact resistance...) even if it looks OK in theory... 😞 

Edited March 4 by IanD

[dmr]

1 hour ago, nicknorman said:

Yes it does seem very good for a cheap product. There is a little bit of calibration drift but it is very small. 10mA in 500A is 16 bit resolution, it would be interesting to know how they did 16 bit AtoD but presumably there is some range switching in the AtoD converter input.

 

I'm sure you know but there are a few chips on the market that do the whole job of shunt measurement (and power calculation) with an I2C interface, I think they are 20bit. I started using one to do my own amp-hour counter (with a positive shunt) but decided it was better and easier to fall into the Victron family.

[nicknorman]

1 minute ago, dmr said:

 

I'm sure you know but there are a few chips on the market that do the whole job of shunt measurement (and power calculation) with an I2C interface, I think they are 20bit. I started using one to do my own amp-hour counter (with a positive shunt) but decided it was better and easier to fall into the Victron family.

 

Yes this does seem to be the case. Of course number of bits is all very well, but one also needs to look at noise, linearity and offset and gain temperature co-efficient etc. But yes there are certainly devices that do fit the bill.

[dmr]

2 hours ago, IanD said:

 

They don't *need* to be the same, but if they're not (different current ratings or rated voltage drop) the current capacity won't necessarily be the sum of the two individual ones -- but the current reading when the RPi adds the numbers up may still be correct. They're just very low-value resistors with voltage sensing terminals... 😉 

 

For example, a 500A 50mV (0.1mohm) shunt in parallel with a 250A 50mV (0.2mohm) shunt will give a 750A 50mV shunt when you add the readings together. But a 500A 50mV (0.1mohm) shunt in parallel with a 500A 100mV (0.2mohm) shunt won't work as expected (it'll be a 750A 50mV shunt) unless you scale the readings in the Rpi before adding them together...

 

None of which is relevant in practice, paralleling shunts is a really bad idea in practice (think contact resistance...) even if it looks OK in theory... 😞 

 

Ahhh I always wonderred what a shunt was 😀

I think you missed the smiley in my post.

Anyway if you did have two shunts in parallel then the sum of the two currents will indeed be the correct total current, I think it was you who was talking about Kirchoff recently???  😀

2 minutes ago, nicknorman said:

 

Yes this does seem to be the case. Of course number of bits is all very well, but one also needs to look at noise, linearity and offset and gain temperature co-efficient etc. But yes there are certainly devices that do fit the bill.

 

A man who understands the difference between resolution and accuracy 😀, Its surprising how many struggle with this.

I note than the Victron displays voltage to two decimal places but Zeus only does one. The Smartgage does one and a half.

[IanD]

17 minutes ago, dmr said:

 

Ahhh I always wonderred what a shunt was 😀

I think you missed the smiley in my post.

Anyway if you did have two shunts in parallel then the sum of the two currents will indeed be the correct total current, I think it was you who was talking about Kirchoff recently???  😀

 

A man who understands the difference between resolution and accuracy 😀, Its surprising how many struggle with this.

I note than the Victron displays voltage to two decimal places but Zeus only does one. The Smartgage does one and a half.

Even though it's a bad idea, the sum of the currents will be correct -- Mr Kirchoff, as you said.

 

But the total rated current won't be the sum of the two individual rated currents unless they have the same rated full-scale voltage drop (e.g. 50mV) -- Mr Ohm gets an oar in here too... 😉 

 

Resolution, noise, accuracy and linearity are all very different things, as you say -- it's fairly easy getting 20b resolution, harder getting 20b linearity and noise, and *much* harder getting 20b accuracy since this is about 1ppm, or 1uV accuracy on a 1V reading. Very expensive DVMs (thousands of pounds) manage it but not much else does... 😉 

[dmr]

45 minutes ago, IanD said:

Even though it's a bad idea, the sum of the currents will be correct -- Mr Kirchoff, as you said.

 

But the total rated current won't be the sum of the two individual rated currents unless they have the same rated full-scale voltage drop (e.g. 50mV) -- Mr Ohm gets an oar in here too... 😉 

 

Resolution, noise, accuracy and linearity are all very different things, as you say -- it's fairly easy getting 20b resolution, harder getting 20b linearity and noise, and *much* harder getting 20b accuracy since this is about 1ppm, or 1uV accuracy on a 1V reading. Very expensive DVMs (thousands of pounds) manage it but not much else does... 😉 

 

Its interesting looking at the various digital voltmeters and current monitors on eBay, or even looking at the Victron specs. A few do display to two decimal places, 10mV and 10mA, but the accuracy spec is sometimes as bad as 3%, especially on the current, and accuarcy is relative to FSD, so the figures do need to be treated with a bit of caution, but can still be very useful if used correctly.

 

The Zeus and the Victron volts and amps agree well. I have not looked at the Zeus amp-hours capability yet.

The current was possibly a little off on the Zeus alternator shunt but I did use a cheap chinese shunt that I had in the parts box. Maybe Arco thought of this as they allow a correction factor to be specified.

[IanD]

6 minutes ago, dmr said:

 

Its interesting looking at the various digital voltmeters and current monitors on eBay, or even looking at the Victron specs. A few do display to two decimal places, 10mV and 10mA, but the accuracy spec is sometimes as bad as 3%, especially on the current, and accuarcy is relative to FSD, so the figures do need to be treated with a bit of caution, but can still be very useful if used correctly.

 

The Zeus and the Victron volts and amps agree well. I have not looked at the Zeus amp-hours capability yet.

The current was possibly a little off on the Zeus alternator shunt but I did use a cheap chinese shunt that I had in the parts box. Maybe Arco thought of this as they allow a correction factor to be specified.

 

A correction factor can take out gain/full-scale errors, but only under the conditions it's calibrated at (and you need something more accurate to calibrate against). If temperature varies then both shunt resistance and ADC full-scale are also going to change, unless there's a temperature sensor on the shunt and a correction curve built-in -- which would be perfectly possible in a smart shunt like the Victron, it's what we often do on-chip.

[dmr]

I suspect the "current shunt amplifier on a chip" does temperature correction if required. I suspect this is what Victron uses I have wondered if the smartshunt suffers from temperature issues, the standard BMV looks to have no electronics on the shunt mounted pcb. I also note that both the BMV shunt and smartshunt obstruct any airflow over the shunt.

I belive that shunts are made from some clever alloy that has minimal temperature co-efficient.