20 minutes ago, frivan said:

Jip. I think I got 180V and 70V but can't remember which way around. I did notice that fluctuating utility voltage had a big impact on the power. Probably because of V*V/R.

I know I actually need 6 capacitors, but wanted to check it it works with the 3, so I should see 118v with the 3. Everything is wired up, just waiting for loadshedding to end then I will check.

IT WORKS. YOU GUYS ARE F**KING AWESOME!!!!

160v with the 3 caps. An unfortunate side effect is that the timer does not work on 160v, so I will have to rethink my circuitry.

But huge thanks to each and every one who assisted.

2 hours ago, sjp100 said:

Thank you. Will I only see a proper voltage drop once I connect a large load on it? Measuring with no load doesn't change the voltage. 230v in and 230v out.

Go back to your formula and use say 600uF vs 100uF and you will see that with caps in parallel you reduce the Zc. 

Your reply was right that you increase the value by going in parallel. 

Zc=1/(2*pi*f*_cap)

52 minutes ago, sjp100 said:

IT WORKS. YOU GUYS ARE F**KING AWESOME!!!!

160v with the 3 caps. An unfortunate side effect is that the timer does not work on 160v, so I will have to rethink my circuitry.

But huge thanks to each and every one who assisted.

A geyser timer should be wired to 230V but the output can supply the series setup of the element and caps.

1 hour ago, frivan said:

A geyser timer should be wired to 230V but the output can supply the series setup of the element and caps.

Why would a timer that needs 90-230V not work? 

1 hour ago, Scorp007 said:

Why would a timer that needs 90-230V not work? 

Sure, that would work. The cheap relays typically don't have such high end power supplies. I would install a Geyserwise with eWe-link (Tuya?) interface in order to monitor the temperature. The problem is just that Geyserwise throws an error if the temperature doesn't increase fast enough. So, you would have to use another relay to drive the heating circuit.

1 hour ago, frivan said:

Sure, that would work. The cheap relays typically don't have such high end power supplies. I would install a Geyserwise with eWe-link (Tuya?) interface in order to monitor the temperature. The problem is just that Geyserwise throws an error if the temperature doesn't increase fast enough. So, you would have to use another relay to drive the heating circuit.

As per your box of tricks I was thinking around a Sonoff where the timer drives a relay. I don't like driving the Sonoff directly to the load even at around 10A purely due to the internal relay used. Temp indication can be a nice add on. 

8 hours ago, Scorp007 said:

As per your box of tricks I was thinking around a Sonoff where the timer drives a relay. I don't like driving the Sonoff directly to the load even at around 10A purely due to the internal relay used. Temp indication can be a nice add on. 

You get the PowR3 which can do heavy loads but this setup should draw around 7A. So, the TH16 or newer temperature controllers should be fine. I liked the temperature fuse of the Geyserwise. You don't want the controller freezing and pushing the geyser to 100 degrees.

Hi guys. I finalized my install over the weekend and added it to my main install post. You can view that here if you wish. once again thank you to @frivan @Scorp007 @markus_m2 @Modina @BritishRacingGreen @GerhardK83 @JustinSchoeman @TaliaB and anybody who I might have forgotten who assisted in this task. I really appreciated each ones input.

I would just use a TRIAC. It is cheaper, and it is possible to adjust the power you want to send to the geiser precisely and easily.

Just now, Mauritius B said:

I would just use a TRIAC. It is cheaper, and it is possible to adjust the power you want to send to the geiser precisely and easily.

This reduces the thermal load on the inverter, but does not alter the peak load at all. So depending on what is limiting the inverter's max output, it may not help.

On 2023/10/20 at 2:09 PM, JustinSchoeman said:

This reduces the thermal load on the inverter, but does not alter the peak load at all. So depending on what is limiting the inverter's max output, it may not help.

Please provide more info on the 2 levels of power mentioned above.

When the triac is triggered at say 90 degrees on the sine wave the DC power used is only 50% of full power.

How does the inverter still have to provide the full peak power?

On 2023/10/20 at 2:07 PM, Mauritius B said:

I would just use a TRIAC. It is cheaper, and it is possible to adjust the power you want to send to the geiser precisely and easily.

I can't see why above curcuit posted by @Mauritius Bcould not work to decrease wattage of the geyser element. 

The average load power is controlled by varying the RMS value of the load voltage. This may be achieved by using inverse parallel-connected SCRs, or by using a single triac.

Alternating current load control using triacs is the most common thyristor control method for low to moderately high power AC loads. The load power is controlled by controlling the RMS value of the load voltage, which in turn is controlled by the trigger angle of the triac. This, in turn, controls the conduction time of the triac.

When the trigger angle is zero, the conduction angle will be 180°, and the load voltage will be a maximum and equal to the supply voltage. As the trigger angle is increased, the conduction angle will decrease, and the load voltage will decrease. When α is equal to 180°, θ will be zero, and the load voltage will be zero.

13 hours ago, Scorp007 said:

Please provide more info on the 2 levels of power mentioned above.

When the triac is triggered at say 90 degrees on the sine wave the DC power used is only 50% of full power.

How does the inverter still have to provide the full peak power?

This is what a triac limited waveform looks like:

For big chunks of the input waveform, the output is 0 - but for the rest the output exactly matches the input.

So, unless you reduce power to less than 50% you will still get the maximum positive and negative peak output values.

If you look at the unmodified sine wave, then the instantaneous peak voltage of the output is ~324V. If you consider a 4kW element (13 Ohm / 230V), then the instantaneous peak current will be ~25A for an instantaneous peak power of ~8kW.

RMS average voltage is 230V, current ~17A and power is 4kW.

As you start blanking the waveform with the triac, you immediately reduce the RMS values. But until you reduce power to <50% you will not have trimmed off the peak of the sine wave, and the inverter will still be outputting an instantaneous 8kW at the peak.

The output IGBTs have both RMS and pulse limits for current.  Loss calculations for power are ridiculously complex, but will also give different continuous/pulsed limits.

The problem is, which of those two limits will bite first?  If it is RMS current, then you are fine.  If it is pulsed current then the magic smoke comes out.

 

53 minutes ago, JustinSchoeman said:

This is what a triac limited waveform looks like:

For big chunks of the input waveform, the output is 0 - but for the rest the output exactly matches the input.

So, unless you reduce power to less than 50% you will still get the maximum positive and negative peak output values.

If you look at the unmodified sine wave, then the instantaneous peak voltage of the output is ~324V. If you consider a 4kW element (13 Ohm / 230V), then the instantaneous peak current will be ~25A for an instantaneous peak power of ~8kW.

RMS average voltage is 230V, current ~17A and power is 4kW.

As you start blanking the waveform with the triac, you immediately reduce the RMS values. But until you reduce power to <50% you will not have trimmed off the peak of the sine wave, and the inverter will still be outputting an instantaneous 8kW at the peak.

The output IGBTs have both RMS and pulse limits for current.  Loss calculations for power are ridiculously complex, but will also give different continuous/pulsed limits.

The problem is, which of those two limits will bite first?  If it is RMS current, then you are fine.  If it is pulsed current then the magic smoke comes out.

 

Thanks a lot for the detail around this way of controlling the power used. 

I thought by using the DC side is a better way as it eleminates the peaks seen on AC waves. 

My results meased from the DC side 12V.

Inverter no load 0.75A

25% power 2.23A

50% power 6.3A

75% power 10.6A

Using these figures it points to the fact that one can even use battery power / PV to slowly heat the geyser at a much reduced power via the inverter. 

The above readings from DC was with a pure sine wave inverter. 

If there is interest I can redo this test on a much higher load. 

 

 

On 2023/10/20 at 2:09 PM, JustinSchoeman said:

This reduces the thermal load on the inverter, but does not alter the peak load at all. So depending on what is limiting the inverter's max output, it may not help.

The author of the post only mentioned the need to match his solar production to geiser power consumption. No referral to any inverter power issue so I do not understand your point. We all have a nice bunch of wave trimming devices at home and no inverter is dying because of them.

1 hour ago, Mauritius B said:

We all have a nice bunch of wave trimming devices at home and no inverter is dying because of them.

You will want to revisit this claim of yours  because you will be shortening the life of the inverter.  Scaling down 4kw to 2kw (50% duty cycle) using pwm on a per 50hz cycle basis is a brutal challenge for  your inverter,amongst other, the HV DC BUS capacitors life will be shortened because of the very high ripple current  its needs to handle. 

Edited October 22, 20232 yr by BritishRacingGreen

15 hours ago, BritishRacingGreen said:

You will want to revisit this claim of yours  because you will be shortening the life of the inverter.  Scaling down 4kw to 2kw (50% duty cycle) using pwm on a per 50hz cycle basis is a brutal challenge for  your inverter,amongst other, the HV DC BUS capacitors life will be shortened because of the very high ripple current  its needs to handle. 

So you mean that I have to stop using my vacuum cleaner (1800w Triac speed control) and also my hair dryer (1000/2000w just with a half wave trim diode), all my dimmable lightning fixtures or my 2000w full electronic kitchen oven just because I am using a hybrid inverter?

DC bus capacitors are there exactly to deal with ripple and TRIAC phase triggering is the most common control system both for domestic and industrial appliances so, maybe inverter designers have taken it into consideration.

This is just my point of view and an alternative to the question raised by the author. If somebody wants to use piles of capacitors in series I will also respect it.

On 2023/10/22 at 11:06 AM, JustinSchoeman said:

This is what a triac limited waveform looks like:

For big chunks of the input waveform, the output is 0 - but for the rest the output exactly matches the input.

So, unless you reduce power to less than 50% you will still get the maximum positive and negative peak output values.

If you look at the unmodified sine wave, then the instantaneous peak voltage of the output is ~324V. If you consider a 4kW element (13 Ohm / 230V), then the instantaneous peak current will be ~25A for an instantaneous peak power of ~8kW.

RMS average voltage is 230V, current ~17A and power is 4kW.

As you start blanking the waveform with the triac, you immediately reduce the RMS values. But until you reduce power to <50% you will not have trimmed off the peak of the sine wave, and the inverter will still be outputting an instantaneous 8kW at the peak.

The output IGBTs have both RMS and pulse limits for current.  Loss calculations for power are ridiculously complex, but will also give different continuous/pulsed limits.

The problem is, which of those two limits will bite first?  If it is RMS current, then you are fine.  If it is pulsed current then the magic smoke comes out.

 

That is exactly the same with a trimmed and untrimmed wave.

That is why we use RMS values in AC. In AC, you enter in the domain of time and the product of V*I changes on every single time interval during wave period.

The effective power that you apply to a load in AC is calculated as an extrapolation to DC equivalent power.

As you start to trim the wave, you also trim the time that voltage and current are applied to lthe load thus reducing the effective power it receives.

OK - Let's try an absurdly over simplified example.

I have a 3kW inverter.

I also have an old-fashioned, thermostat controlled 2 plate stove. Each plate is rated at 2kW.

I turn one of the plates to setting '1', and use an accurate power meter to measure the RMS average power over 10 minutes, and it is 200W.

I use another power meter which measures instantaneous power, and it shows something like: 2kW ... 0kW ............2kW ... 0kW............2kW ...

I can clearly see the plate turning on for a few seconds, and then off again to maintain the target temperature.

I turn the first plate off, and do the same test with the other plate, with the same results.

(Now I need to cheat a little and assume both plates are still at operating temperature to avoid a long 'on' period to heat up the plates...)

I now turn both plates on.  I can see the instantaneous power goes: 2kW ... 0kW ..... 2kW ... 0kW ... 4kW - and the inverter trips.

Each plate individually only uses 200W, but the combination trips a 3kW inverter because the instantaneous load is 4kW.

 

Exactly the same thing happens in triac controlled loads - just in much shorter intervals.  The average RMS power may be 200W, but there are still short 2kW peaks.

In reality, things are a little easier in the triac case, as the inverter output IGBTs typically have almost double the pulse rating vs continuous rating (although some of that margin is used for DC bus regulation).

 

As a side note, the above example is a little better if we use capacitor dropping to reduce the plate power to 200W.  The would result in a PF of 0.31, and the inverter would need to deliver 300/0.31 = 632W per plate, or nearly 1300W for the two plates combined.  No where near the reduction you would expect.

If you are relying on derating (triac or capacitor) to connect bigger total loads than what the inverter is rated for, then the results may not be entirely what you expect - even if the RMS average power of the total load is less than the inverter rating.

This does not only affect inverters.  Many utilities will bill large industrial users based on load harmonics and PF to compensate for the effective power they need to deliver to meet the real load requirements.

@JustinSchoeman Is correct on this. Many inverters warn against the use of phase angle devices like this in their documentation - in fact you can cause far more complicated issues than is covered here. Luckily most inverters handles these issues remarkably well, but in general phase angle triggering is a bad idea on this power level even if you are just on the grid.

11 hours ago, JustinSchoeman said:

OK - Let's try an absurdly over simplified example.

I have a 3kW inverter.

I also have an old-fashioned, thermostat controlled 2 plate stove. Each plate is rated at 2kW.

I turn one of the plates to setting '1', and use an accurate power meter to measure the RMS average power over 10 minutes, and it is 200W.

I use another power meter which measures instantaneous power, and it shows something like: 2kW ... 0kW ............2kW ... 0kW............2kW ...

I can clearly see the plate turning on for a few seconds, and then off again to maintain the target temperature.

I turn the first plate off, and do the same test with the other plate, with the same results.

(Now I need to cheat a little and assume both plates are still at operating temperature to avoid a long 'on' period to heat up the plates...)

I now turn both plates on.  I can see the instantaneous power goes: 2kW ... 0kW ..... 2kW ... 0kW ... 4kW - and the inverter trips.

Each plate individually only uses 200W, but the combination trips a 3kW inverter because the instantaneous load is 4kW.

 

Exactly the same thing happens in triac controlled loads - just in much shorter intervals.  The average RMS power may be 200W, but there are still short 2kW peaks.

In reality, things are a little easier in the triac case, as the inverter output IGBTs typically have almost double the pulse rating vs continuous rating (although some of that margin is used for DC bus regulation).

 

As a side note, the above example is a little better if we use capacitor dropping to reduce the plate power to 200W.  The would result in a PF of 0.31, and the inverter would need to deliver 300/0.31 = 632W per plate, or nearly 1300W for the two plates combined.  No where near the reduction you would expect.

If you are relying on derating (triac or capacitor) to connect bigger total loads than what the inverter is rated for, then the results may not be entirely what you expect - even if the RMS average power of the total load is less than the inverter rating.

This does not only affect inverters.  Many utilities will bill large industrial users based on load harmonics and PF to compensate for the effective power they need to deliver to meet the real load requirements.

Thank you!

It is a nice example but I would like to clarify the concepts of power and energy a little bit more.

The plates will have a rated power of 2KW no matter what you do, but the energy they use is 200Wh because they are not on all the time. If they remain on for a full hour then they will use 2KWh each.

If you have a 3KW rated inverter and both plates are switched on at the same time, the instant power drawn is 4KW and the inverter trips, but you are still measuring 400Wh of energy usage on your meter.

Again, the concept of time is appearing on the equation because we are talking about energy and not only power.

Let me remind again that the author only wants to derate the geiser because he wants to match it with solar PV production, not because of inverter rated output power.

If we go back to the capacitors, when your PF is different than 0 you have undesirable effects on your distribution lines like current increase that forces utilities and users to increase wire sections. Same thing will happen to the inverter.

We are concerned about a trimmed wave effect on the output stage of the inverter but nobody has mentioned the inrush charging current of a capacitor that the output stage will also supply on every wave period.

22 hours ago, Mauritius B said:

So you mean that I have to stop using my vacuum cleaner (1800w Triac speed control) and also my hair dryer (1000/2000w just with a half wave trim diode), all my dimmable lightning fixtures or my 2000w full electronic kitchen oven just because I am using a hybrid inverter?

I have revisted my posts and no I never said that. Appliances that chops AC power at relatively low levels  and that is on for 30 minutes to 2 hours a week is ok by me. But chopping 4Kw for 30-40 ‰ of the day matters. 

 

22 hours ago, Mauritius B said:

DC bus capacitors are there exactly to deal with ripple and

Yes, thats what the caps are there for, I agreee. But it shortens its lifetime if the ripple current is significant on a long time. It is in  just the same manner that a car shock absorber takes care when you hit that pothole. But when you consistently hit 10 of them every day, then it dramatically shortens the shocks usefull lifetime. In the same manner is temperature and ripple current a 'shock' to the Electrolic cap. 

I am all for innovation, but I am of the opinion that replacing the geyser element from 4 to 2kw rating is the best solution in the long term.  Its going to cost more than the derating techniques offered here, but it is superrior and inverter/battery/solar friendlier. 

Edited October 24, 20232 yr by BritishRacingGreen

1 hour ago, Mauritius B said:

Again, the concept of time is appearing on the equation because we are talking about energy and not only power.

I think you may not understand how RMS works.  RMS is a technique for time-averaging a changing reading in such a way that it preserves the effective power transmitted.

If you look at your typical AC sine wave, and you measure power at the zero crossing, you will see V=0V, I=0A and P=0W.  If you measure at the peak, you will see V=325V, I=25A and P=8125W.

100 times every second, power is 0W and 100 times every second, power is 8125W.  If you take the RMS average over one second though, then you will see an average power of 4000Wrms.

RMS is an average power over time, whether you take that average over 1 second or 10 minutes does not change it from power to energy.

I want to invite our resident guru @Coulomb to lay my fears at rest regarding using grid-tied inverter technology to power a phase angle chopped geyser element. 

Let us assume we have a grid-tied inverter eg SunSynk  (SS) that is allowed to export , but limited  to export only within the non-essential domain . In this domain their excist a geyser that has a phase angle chopper to derate from 4kW to 2kW  using a 4kW resistive element. So in practice the first 90 degress of each half cycle is wasted , whereupon the last 90 degrees carry the full burden of 4kW power. 

I am of the opinion that the SS cannot handle the vast disruptive changes in power on a per 50hz cycle basis.

My assumption is based on the fact that the close loop response  in order to refresh the inverter phase angle in relation to the grid reference  cannot be changed within the execution of the current cycle. Therefore I assume that whatever the inverter phase angle will be , for the first 90 degrees of the half cycle will the exported power on offer be 'burped' into the Eskom grid, because its not taken by the geyser chopper. Then the  current clamp will tell the inverter to back off  for the next cycles , making things even more complicated.

I would like to have your view on this .

EDIT : if this is true that the chopper will confuse grid-limiting , then its also applicable to 'offgrid' axperts , as the same technique of grid-tied  blending is applicable .

 

 

 

 

 

Edited October 24, 20232 yr by BritishRacingGreen