Immersion Heater - PV electricity

[inie meanie]

I have been using the Triac 10PCV2425 from Crydom. It has a built in snubber and comes as a complete package. I have been using it to run two immersion heaters. When the first is up to temperature I switch a contactor and bring in the second tank. I have just bolted it to the side of a metal case with a small copper shim and a small thin heat sink from a pc. At the very worst time I have only ever felt a very small rise in temperature so I can honestly say that heat is not a problem, and I have been working it hard from just a couple of watts right up to 3 kw continuously throughout the day. The discussed interference has not manifested it's self in any way and I don't think that it is a problem. I am sure that the Emma box uses the same off the shelf component and seem to not employ any extra supression.

 

[suntrap]

Looking at the relative phase of current and voltage gives you the direction of power flow, e.g. if the current waveform is positive while the voltage is negative, power is flowing in one direction; if the current is negative while the voltage is negative, it's the other. In practice it's a bit more complicated because of power factor so you integrate over a cycle (take many rapid samples of voltage and current, multiply them together, and add the results) to work out the net power flow.

I don't have any references; I might be talking rubbish But there is a lot of stuff out there: I found this:

"The general requirement in IEC 1000-3-2 prohibits power control systems that generate low-frequency harmonics. Specifically, systems that generate switching operations less than or equal to 40 times in a half cycle are prohibited in the control of power supplied to heating elements and thermal devices. This relates to equipment that has power current control in the form of SCRs."

And look inside a dimmer pack to see how much ironmongery it contains. As for power dissipation from a triac, well they drop a volt or so which is 13W at 13A RMS.

 

[bradburts]

And look inside a dimmer pack to see how much ironmongery it contains. As for power dissipation from a triac, well they drop a volt or so which is 13W at 13A RMS.

It good to be back to design, patents were a bit boring.
I think that the max voltage drop is at peak voltage and when on. AC is 240V RMS of course so it won't be 13W. Too late to do the maths.
The regulation you quote is interesting. So we need to switch faster than 40 times in a half cycle. That's easily achieved. You trigger on the zero crossing and then fire a (hardware) PWM stream at the duty cycle you need. A little maths to figure out the various pulse stream duty cycles required for a given % power but a small lookup table on the microcontroller would do it. Say you have to switch at >10,000 times within the half cycle, you are still talking of 0.5MHz which should be straight forwards with a PIC or whatever.

inie,
Thanks for reminding us of the Triac 10PCV2425, which proves that phase control is possible for $100 or less.
Does the Crydom switch rapidly within the cycle for a given demand? I would be a little wary of trying to switch the Crydom rapidly as it has an analogue input.
Apart from snubber (which protects against false firing?) do you know if the Crydom has any EMC filtering?
The reason is that the Crydom is still expensive. If this is to be done then best done with a few £ of SCRs built from the ground up rather than a $100 package. Convinient package granted.

 

[suntrap]

I think the dissipation is still 13W, or think of it this way: if you've got a heater dissipating 3kW on 240V that's 12.5W per volt. You could put 240 resistors in series, each dropping a volt, and each would dissipate 12.5W. A triac dropping a volt would do the same thing; in fact it would dissipate more because it is non-linear so will drop a volt over nearly the whole cycle. Yes, it will only do this at full pelt but if you want to be able to drive the heater continuously at high power you will need to provide heatsinking accordingly (the hot water cylinder being an obvious choice).

inie: I think I've already said that the snubber network is not an interference suppression system; it merely prevents false firing of the triac and has no effect on poor power factor and I would suggest little on electromagnetic interference. If it were adequate, you wouldn't have to fill dimmer packs with expensive inductors. I've definitely said that just because you haven't experienced interference doesn't mean that it is not being generated. Those solid-state relays/phase angle controllers do not have interference suppression or power factor correction.

If I ever get round to building such a thing I can see myself going for something based on an Arduino microcontroller as discussed on the openenergymonitor site as well as here. I don't see why this shouldn't be able to calculate the grid current (and power direction) over one cycle and provide the firing pulses for an opto-isolated triac (about £5) for the next cycle. Only one current transformer needed, on a meter tail. This could not be made into a commercial product though.

 

[bradburts]

Yes the hot water tank would be a good heatsink, except when hot! The Crydom operates to 80C and I guess the tank goes to 60C+.
13W would be a fair sized heat sink passively cooled in an airing cupboard, say two hands surface area.
Best to add a thermal trip as well, too easy not too.
Still not sure that the 13W is right, if you look you can find an Rds of 0.03, which is around 5W, say the BT139. Bet you could go lower. Isn't some of the V drop junction effect and so is not lost power?

The Arbuino is a good choice, lots of support and worked examples which is what you need. A PIC may be a couple of £ cheaper. Lots of choice though. I'd go with the brand that has the most/best examples similar to what you're doing. The PIC has a PFC example circuit for example.

Why do you not think that you could make the discussed design as a commercial product?
I am sure that I have delivered worse than that before
I cannot see anything too hard here. Also I have seen products which switch 3KW @£80 or so, they may ignore the rules and just stamp CE, but I hope that its shows all is solveable.

 

[suntrap]

The hot water cylinder should be OK especially if you only put a triac there rather than everything else (such as electrolytic capacitors). You'd attach it at the bottom which would probably remain relatively cold. Difficult to ensure good thermal contact to a curved surface though. If the regulation I quoted applies, you would not be able to use an SCR/Triac to control your immersion heater; you'd have to use a gate turn-off device or a MOSFET which could switch more rapidly than twice a cycle. If you could point me at one of these commercial heater controllers I'd be interested to look. Perhaps they use burst firing, which wouldn't fool an electricity meter (although that reg implies that burst firing isn't allowed either).

 

[suntrap]

First hit for BT139 brings up a data sheet which has a clear graph showing power dissipation for various phase angles. It agrees with what I said: worst case (full tilt) about 17W at 13A, and still nearly 10W at 30 degrees conduction angle. These devices might have a resistance figure but this is additional to the non-linear fixed voltage drop; both factors cause heat dissipation. This is why plug-in timers use relays.

 

[bradburts]

Oops, you're right of course.
I did not check & assumed that power loss was all Rds.
Is this the case with all semis? Worked on a high efficiency DC project once where it seemed that all the hardware ppl could think about was Rds.

Yes a triac would only fire until zero current (the remaing phase) so MOSFET or similar would be need, if the standard applies.
You say burst firing is not allowed, do you mean burst firing in the sense of zero firing where you miss so many cycles?
I was expecting to burst within the half wave as suggested by your quote, using a MOSFET etc.

I don't have links to the commercial product, it was a studio lighting controller, think the brand was Evans. I will see if I can find again.

Anyway, what do you think prevents you from making a product?
You certainly can pick me up quickly on my mistakes & lazyness!

 

[suntrap]

Rds presumably means resistance drain to source, which is for MOSFETS, not triacs. That might explain the anomaly! It is indeed very important for those things. My solar inverter wouldn't be able to get 96% efficiency and a reasonable sized heatsink without nice low resistance FETs (I presume - insulated gate bipolar transistors seem to be another fancy semiconductor used in such applications). They don't have this pesky fixed voltage drop and so behave more like resistors.

Yes, I mean firing for a certain number of complete cycles when talking about burst mode.

I wouldn't be interested in jumping through the regulatory hoops to make a commercial product; besides which, I have a day job!

 

[bradburts]

Yes it was for a buck/boost intelligent psu.
So using a MOSFET with a low Rds in this application would reduce power loss down to 5W or so?

The link I need to find is for a lighting controller & so I doubt that it was burst/zero firing. Would give the viewers a headache! Also I don't think that it was an autotransformer. Probably wrong on that one though.

Still not sure that you have many hoops to jump through, once designed to meet conditions then a design file for CE and lots of testing?
I have experience of companies where the low voltage regulations have been a mystery to those designing 240V, 440V equipment and their managers. Even worked at one. EMC problems, lets hop frequencies! Not suggesting you go that route but I do wonder how much product truely conforms.
BTW, what field do you work in ?

 

[pmcalli]

emma circuit diagram shows a filter which appears to be mounted under the cylon controller. the crydom incorporates a PWM controller which switches at a high rate. this is necessary both to comply with the regs and to make sure that the inverter has enough power storage (hold up) to supply the 3kw load even though it might be generating a few watts on average.

 

[suntrap]

Thanks for clarifying, pmcalli - explains the Crydom's eye-watering cost, and it would be nice if the data sheet mentioned this (and talked about filtering too). Crydom also needs a low-voltage supply, which is irritating if operating it from a system designed to consume minimum power.

Interesting too about the problems of a spiky load for a solar inverter.

Would love to see the Emma circuit diagram!!

bradburts: I supposedly work in electronics but it's all software nowadays...

 

[bradburts]

bradburts: I supposedly work in electronics but it's all software nowadays...

You should ask for a pay rise then

I take it that the cylon controller is the industry standard energy management system (including a lighting controller?) rather than a evil scientist?

Which Crydom range are you refering too?
Did anyone find a FET capable of switching 3KW using AC? I didn't find anything. All I found had Amps dropping off very quickly as rating goes above 80V or so.
I am assuming that there is something (and with a high efficiency) as you can get 240V to 220V convertors fitted to your consummer unit for a few hundred pounds.

 

[suntrap]

You can put FETs in parallel as they self-regulate, so it possible this is done in some cases.

The Crydom range is here.

 

[bradburts]

Thanks.
Looked again, found loads. Must have been selecting gate voltage! What a numpty.
Lowest Rds was 70mOhm.
You sure that the linked Crydom switches at high frequency? I thought it was just a phase controller?

 

[suntrap]

I'm doubting that it can be high frequency, especially when the thermocouple version especially designed for heaters is zero crossing burst firing. pmcalli, which devices did you mean?

 

[Andysimm2000]

Would love to see the Emma circuit diagram!!

I have been on the Emma site and as well as a circuit diagram they show a full colour picture - the only bit that they are keeping under wraps is the software in the PLC.

Andy

 

[suntrap]

Sorry, I can't find the full circuit diagram - could you point me to it, please?

In the picture of the SP15 (smallest one) here there is a Kyotto KC2015AX (sadly apparently unavailable retail). This is definitely not a high frequency controller, and there is a nice big filter. (The black lump called Lambda is a DC power supply.)

In the PDF showing the wiring diagram, things are rather different. There is a Crydom 10PCV2440, similar to the Kyotto, but no filter. Unsure what unit this is, but as it's only a 40A controller and doesn't appear to have a heat sink, I assume it's also the smallest.

 

[bradburts]

(Posted as suntrap posted)
Checked the website & they show a schematic which is pretty much a circuit diagram as all the components seem to be COTS.
Think the Crydom shown is a 10PCV2440, which is a PFC.
There's a AC filter module. Cannot make it out though.
Any thoughts on what the DIN AC filter module is anyone?
Its clearly not designed for cost. Easy to build a DIN rail system but the components cost a packet.

So meeting regulations cannot be 2 hard if all you need is a COTS filter.

 

[suntrap]

Ah sorry, should have read pmcalli's post properly - yes, the high current wires disappear under the control unit, so presumably there is a filter there. I notice in the wiring diagram that you can't have your PV connected to the existing consumer unit - but then if the consumer unit is going to be full of relays it'll need replacing anyway.

 

[scoobiedoo]

Tryed to send a massage to see if you are still suppling your immersion units but get this massage (echase has exceeded their stored private messages quota and cannot accept further messages until they clear some space.)!

 

[pmcalli]

Ah sorry, should have read pmcalli's post properly - yes, the high current wires disappear under the control unit, so presumably there is a filter there. I notice in the wiring diagram that you can't have your PV connected to the existing consumer unit - but then if the consumer unit is going to be full of relays it'll need replacing anyway.

The emma unit uses a PID function in the cylon controller to balance the outputs of the two CT's which measure the power from the solar panels against the power consumed in the house including the variable load immersion heater. If the PV system is wired into a spare way in the consumer unit they recommend that it is seperated by fitting a henley block and a second consumer unit. This is not stricly necessary as there is a sneaky way to install the CT's without adding a second consumer unit. If you have 16mm2 meter tails then you can clip the CT around both the incoming meter tail and the PV wire. As the meter tail is carrying house load - pv gen putting both wires through the CT results in house load -pv +pv = house load. If you have 25mm2 tails then you may need to buy the bigger CT's.

 

[suntrap]

Someone on the openenergymonitor forums gave me that idea until I decided that one CT on the meter tail with a sniff of the voltage to determine the direction of current flow (not to mention give more accurate readings) would be a more elegant solution!

 

[pmcalli]

Accurate power readings are not necessary in a two ct set up as you are only make two things the same. the control function for this mode is very simple to implement. If you only sense the composite current you are trying to reduce the value to zero. current transformers are very inaccurate a low levels and the function is non trival to implement.

 

[suntrap]

Good points, although I was really considering a more sophisticated system based on a microprocessor so that I could log things. Maybe just keep it simple, although it might be difficult to stabilise the loop. It does require the solar feed T&E cable to be split.

 

[pmcalli]

I moved the RCBO for the solar feed to be next to the main incomer in the cs unit and then put the CT around the the incoming meter tail and the solar feed live which are now right next to each other. second CT just goes around the solar feed live all nicely contained in the CS. My CS is a large MK sentry so there is plenty of room.
I am also thinking of using a micro controller type microprocessor and implement the PID function in code there are plenty of code examples for PID loops. I use them at work regularly and finding the right loop values can be done by a simple set of experiments. I am also looking at using an off the shelf PID controller and the crydom unit some of them take CT inputs directly. An expensive solution but only needs wiring up. I did try to see if i could get hold of the controller emma uses but cylon wouldn't supply details just referred me to their partner integrators

 

[inie meanie]

emma circuit diagram shows a filter which appears to be mounted under the cylon controller. the crydom incorporates a PWM controller which switches at a high rate. this is necessary both to comply with the regs and to make sure that the inverter has enough power storage (hold up) to supply the 3kw load even though it might be generating a few watts on average.

Hi guys
Just to let you know the 10PCV2425 is not controlled by PWM. It is a triac and as such, as soon as fired will continue to conduct until zero cross over. the EMMA uses this unit and therefore definitely does not use pwm. This has been confirm to me today by an application engineer at Crydom.

 

[suntrap]

Good stuff - please keep us posted!

One of the things that baffles me is the apparent lack of heatsink in the EMMA unit. I'd have thought the phase controller would have to dissipate about 15W going full tilt.

Edit: thanks inie meanie - your post crossed mine.

 

[pmcalli]

Data sheet I had on the crydom said it incorporated a proportional power control which I incorrectly interpreted as being a PWM. Sorry for the bum steer. I just downloaded the full data sheet from USA site and it doesn't make pleasant reading. voltage drop is specd at 1.6V which I make 20W dissipation at full load. Looking at EMMA pics it appears to be mounted on a DIN rail. The EMMA case looks like it is stainless steel. using the derating curves from the data sheet and some guesswork on the watts per degree C of such a mounting it will run hot but within max temps. This might explain why the EMMA unit has vents in the case top and bottom of the crydom. If it was mounted directly on the metalwork it would run stone cold. Looks like they could do with some lessons on thermal design. I am surprised that the solar inverters have enough hold up to support simple phase control. The data sheet specifies a minimum current of .15 A but the graphs show it going to zero have to assume the .15 minimum means that this is the lowest controllable current and after that it goes straight off meaning lowest usable power control is 36 W not really a problem.

 

[echase]

Tryed to send a massage to see if you are still suppling your immersion units but get this massage (echase has exceeded their stored private messages quota and cannot accept further messages until they clear some space.)!

Sorry, have now deleted some.

 

[inie meanie]

Data sheet I had on the crydom said it incorporated a proportional power control which I incorrectly interpreted as being a PWM. Sorry for the bum steer. I just downloaded the full data sheet from USA site and it doesn't make pleasant reading. voltage drop is specd at 1.6V which I make 20W dissipation at full load. Looking at EMMA pics it appears to be mounted on a DIN rail. The EMMA case looks like it is stainless steel. using the derating curves from the data sheet and some guesswork on the watts per degree C of such a mounting it will run hot but within max temps. This might explain why the EMMA unit has vents in the case top and bottom of the crydom. If it was mounted directly on the metalwork it would run stone cold. Looks like they could do with some lessons on thermal design. I am surprised that the solar inverters have enough hold up to support simple phase control. The data sheet specifies a minimum current of .15 A but the graphs show it going to zero have to assume the .15 minimum means that this is the lowest controllable current and after that it goes straight off meaning lowest usable power control is 36 W not really a problem.

I have been using this module and have it bolted to a sealed case with only a slight perceptible rise in temperature. I think the EMMA uses a heat sink on the rear or the triac that then clips to the din rail. I have been driving it hard as I switch between immersion heaters on different tanks automatically via a contactor change over. I have monitored current down to 10 w when the control voltage is 2.4 vdc

 

[bradburts]

You can tune a PID using:
https://controls.engin.umich.edu/wiki/index.php/PIDTuningClassical
Usually enough just to have a PI system.
I think that in this case you should be careful how you tune though. I would be more interested in not importing electricity than having a control system which does not oscillate. The sun goes in & your immersion power should drop sharply is more important than slowly integrating to a fine balence.
I would have thought that a simplier control system would be more relevant. Suppose you could use an integral to approach ideal but when the sun goes in the integral will have you importing with a true PID.
Perhaps EMMA use a PID because thats what comes with the DIN rail controller.
Is CT accuracy much of an issue? I would be happy using all but 200W or 100W as a design aim, thats quite a small percentage 'loss' after all. 200W is 1 Amp which should be easy enough to measure?
I have not used CTs before, always used shunts. What range of accuracies would you expect?

 

[inie meanie]

You can tune a PID using:
https://controls.engin.umich.edu/wiki/index.php/PIDTuningClassical
Usually enough just to have a PI system.
I think that in this case you should be careful how you tune though. I would be more interested in not importing electricity than having a control system which does not oscillate. The sun goes in & your immersion power should drop sharply is more important than slowly integrating to a fine balence.
I would have thought that a simplier control system would be more relevant. Suppose you could use an integral to approach ideal but when the sun goes in the integral will have you importing with a true PID.
Perhaps EMMA use a PID because thats what comes with the DIN rail controller.
Is CT accuracy much of an issue? I would be happy using all but 200W or 100W as a design aim, thats quite a small percentage 'loss' after all. 200W is 1 Amp which should be easy enough to measure?
I have not used CTs before, always used shunts. What range of accuracies would you expect?

I have calibrated my system which is totally analogue to export slightly all the time. I too did not want any import to be used as I seem to have plenty of available power. The CTs do have a slight curve at the lower end but I have managed to just export 10 or so watts at these points but then I have to export a 100 or so when power is in the 1 to 2 kw range and I was quite happy with that. The control function is only proportional as I would need to go digital to introduce and integral action. I think I will be investigating a digital setup in the future, now I have everything running really well, I can relax a bit. :- )

 

[pmcalli]

You can tune a PID using:
https://controls.engin.umich.edu/wiki/index.php/PIDTuningClassical
Usually enough just to have a PI system.
I think that in this case you should be careful how you tune though. I would be more interested in not importing electricity than having a control system which does not oscillate. The sun goes in & your immersion power should drop sharply is more important than slowly integrating to a fine balence.
I would have thought that a simplier control system would be more relevant. Suppose you could use an integral to approach ideal but when the sun goes in the integral will have you importing with a true PID.
Perhaps EMMA use a PID because thats what comes with the DIN rail controller.
Is CT accuracy much of an issue? I would be happy using all but 200W or 100W as a design aim, thats quite a small percentage 'loss' after all. 200W is 1 Amp which should be easy enough to measure?
I have not used CTs before, always used shunts. What range of accuracies would you expect?

I agree PI with D set to zero will be fine. if you don't use the I term there will be an error to the set point. This error could be above or below the set point depending on what direction you have aproached from so its important to use the I term even if its set very low to give a rapid response. We do not know if EMMA uses a PID controller its just a guess. the maker of the controller only releases the information to integration partners.
CT's are usually designed for a current range but this can be adjusted by modifying the so called burden resistor ( load ). if the burden resistor is high you get a higher o/p voltage which helps with the low values but it becomes non linear at high currents. For the immersion requirement I would reccomend a high burden value as this gives the resoultion you want at low values. if you are using a ct on both house and pv the high current non linearity is the same for both and so has no effect if you use a PI controller. you need to be careful about o/p voltage at high currents on the house CT and i would suggest that clamp diodes or a zener is used to ensure that the i/p to the controller is not exceeded. the clamp should be fitted for saftey reasons anyway and some CT's have them built in.

 

[pmcalli]

I have calibrated my system which is totally analogue to export slightly all the time. I too did not want any import to be used as I seem to have plenty of available power. The CTs do have a slight curve at the lower end but I have managed to just export 10 or so watts at these points but then I have to export a 100 or so when power is in the 1 to 2 kw range and I was quite happy with that. The control function is only proportional as I would need to go digital to introduce and integral action. I think I will be investigating a digital setup in the future, now I have everything running really well, I can relax a bit. :- )

You don't need to go digital to introduce the I term. A simple change to your amplifier to put a second order term in with give you a PI loop. If you have used an op amp as the gain element the feedback resistor just needs replacing with resistor+ capacitor so that it acts like and integrator with the slew rate limited by the resistor.

 

[suntrap]

More good stuff. pmcalli, have you seen the openenergy monitor forums? A lot of what you say, such as about clamping diodes and burden resistor selection, is discussed there. inie: I presume you're keeping your circuit diagram to yourself. :smiley2:

 

[inie meanie]

I have been using the AT20B10 coils the same as the EMMA, well no point in reinventing the wheel The good thing is they give a 0 to 10 vdc output for the specified range. No worries about a peek voltage. A bit more expensive but it makes things very easy. So far the slope I have achieved is so close it's not worth worring about the off set. As stated Integral can cause an over shoot and I dont want any of that :smiley2:

 

[pmcalli]

More good stuff. pmcalli, have you seen the openenergy monitor forums? A lot of what you say, such as about clamping diodes and burden resistor selection, is discussed there. inie: I presume you're keeping your circuit diagram to yourself. :smiley2:

just had a look very interesting. The only fear I would have with some of the circuits suggested is there is very little protection and no filtering none of them would pass CE. As to circuits I have been experimenting for a while with various options from expensive like the EMMA to cheap using entirely analogue control loop and discrete thyristor. My conclusion based on the prototypes and entry's on this forum is to be lazy and go for the AT20B10/ AT50B10 sensors, crydom phase controller, i/p filter, packaged power supply and an off the shelf PID controller ( not yet selected any suggestions welcome ). This is very expensive but I don't have the hassle of designing and building a pcb. The analogue cts were std stuff. I used precision rectifiers for the ct interface and then a simple type 2 op amp integrator

 

[pmcalli]

I have been using the AT20B10 coils the same as the EMMA, well no point in reinventing the wheel The good thing is they give a 0 to 10 vdc output for the specified range. No worries about a peek voltage. A bit more expensive but it makes things very easy. So far the slope I have achieved is so close it's not worth worring about the off set. As stated Integral can cause an over shoot and I dont want any of that :smiley2:

EMMA uses AT20B10 for the PV sense and AT50B10 for the house sense. these sensors are only rated for continuous overload of 120% so peak voltage is a problem

 

[pmcalli]

I have been using the AT20B10 coils the same as the EMMA, well no point in reinventing the wheel The good thing is they give a 0 to 10 vdc output for the specified range. No worries about a peek voltage. A bit more expensive but it makes things very easy. So far the slope I have achieved is so close it's not worth worring about the off set. As stated Integral can cause an over shoot and I dont want any of that <img src="http://www.electriciansforums.net/images/smilies/smiley2.gif" alt="" title="Smiley2" smilieid="329" class="inlineimg" border="0">

EMMA uses AT20B10 for the PV sense and AT50B10 for the house sense. these sensors are only rated for continuous overload of 120%