 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Electronics : High Power Solid State Inverter Relay
Page 1 of 2   |
|||||
| Author | Message | ||||
| Solar Mike Guru  Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Next item in the do do list for my inverter project, is a solid state electronic switch to replace the various gigavac high voltage relays. These somewhat expensive relays are great and efficient but I require something more generic that can be used elsewhere in other projects with differing currents and voltages; so some sort of very low resistance mosfet switch is required, with limited functional intelligence. The switch must handle anticipated load currents approaching 300 amps for short periods and perhaps average 100a for extended durations, voltages are for a 50 volt bank but must be able to be used at other volages with minimal modification. I want the switch to be reasonably intelligent in that it should perform the following actions: Sense battery limit voltages and switch off if outside set limits, especially low voltage. Sense heatsink temperature and control fan speed propertionally. Allow automatic current limited pre-charge delay for charging the humongous caps in the inverter at turn on. Opto isolated control inputs. Easy to change settings. High side operation. For the prototype, I have come up with a pcb layout for the high power components and a seperate pcb for the controller, note on the control schematic, connectors are miniture spring loaded board sockets, they do not plug into each other directly. As I have a quite a few picaxe chips lying around and they are simple to program, have used a picaxe 14M2 here for the cpu, this gives me another pcb with generic functionality that can be re-used elsewhere. Main power cct contains the paralled IRFP4468 mosfets and an auxillary mosfet acting as a current limited device, along with a basic 12 volt pre-regulator. Mosfets are switched by high voltage optimos isolated relay chips, no fancy high speed high current drivers are required here; different mosfets could be used depending on expected current load; probably I will only require 5-6, but have allowed for 8. So I threw this prototype design together without any prior testing and am about to design pcb's:  Controller:  If it works ok, will perhaps add additional load current sensing to the control board (as cpu has unused pins) using a 500 amp DC through hole device to fit over one of the copper busbars, this will allow shutting down the inverter at too high a current load. Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
This could also be done reasonably economically with a 100 amp dc rated circuit breaker fitted with a shunt trip coil. The shunt trip coil could be energized from a commercial digital voltmeter that has a pair of programmable voltage threshold alarm relays, for under and over voltage disconnect. A circuit breaker is going to be a lot more economical than Gigavac relays, and this way you also get the overcurrent protection, and voltage monitoring, and manual disconnect, without really requiring much else. Its also simple with very little in it that can go wrong. Over temperature could also flick the circuit breaker off with something as simple as a bimetallic thermal switch bolted onto your inverter heatsink. The above circuit really needs some pretty robust transient voltage suppression. Breaking 100 amps or more very fast, is likely to produce a pretty fearsome mosfet killing voltage spike. Cheers, Tony. |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Thanks Tony for your input concerning the transient suppression, I have simplified the main switch circuit to incorporate a 10 sec on time for pre-charge of inverter capacitors and an off time of 50ms. As transients are caused by fast di/dt transitions on wiring inductances etc, by slowing down the turn off of the mosfets so they remain in partial conducting state for a longer period, any transient spikes should disappear, also placed a couple of transorb devices across the outputs.  I looked at purchasing dc circuit breaker with shunt coil, would cost over $500 here in NZ for a 250a breaker, plus another $80 for programmable voltage switch etc; as I require a number of these I would rather make something for $100 and be more versatile. Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Great stuff there Mike. Slowing it down is an excellent idea, it will help hugely. Large dc circuit breakers bought new over the counter can be amazingly expensive, but occasionally something turns up on e-bay if you are both patient and lucky. I only have a small system here, 100v 6Kwh (30 cells) Lithium battery. I bought a Chinese e-bay two pole 63 amp breaker, 10KA rated breaking current, vacuum contacts (similar to the one below)for about $30. http://www.ebay.com.au/itm/32A-Schneider-Electric-C65H-DC-Miniature-Circuit-Breaker-2P-250V-10KA-Vacuum-/201716234184?hash=item2ef73977c8:g:NKMAAOSwAL9Uj5WJ The shunt trip coil was a one off freak, also e-bay for around $16 plus postage. This sits in beside the two pole breaker so it looks like a three pole breaker. And the programmable digital voltmeter $38 US plus postage (not e-bay). http://www.lightobject.com/Programmable-4-Digit-Red-LED-ACDC-Volt-Meter-with-Dual-Control-Good-for-HHO-System-P408.aspx. This displays voltage 00.00v to 99.99v with two alarm relays that have independent programmable pull in and drop out voltages in 10mV steps. The only thing I added to that was a small push button, a relay, and 30 ohm 200 watt metal clad resistor for initial pre-charging of the capacitors in the inverter before closing the breaker. It took some time to source all the parts at a friendly price, but it adds a further final layer of protection to the battery quite independently of the usual controls and shutdowns in the BMS, inverter, and charger. It all works wonderfully well as last resort battery protection, and am quite pleased with how it has turned out. As far as I know, nobody else has done it this way. Cheers, Tony. |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Yes that looks like a good meter, I have used the JLD404 in the past which has the DC voltage, current and amp hour programmable set points. The large and expensive Lifepo4 battery bank certainly needs multiple layers of low voltage failure redundancy built into the system and a programmable meter with relay outputs would help here. The LAA127 optimos switch pairs used here in the Inverter Switch have 250 volt ratings, I'm also using them switched by a mux to sample each cells voltage to a single high voltage common mode rejection op amp for bank cell monitoring. What I would really like is to purchase a postage stamp sized switch power supply that can run off 40 - 80 volts DC and have an isolated 12 volt outputs at 200 ma or so, to replace the clunky 12 volt pre-regulator. I could design one, but no time...so if anyone knows of such a device? Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Great minds think alike Mike. That programmable voltmeter is a real ripper and highly recommended, although initial programming to set up all the options is a bit fiddly. But that only needs to be done once. It would work great with one or two Gigavac relays. I am using something very similar to the LAA127, a PVA3354 which has a 350 volt rating, its also faster, but the on resistance is not nearly so good as the LAA127. Each Lithium cell has a relay and two small fuses mounted on a small circuit board bolted between the battery terminals to select the cell to be monitored. I am using four of those PVA3354 mos switches in the "flying capacitor" configuration to directly drive the analog to digital converter input of a microcontroller. There is a 10uF capacitor that is alternately connected to the the cell being monitored, and the A/D input to the microcontroller directly, so that voltage measurement at the cell is fully floating, and is ground referenced when switched to the A/D input. A non overlapping clock switches the four PVA3354 mos switches in pairs. It works great with battery cells up to at least 300v above ground, with only hundreds of microvolts of dc error. That error is far less than one bit of the A/D so it can be ignored. I prefer to use relays for selecting the cell, because I am also charging or discharging individual cells for balancing, determined by software. There are a great many small postage stamp sized isolated switching power supplies around that are made to operate over the worldwide mains voltage range. The nominal mains supply in Japan is only 100v and may go as low as 85v. These invariably use a small flyback supply which you will find starts up and runs at a surprisingly low dc input voltage, provided the supply is not heavily loaded. Most come to life at about 30v to 35v dc input and would probably work fine in a 48v system. I am using a lot of these now for powering all kinds of things from dc inputs over about 50v up to about 350v Plenty of these on e-bay. If you get something rated at about 12v 1 amp I bet it will start up and run fine with only 200mA load. http://www.ebay.com.au/itm/New-AC-DC-Power-Supply-Buck-Converter-Step-Down-Module-5V-12V-3-3V-9V-24V-500mA-/262807265342 ?var=&hash=item3d3088a03e:m:mz_hHYGfJCLVc24zslyFqoA Cheers, Tony. |
||||
| noneyabussiness Guru Joined: 31/07/2017 Location: AustraliaPosts: 513 |
1PCS HLK-PM01 AC-DC 220V to 5V mini power supply module,intelligent household switch power supply module http://s.aliexpress.com/UN7RfIZf (from AliExpress Android) Something like this but 12v version, which there is plenty of. I think it works !! |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
There are plenty of these amazing postage stamp sized power supplies around, and so very cheap. They all seem to be copies of each other in the best Chinese tradition. They are all fully isolated input to output, and have excellent output voltage regulation. One thing, when running at very low input voltage, these are flyback supplies and draw quite high peak current, especially at very reduced input voltage. The energy stored in the input filter capacitor (after a bridge rectifier) is proportional to dc voltage squared, so at very low dc input voltage this 400 volt rated capacitor is not going to be very effective. The whole thing needs a reasonably low impedance source of power, which is usually not a problem. If it is, I have sometimes replaced a 47uF 400 volt electrolytic with something with higher capacitance but lower rated voltage. Probably not required, but its something to keep in mind if you are pushing the limits. I now sometimes piggy back these direct onto circuit boards I get made, very simple, convenient and low cost. Cheers, Tony. |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Yes I have looked at those, a lot of these products have a main switch chip that has an internal low voltage lockout that stops the thing working at voltages less than 80 volts, so they dont work, depending on the type of device they use. I have one similar to Warpspeeds suggestion in a previous post and it doesnt work at 48 volts.  Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
I have used quite a few different ones (5v, 12v, and 15v) and all of them would start up with about 35volts, and once running would continue to run with slightly lower input voltages. Not saying you are wrong, but my experience has been different. I will test a few right now, I have quite a collection here, and come back with some results and the origin of the part. Cheers, Tony. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
What I tested is a 5 watt supply advertised as 12v 450mA (from e-bay). With no load at all on the output, it starts up at 35 volts. Either slowly rising or quickly switched on makes no difference. Once running it keeps going down to 22 volts input. With 220 ohm load it starts up at 35 volts and continues to run down to 22v input no real change from the unloaded condition. With 120 ohm load, startup rises to 58 volts, and runs down to 23 volts input. With 68 ohm load, startup requires 60v and runs down to 29 volts. I have mainly use these for powering cmos circuits, mosfet floating gate drivers, op amps and such, where the load is absolutely minimal. I have also used them to power LED panel meters, but have not tried to do that from very low input voltages. Anyhow, its definitely worth looking at for a 48v system for some very lightly loaded applications. Cheers, Tony. |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Hmm interesting, possibly the one I tried had too much load on it, prob a 12 volt fan, so it wouldnt start up, will try another type... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
A "stalled" fan motor would probably be about the worst kind of load for starting it up into. Try it first with no load at all, and see what it does. This 5 watt one is only a baby, they are available in higher power ratings. Another thought. You may be able to get away with something sneaky. Switch the fan load on only after the supply comes fully up by means of a series mosfet. Have the gate voltage very slowly rise with some sufficient time delay. Big series resistor and electrolytic on the gate maybe ? A super soft fan startup. Starting up seems to be very critical. Once the beast is actually running, it appears to be able to support a much lower input voltage under load before it drops out from starvation. Cheers, Tony. |
||||
| noneyabussiness Guru Joined: 31/07/2017 Location: AustraliaPosts: 513 |
If it is certainly a flyback design then they commonly have a aux winding to boostrap itself. .. hence once started usually all good. I was going to suggest if you can gain access to the main pwm chip (uc3842 or equivalent usually ) u can change the startup resistance to a lower value. It will no longer be safe for high voltages but will respond at a lower voltage. I think it works !! |
||||
| noneyabussiness Guru Joined: 31/07/2017 Location: AustraliaPosts: 513 |
http://danyk.cz/iz_osc_en.html Just an example, Change the 220k to maybe 75k or there abouts. When i do a diy flyback at these voltages i usually use a 10k as startup... I think it works !! |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
I have not reverse engineered any of these supplies, but a flyback topology is the only practical means to get such a very wide input voltage range. Its the startup method, especially under load that seems to be the limiting factor for low input voltages. Agree that its quite probable that there is a series resistor in the startup circuit that could be reduced in value for much better results with low input voltages. There seems to be a large sudden change in start up characteristic between very low load and medium to higher load. That suggests the self powering bootstrap has some threshold where it fails to sustain the system after an initial try at powering up. My own system here uses 100v nominal battery, so I have had none of these problems myself, so have not investigated too deeply into it. But for 48v systems,I feel there could be a lot to gain with probably a very simple modification to the start up circuit of these Chinese supplies. Cheers, Tony. |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Exactly, got out the magnifier and checked out the driver chip on my non starting PSU board, it is an AP8012, google it and voila AP8012  There is a lot of electronics inside that little chip, looks like the startup resistor R1 could be modified to allow it to run at a lower voltage. The PCB is 5cm x 2.5cm so I could mount it on my main mosfet PCB to replace the clunky 12 volt supply components. Cool will do some investigation, it may even be worth while creating a pcb using that chip with + 5v and 2 isolated 12 volt rails, as I really need the 3 supplies, could use that in all sorts of applications. Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
R1, C1, and D1 appear to be a snubber network across the primary winding, as D1 is pointing the opposite way. Bootstrap power for U1 comes via D2 and R3, but if the output is very heavily loaded that will not work very well during initial startup. From that schematic, the startup system (whatever it is) would appear to be completely contained within the U1 chip, so we are screwed. There are not too many choices, but a protective zener across CD3 and an added external startup resistor from CD3 to the incoming positive on CD2 might do the trick. That would supply some additional initial power to U1, and it may only need a very few milliamps to be effective. Cheers, Tony. |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Yes of course R1 C1 it is a snubber.... According to the data sheet CD3 on pin 4 must reach 15.5 volts in order to start, supplied via power on pin 5, once it starts then the feedback winding comes into play. I will try the extra resistor, zener to CD2.... |
||||
| noneyabussiness Guru Joined: 31/07/2017 Location: AustraliaPosts: 513 |
Actually if that the case then i would just start with adjusting R3 lower, at 48v (or there abouts) its getting over its uvlo threshold. I would bet that because the aux winding low for higher voltage, its not getting enough juice to maintain uvlo.. so reducing R3 would mean more current available for chip, regulation still with original opto etc. Just a thought. .. I think it works !! |
||||
| Page 1 of 2 |
|||||
 Print this page |
