 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 : time to build a replacement inverter
  Page 7 of 8   |
|||||
| Author | Message | ||||
poida Guru  Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
I don't think so. In one of the above photos you can see the copper of the Drain revealed. The Drain only goes down to about the level of the heatsink anyway. And the silicon chip will have heat very quickly conducted away via the copper Drain plate. Thermal conductivity of Copper, etc... wronger than a phone book full of wrong phone numbers |
||||
| tinyt Guru Joined: 12/11/2017 Location: United StatesPosts: 438 |
I don't know if excessive ground bounce is happening in your setup. The idea of CN1 connection is to minimize it. .jpg) My theory is any switching current drawn from the Jx pins 1 and 2 by the nanoverter will raise J1 pins 9 and 10 from Jx pins 9 and 10. The effect is 0 volt signal from the IR21844 will not be exactly 0 with respect to the power board ground. CN1 connection will minimize it, I think. |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
TinyT I will do this in the next attempt. The changes made for the next attempt: no ferrite beads on Drains Add the thick ground lead Test all remaining parts on board. - all 16 gate diodes,16 10R gate resistors, 4 snubbers (47R and 103 caps) - other resistors - the 2 UF4007 were tested and proven good, but I replaced them with FR308 which were available. - carefully tested the 4 20R gate drive input resistors. Been tripped up previously by them Replaced: - 18V zener, TIP35, 120R 5W, all blown - all 16 MOSFETS, all blown - all 4 TIP 41, and 4 TIP42, all blown Test and verify approx. 18V across emitters of TIP41/42 as seen on the evaluation test board. I only see 16V. Why? I see 18V across the zener. Volts across the 120R 5W is 0.4V on standby, 10V when running. This is about 0.15W on standby and 4W when running. It works on the bench, pulling 0.29A at 52V into the test toroid/choke combo. This is about the same as the other boards I have when run at 52V. wronger than a phone book full of wrong phone numbers |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
>> Test and verify approx. 18V across emitters of TIP41/42 as seen on the evaluation test board. I only see 16V. Why? >>I see 18V across the zener. They are acting as emitter followers, so cannot pull up to the rail, 16v is fine. Cheers Mike |
||||
| tinyt Guru Joined: 12/11/2017 Location: United StatesPosts: 438 |
if the TIP41/42 emitter followers cannot pull all the way up to the positive rail, maybe they also cannot pull all the way down to the gnd rail. |
||||
mackoffgrid Guru Joined: 13/03/2017 Location: AustraliaPosts: 460 |
Yes, to both. The emitter followers can't get to either rail. Close enough it seems. It has surprised me that the low has been low enough. Driver IC's go to some trouble to get close to the ground rail. Mikes, Wiseguy, solution of having a negative supply is a nice improvement on that. whether it's the difference, given the number of working machines??? Cheers Andrew |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Here is a good explanation why this driver is good to use:   Good value for "R" is 10 ohms to isolate the driver chip supply from the output pulses, with the LHS cap 100uf and the RHS cap 1 to 10uf depending on the number of mosfets being driven, ideally one totem pole driver per mosfet. Cheers Mike |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
And so I need to take the next step. Repeating the same thing and expecting a different result is insane. What to do.. I propose I assemble a simple test, located NOT in the house but the workshop. It is to use a 1.5kW toroid, with a primary suitable for 48V supply, a known good choke and no load. I will power it for a week, or two, with no load. The power board will have the ferrite beads removed, all components proven to be good. It will never need a fan. It will just sit there and hum a boring 50Hz tune for days. This will show if there is a problem with the gate drive, IR21844 or nano1, or power supply, etc. No thermal issues will become relevant since there is no load I expect approx. 13W to be consumed by the toroid, choke and power board. Let's just prove the concept first. How do I power it in my circumstance? I have a spare 200Ah 48 battery that is unused and so I will use this, and charge it via the existing solar system. Use a diode to permit the test battery to remain undischarged by the house inverter. No probs, just need a decent sized diode, good for 20 A or something. 24 hrs at let's say 20W will be 0.4A x 24 = 9.6Ah per day. I would like to think the following day's solar power charge would be small, easily handled by a 35A full bridge rectifier, mounted on a heat sink. What do other people here think? wronger than a phone book full of wrong phone numbers |
||||
| arthur8 Regular Member  Joined: 08/05/2019 Location: BrazilPosts: 69 |
I think i will be doing this on my inverter, also. Also, i suggest to everyone this little thing: https://www.aliexpress.com/item/63A-230V-Din-rail-adjustable-over-and-under-voltage-protective-device-protector-relay-with-over-current/32863099171.html?spm=a2g0s.9042311.0.0.2742b90asMAbDi It already saved my house equipments when the power gone out and my cheap on grid inverters started to feed the AC line with 290V AC. Had them monitored by a camera (not the best way to monitor things, i know) and discovered why the home turned off. The power was ~220v and then fast rised up to ~290v AC and the little device turned everything off because of overvoltage. Keep up the good work! |
||||
| arthur8 Regular Member Joined: 08/05/2019 Location: BrazilPosts: 69 |
Poida, I will be using a setup with multiple transformers and i'm thinking on a way to soft start the second and the third transformer groups. Do you maybe know what Victron uses to turns on the second transformer? |
||||
renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678 |
Good plan, build it stock standard. Feeding it with a battery bank that is not connected to other things will rule any other variable out. I probably asked this already, but have you got a tuning cap on the torroid, if so has it been tuned to 75hz? Cheers Caveman Mark Off grid eastern Melb |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
Sorry, I don't know how it's done. I suspect it might use 2 relays. When I tore it down I saw many relays, AC switching type. So maybe one relay (#1) to switch in the primary of the second unloaded toroid. The second relay (#2) to switch in the secondary of this second toroid, to become parallel with the first toroid. It's possible to have the switching occur at zero crossing time. So it's close #1, wait a few cycles, then close #2 for when the second one is needed. When power levels drop and the second is no longer justified, open relay #2 and a few cycles later, open relay #1. With syncronised inverter boards, it might be a lot easier, just needing relay #2. I would like to know how Victron do it. I have thrown out much of the guts of the dead 3kVA Multiplus inverter/charger. wronger than a phone book full of wrong phone numbers |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Poida, do you have a logic analyser, or four channel digital storage oscilloscope you can set up to monitor all four gate drive waveforms. If/when it blows, use the resulting current spike as the trigger point to freeze the stored display. You should then be able to see events immediately leading up to, and during "the event". Four reasonably fast opto isolators will be required to fully isolate the sources from the monitoring side. A bit of forensic evidence at the actual point of death might be quite revealing. Cheers, Tony. |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
I have chosen to perform the following as a first test. I have rebuilt the board with all new semiconductors, verified all resistors and gate diodes, new caps for the totem pole pairs. I verified correct and good operation of the nanoverter board with the test power board. Now I have set up the above boards with a 48uH ferrite choke and 1.5kW toroid with a 27V AC RMS primary. DC supply is via a separate 200Ah 48 battery, isolated to some degree from the home solar power. (I charge the 200Ah battery via a single diode and 4mm2 cable, about 3 meters long. This places a big buffer between any energy spike sources.) DC supply to the inverter test rig is via 1mm2 wire, a 5A fuse and a switch. The aim is to run the inverter with zero load for a week and see what happens. Idle DC current is about 0.4A so the 1mm2 cable also isolates the inverter power board well from energy spikes. The power board has all 6 10,000uF bulk caps fitted so the system would be expected to be running with a quite stable DC voltage in the bulk caps. I started this 2pm Saturday, it's 12:30 Monday, so nearly 46 hrs completed successfully. If it blows with zero load, it means the failure reason is not thermal or high current caused. The reason will be due to poor gate drive timing, shoot through, too low or high gate drive voltage, too slow/fast gate voltage swing and a heap of other things. I like the idea of recording all 4 gate drive voltages and triggering storage on a failure. The DSO can be driven via USB commands, including stopping recording in horizonal "roll" mode. There will be enough data there to see the failure. Since MOSFETs fail with Drain - Source going short circuit, I would build a circuit that takes the voltage difference of Drain to Source and produces a 5V logic signal when the voltage drops to zero on any of the 4 legs of the bridge. This 5V will trigger the output of the "Stop" command to the DSO and then command it to dump the entire data stream to file. wronger than a phone book full of wrong phone numbers |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
I have had another look at the question and my view is now different. The Vicron has two toroids. Both are bespoke units (that is, off the shelf but maybe custom specification). They have one secondary winding, 240V AC RMS. They have two primary windings, 33V AC RMS. Two windings? Yes, two separate secondaries, maybe 2.3mm diameter single enameled copper. This is 4.15mm2 per winding. The toroid is 0.437V AC RMS/turn Length per turn is about 200mm My idea how Victron switches in the second toroid is that when running with low power loads, one winding of both toroids is connected to the MOSFET bridge. Then when higher loads are present, the second winding is switched in. I'm OK if this idea makes no sense. It is a possible setup given the 2 separate primary windings. low power through 1 x 2.3mm diameter primary high power through 2 x 2.3mm diameter primary At all times both toroids are delivering power. Time to do a little math.. wronger than a phone book full of wrong phone numbers |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
I proved the rebuilt board for about 7 days continuous idle. It was connected to the test bench inverter’s choke and toroid and cap. I had it running through a 5 amp fuse, and just forgot about it for a week. I had a thought, find some fact that is different in the build of inverter #1, which just keeps on running no problems with it's 16 HY4008 FET board made in China, controlled by the EGS02. ..and inv #2 which has blown Chinese 16 HY4008 FET boards twice and 16 HY4008 Mad power boards twice. What is different.. I finally got to the stage of testing the resonant frequency of the load the inverter boards see. Inv #1, that keeps on running: 50Hz Inv #2, busted 4 boards (and counting): 17Hz Test inverter on the bench was 35Hz How to test to find resonance: Need a function generator, with a 50 Ohm output. A DVM that does true RMS. (maybe TRMS is not needed, dunno) 50 R resistor. First, hook up the primary winding and choke as normal. Have the capacitor on the secondary in place and also have the 12 AC transformer in place. It’s just like you are going to connect the inverter board to it. But you don’t. Now for the function generator, put the 50 Ohm resistor across the fun gen output, to load it properly. Then take two wires from the fun gen output and run them into the DVM. I set my fun gen to output 2V RMS and see about 2V RMS on the DVM, from about 15Hz up to 100Hz. This is just measuring the voltage across the 50R resistor at this stage. Then take another pair of wires, just as you lead them to the DVM, but this time connect them to the choke and primary winding. Change the frequency and note the AC voltage across the 50R resistor. The resonant freq. is where the DVM shows the highest voltage. You probably won’t see 2V RMS any more once you connect the choke/primary. I used the oscilloscope instead of the DVM. Not going to change anything but it lets me see the phase change as the freq goes from below resonance to above resonance. It’s clearer to see the point of max resonance, where there is no phase difference. Anyway, seems I was an idiot and ran the inverter at 17Hz which is a bad thing. I used the 10uF cap that comes out of the 3kW Aerosharps. To get to 75Hz, I changed it to a 2.2uF cap. Trial and error. Maybe this $6 purchase from Jaycar is all that’s needed to prevent blowups. RenewableMark wants his builds to have a freq of 75Hz. I have no idea why inv #1 DOES NOT DIE with it’s 50Hz resonant freq. The watt meter on it shows about 1.4MWhr and about 7kWhr per day. No blowups. I must have held my tongue at the exact correct angle when building that one. wronger than a phone book full of wrong phone numbers |
||||
| renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678 |
Perhaps Tony might elaborate here on how he does the test I won't even try and explain the 75Hz tune, I think I understand it, but wouldn't explain it properly. Cheers Caveman Mark Off grid eastern Melb |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
The way I do this is to use a sine wave function generator to feed the toroid secondary through about a 10k series resistor. There should be nothing at all connected to the primary. I then monitor the sine wave voltage directly across the secondary with an oscilloscope. At the parallel resonant frequency of the secondary winding, a definite peaking of the output voltage will be seen. Fit as much capacitance as may be required to bring this resonant peak down as close to 75Hz as you can get. Ideally within about +/- 1Hz is usually possible. A digital frequency counter may be required to get sufficient frequency accuracy. Without any deliberate capacitance at all connected across the secondary, the secondary will still have a self resonance, usually at a very few Khz. Under light inverter loading, harmonics of 50Hz (harmonic distortion) can excite this resonance producing stationary "wobbles" on the waveform. These can range from barely perceptible, or pretty horrible, depending on your luck... The way to fix this is to add some deliberate capacitance to lower this resonance, but we need to be very careful how we do this, or we can introduce other problems. As we get down closer to 50Hz resonance the wobbles disappear, but we can get a very large undesirable buildup of energy at 50Hz which can cause the voltage feedback to lose control of the inverter output voltage. Resonance below 50Hz will increase the idling current in a pretty useless way, so we must avoid that. Anyhow, there is a sweet spot right at the inverter frequency times 1.5, or 75Hz for a 50Hz inverter. There is a curious effect right at this specific frequency. Any tendency towards an uncontrolled resonant buildup of energy is self damped, because it will be exactly out of phase with the following cycle. We gain two advantages. The wobbles disappear, and we can get increased attenuation of the 23Khz pwm switching frequency. Idling power does not seem to be altered either, so it usually produces an excellent clean output waveform under all but the very worst non linear inverter loading conditions. Cheers, Tony. |
||||
Ralph2k6 Senior Member  Joined: 24/09/2017 Location: AustraliaPosts: 129 |
Sorry to jump in, but I'm looking at an Ebay function generator driven by the xr2206 chip. Very cheap to purchase, any concerns going down this path to check transformers? ebay function generator kitset Ralph |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
Ralph, I suppose I would see if it can drive about 5V into a 50 Ohm load. We need to put a bit of power into the load so the DMM can easily measure the approx 1V results. wronger than a phone book full of wrong phone numbers |
||||
| Page 7 of 8 |
|||||
 Print this page |
