 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 : Hopefully? Another 48vdc-240vac Toriod Inverter build.
  Page 8 of 26   |
|||||
| Author | Message | ||||
Murphy's friend Guru  Joined: 04/10/2019 Location: AustraliaPosts: 648 |
Mike, keep that one hour walking task going. I have been doing that for about 12 years since a medical scare and it *really* makes a difference to be a little fitter than the average 77yo couch potato. I found the opto isolated drivers are easiest built as half bridges, using two identical heatsinks and the mosfets isolated on mica washers. With my lay out in a tallish cabinet it just requires a single fan that only runs on very hot days. I can empathize with "hoarded crap", have plenty of that here but find it very difficult to part with  . |
||||
KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
I totally agree how difficult it is to part with stuff, a lot of old electronic bits and pieces that will never be used, I / we hang on to so much stuff for future project that never come. My wife and I are all about de-cluttering now and the shed is the only place left to get stuck into. I have actually walked almost every day for the past 30 years, before that, high level ballroom dancing 3 to 4 nights a week for hours at time and up until 10 years ago 2 nights a week and all Saturday morning. Likely why I survived, apparently I should not be here. One lousy artery, the rest are fine, anyway it is what it is - I'm also in my 70's. I have a few ideas on heatsink layout and mounting but until I get the board partly built and do a test layout I'm not locking myself into any preconceived layout. I'm glad I spent so much time on the test driver build, running 24/7 has made me realize that on stinking hot days running around 3.5kW for a few hours builds heat gradually and although that 3 stack toriod is maintained under 43 deg and feels only warm to touch mainly the primary winding and one choke, but the bulk / weight of Toriod holds that heat for hours, it made me realize how important a decent heatsink for the driver is under these conditions. So the new driver will have some generous heat-sinking, which fortunately I allowed for when I built the cabinet. I think you and and few others mentioned that along the way. It's all too hard. Mike. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Just some more ramblings and thoughts and testing. I've been testing current clamps finding a way to test the limits of same at high currents. Along the way I've discovered some more interesting peak currents from the Batteries to the Inverter. I was questioning myself and the Current clamps so I went down a bit of a rabbit hole, rewriting the Averaging code in my Solar monitor and resolving a way to show true instantaneous current peaks on the Graphs while maintaining averaged Digital display values that agreed with commercial monitoring devices. The Clamp I'm using is rated at 500A @ 4v with 750A upper limit, it has a 12V rail and can run to around 10v for about 1300A at unknown accuracy. Link to Clamp A commercial peak hold (slow decay) clamp meter agrees with my Current clamps peaks and the waveforms produced by these clamps. It appears that some AC Loads are constantly oscillating around the averaged current point for that load depending on the load, it can be around +- 10A of swing at the Inverter DC input. The CRO does NOT show any spikes on the DC line, just some noise superimposed on the AC waveform from the current swing into the Toriod drive and load, < 200mv PP to < 500mv PP depending on total load current. Now is this input current deviation the inverter modulating the AC output voltage (won't / does not show on a standard meter or DSO) or are various AC loads on the inverter interacting causing the load to vary. NOTE: All these tests were done with the House running off grid and on this inverter, in other words a lot of AC devices connected. If you try and measure instantaneous peaks with averaging you obviously will fail on narrow Pulses. So why are these narrow peak currents important to me? Well, if it can be measured then the Inverter is pulling that current, if only for a few ms, and once again, our workshop equipment has Load power on peaks in excess of 16 kW !!! This is why I want a Choke that does not completely saturate and turn into a piece of wire until at least 400A. The other day we ran the inverter for over 3 hours at just under 5 kW continuous load with peaks of 290A every 10 to 15 seconds for 15 minutes as small logs were carved up on the Band saw. This B-saw has a fast stop foot break and we stop and start it for safety as we position the logs after every cut, 300mm of aggressive spinning naked blade is not something I want to shake hands with. The inverter hardly makes a peep with this load and the Toriod was held at 37 deg with the Inverter at 36 deg, both cooled by the case fan. The Toriod temperature is read from between the Stacked Cores and sealed on to the Secondary windings and sealed from any air flow. One choke was a problem though, it's a Toriod ring choke wound with smaller gauge cable to get the number of turns required for the 37uH inductance I was after. I converted it to a 12 stack Toriod ring choke and just squeezed 4 turns of heavy gauge cable through it, that has totally fixed the heat issue when running at 5kW. Measurements with this choke are interesting with NO real knee out to the limit of the peak Current from the Sat tester (over 1800A) The "Resistive Current shunt" voltage drop was 1800A and the 500A Current Clamp was clipped at 1320A on the DSO. This brings me to this question again, could some of the unknown inverter self destruction be a result of these large transient peak currents. A simple less problematic example is switching off a 2kW resistive load, it can cause a big current spike if, like a lot of gear these days, it does not have protection from switch arcing, also consider this when switching of at the wall socket. I'm sure most people are over pictures of Choke Saturation curves so I won't post any unless someone actually wants to see them. It's all too hard. Mike. |
||||
| Murphy's friend Guru Joined: 04/10/2019 Location: AustraliaPosts: 648 |
Mike, somehow I think you are the only one here torturing his inverter with outrageously high loads. Mine does just the domestic loads plus some smaller workshop equipment. But I have added the second choke like you did so now there is a choke in each primary leg. I'll find out soon how this takes any punishment I have on hand here. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Yes I agree, the perils of having some large woodworking equipment that my wife uses daily, sometimes with 3 or 4 running at the same time. This I though would be impossible on an Inverter a year ago. I'm convinced that the robustness under high loads really comes down to the Toriod Transformer design, then the choke and finally the wiring between those components and the Inverter Driver board. I say that because of the numerous YouTube videos I've seen of people blowing up these Chinese Inverter boards. The ones I see that don't blow them under high loads are almost always made by a builder with experience or the know how to build a correctly sized Toriod transformer and choke and the layout to match his loads. I think the Choke design will change when I finish wiseguys driver board, but it should be a good starting point. Finally got some bench space and components in marked bins ready to assemble. Yes, I run the second choke made to not saturate under high loads, it's characteristics are different to the smaller choke, they are both the same inductance but behave differently when placed in, what is, the HI side of this inverter. That obviously changes with inverter driver design. It's all too hard. Mike. |
||||
InPhase Senior Member  Joined: 15/12/2020 Location: United StatesPosts: 178 |
I don't have nearly the experience that these guys have here, but in my testing, the FET drive was the key. I went through the iterations pretty quickly. I started with the charge pump high side supply, and it worked, but occasionally FETs would blow, especially if even a small load switched rapidly. Then I moved to a charge pump + totem pole drive, and that solved it for small loads, but rapidly switching induction motors or other heavy loads would kill the FETs. Finally, I went to isolated power supplies and optocouplers and totem pole, with a -5 V turn off voltage. Now nothing kills it. Having a strong drive and a negative off state really holds it together. Edited 2023-04-05 22:56 by InPhase |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Those improvements in various iterations are incorporated in the the newer Inverter design that I'm changing over to. The Chinese board I'm using was my test board when building this Toriod inverter, however apart from totem pole drive it's a conventional board. It has been abused almost from day 1 with 10kW to 16kW starting power surges and running the Workshop and Home, and it still appears unbreakable in this build with this Toriod. If you get the combination of Toriod, choke and wiring right, even an older driver design will work, as some have for years in harsh conditions like mine. This was the case with an earlier conventional design similar to this China board, however a simple mod was made to the EG8010 control board to stop it self destructing under certain conditions, I made that modification to this Board as well. Even with that, some people blew up a lot of FETS trying to get a Toriod design to work at power. A few things that were not a constant between each build were: 1: The Home made Toriod 2: The Choke 3: The interconnect wiring 4: The way the Driver board PCB was assembled and how FETS were mounted and soldered. 5: The quality of the FETS and components purchased for the build. Get one wrong and it's a failure waiting to happen. The workshop has at least a 10 big devices with induction motors, one is a 3Hp dust extraction system three meters high and housed in a dedicated enclosure outside the shed. No start cap, no start winding, just a huge motor that slowly grinds a massive 14" iron flywheel up to speed. The inverter gives a little hum when it starts it, and it starts it like it's on mains. That happens around 4 to 5 times a day, along with the other equipment. I often had nightmares thinking about those tiny FET internal connections doing that. Now I don't even give it a second thought. And the bare inverter idles at 17 watts. Thanks for the info, FYI I have been reading your posts and enjoy the knowledge and real world insight you share here. . Edited 2023-04-06 10:09 by KeepIS It's all too hard. Mike. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Need some help here getting my aging one brain cell working: To start with I'm feeling a bit embarrassed as I realized that measuring the noise on the 0 Gauge DC cables feeding the inverter was incorrect, yes it did change with load but I never placed a big load over 3kW on the inverter and I was simply measuring across the supply lines with a DSO. This is where the cycling is coming from out of the Clamp Meter and not 200ms to 500mv as displayed on the DSO directly reading between the + - DC cables. The DC current flow to the inverter is a reflection of the current consumed to produce the waveform that the inverter is driving into the Toriod and the Load. It stands to reason that it will end up being something like an AC 50Hz waveform (is my assumption wrong) the 500A Current clamp output indicates that this is correct. The Magnitude of the waveform is proportional to the Current drawn by the Load. Now we all know trying to measure anything across a resistive Current Shunt on the DC input to an Inverter is never going to work without some serious active filtering, and considering that the voltage is only 50mv for 200A and filtering my distort the true fast rising current peaks, then a quality Isolated current Clamp is the way to go, and this unit has very little HF noise at the output and produces 125A per Volt, you can use unshielded leads to the DSO and it's still clean. I thought my Code for reading the Clamp meter on my ARM H7 controller was wrong as it showed very high peak current and needed to be averaged at least 3 times to match the Commercial peak reading Current clamp meter. My problem and question is this: The clamp is unidirectional and only reads the positive going current, but obviously will still produce a sine wave if the current drawn is a high power sine wave current. A: Is the positive "DC current" excursion flowing into the Inverter the Peak current? B: Does it need to averaged because of the Wave shape. C: What is the expected current waveform at the input to the inverter under load. D: Is a P-P current that needs to halved and converted to RMS current? Like I said, brain not working and the reason is the Magnitude of the current and measured on an 0 Gauge cable close to the Battery Bank. I'm going to try an active 50 Hz notch filter or Hi Pass filter on the output of the Clamp and see if this gives a better result with respect to what I'm seeing. These are the DSO captures from Powering up the 3HP Dust Extractor: DSO 1V per division = 125A per division from the DC Current Clamp. Below: The current at almost 500A and the startup duration of 2.2 seconds before the current drops below 375A. That's around 500A * 47V = 23.5kW which cannot be correct, 12kW I can believe.  Below: Compressed startup waveform to give a idea of the surge current duration and shape.  Below: The AC current flowing through the 0 gauge cable.  . Edited 2023-04-06 14:56 by KeepIS It's all too hard. Mike. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
You know, thinking about only the peak current, the Current Clamp is showing a transition from around 7A load to 500A peaks through the cable. If I were calculating average power and current input then I would be using standard AC calculations. But to me it really does look like the peak current is reaching 500A DC from the batteries, considering the Voltage at the inverter input has dropped from 56V to 46.7V on those peaks. I capture the highest instantaneous current and lowest voltage for Transient peak readings. Even true short term 200A current draw only drops the voltage to 51V. That's from 25kW of LiFePO4 batteries with a constant discharge current rating of over 300A and a short peak capacity of around 700A through 1M 0 Gauge + - cable. EDIT: Reversing the Clamp obviously results in absolutely no current flow detected in reverse direction in the cable. The AC waveform, as mentioned before, is a result of only a positive current flow cycle from the inverter, have I answered my own question? FYI The Huge Compressor shows a 563A Peaks at 47V for 800ms - you do the math - what am I missing. This clamp produces 500A at 4V, the Clamp output was 4.5v on the Comp. BTW I have confirmed the clamps readings are accurate to 700A when compared to a high quality Resistive shunt on my Saturation tester. . Edited 2023-04-06 16:57 by KeepIS It's all too hard. Mike. |
||||
Revlac Guru Joined: 31/12/2016 Location: AustraliaPosts: 1024 |
I'm no good at maths either, I can get some very high current readings if I put a clamp meter on the transformer primary and know its a false reading compared to power in power out, If it is high current pulses from the batteries, power must be wasted somewhere? I have some wood working gear but all of it is 3 phase, compressors as well. Cheers Aaron Off The Grid |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Hi Aaron, the math isn't the problem, it's the power it comes out at which, BTW, is 26.5kW  I could understand if it was spikes causing it, but it's not, as shown by the waveforms, the first thing I went looking for. The current Sensor is accurately tracking the current drawn by the inverter to enable the Toriod produce a sine wave. Yes, a current sensor on the Toriod primary is going to show a lot of HF and digital crap from the inverter drive waveforms, but the Battery DC input has no crap, again as shown by the waveform, the DC input is more a reflection of the AC waveform current swing in the Toriod. Of Course I could be totally wrong and missing the bleeding obvious, hence my plea for help from someone with a clear perspective. But one thing I have proven, that the measured current has to flow through the cable for this Current Clamp to produce this output voltage. This Sensor is fast enough to display microsecond pulses with clarity. There are none. BTW The efficiency of this inverter is over 92% at high power levels. The last photo is worded incorrectly. It should read: Shows the Current waveform flowing through the 0 Gauge cable. . Edited 2023-04-06 18:51 by KeepIS It's all too hard. Mike. |
||||
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Mike, I have not really been following what you are doing and I might be confused but maybe I can try to add some clarity ? (or confusion) The last picture shows a peak about every division and a bit. The time scale indicates 8mS per division (a strange number what happened to 5 or 10.....). So it appears that there is a positive peak ~ every 10mS. This is what I would expect as the current ramps from close to zero to maximum every 10mS. It starts from zero, drives to a positive peak over 10mS & returns to zero and then drives to a negative peak over the next 10mSecs and returns to zero. The current ripple should be at a 100Hz or 10mS rate. Now to work out the true power, if you draw a line at approximately half the peak amplitude and fold it down then we have filled in the power envelope considerably so it now looks like inverted 100Hz ripple, Now if you multiply that result by 0.7071 that is how to convert the ripple peak to RMS. So if we are seeing 500A peaks I would average that at 250A and then multiply it by 0.7071 resulting in 176 amps RMS, lets multiply that by 48V and the result is ~ 8,500W. What I dont get is that surely you know this and I'm missing something as you would need to know all this to calculate the inverter efficiency at ~ 92% ? I assume you have an RMS AC amp clamp meter that will confirm the secondary power and > 90% sounds like the right ball park figure else there would be a lot of heat generated somewhere. I do agree that the power peaks are ~ 24kW but the RMS is still around the 8.5kW mark? I dont know if this helps or maybe you can confirm what you are trying to work out ? If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Hi Mike, thanks for popping in, yes I mentioned calculating the power a few posts back. The thing I'm trying to confirm is only the instantaneous peak current that would flow into the Inverter. To me the current peak during startup of some of these big induction motors appears to be very high at 500 to 563A, higher than I had originally thought, due to averaging and my early Current Clamp limitations. I needed to put this in a post as it gives me some clarity of sorts to think it through. If the brief instantaneous Peak current really is 563A, and I confirmed that the Current clamp is accurate, then it's truly mind boggling to me that the FETS can take that abuse. Especially as at a slightly lower current those peaks continue for 2.2 seconds before dropping down to the Machine running current. Edited 2023-04-06 20:14 by KeepIS It's all too hard. Mike. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
If the load was ~ the 8.5kW mark as far as I'm concerned what you are seeing is the true peak current. - Yes you dont have to do too much wrong at those levels to let the smoke out. That is why choke design to avoid saturation is really important. If they are HY4008s and there are 4 of them then its ~ 125A peak each and they are rated for 200A but I do agree it is all a bit scary. Edit: I just went back a few pages and saw 8 FETs per leg - crikeys its just idling at 500A peak  Out of curiosity, are you planning to build the new power stage with negative bias ? If so you will be a bit of a pioneer - in the previous design I used a negative power module. In the new design if you use a 15V supply and include the 12V zener clamp on the rail you end up with ~ -3.0V of negative bias. I have not actually built one up to test yet - tell me if it works or if it all went bang lol. Hopefully I will get to test it before you finish yours - I am very confident it will work just fine. Edited 2023-04-06 21:16 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| analog8484 Senior Member Joined: 11/11/2021 Location: United StatesPosts: 108 |
What is the power factor of the AC power to the dust collector? |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Funny thing, I was thinking about the number of FETs on the board whilst in the Shower last night and realized the same thing, at the same time I thought as long as they are load sharing equally. I think it's safe to say they are in this case. I was going to ask you if you found the neg bias option worth while, so it looks like you may get to test it before I can as the next two weeks are a bit of crap shoot for me. FYI The Choke I just made does not appear to saturate until way past 600A. I had to remake it as it was getting hot running at 5kW for a hour or so, and the only thing that got hot. I ended up with a 10 Stack Toriod Ring core and a few turns of High current cable for a 37uH choke, it barely gets warm at those power levels. From memory the mains power factor with the DE running is .96, our huge 25 year old Air Compressor draws more current than the Dust Extractor though, however for a much shorter duration. BTW the question of Peak current was only for the DC supply current and not the AC output currents. . Edited 2023-04-07 08:43 by KeepIS It's all too hard. Mike. |
||||
| pd-- Senior Member Joined: 11/12/2020 Location: AustraliaPosts: 122 |
The peek Discharge currents seen on my current shunt between the battery & inverter are routinely scary Over the past few days Discharge 1294A Charge 78A Discharge 1270A Charge 76A Discharge 1444A Charge 89A Discharge 1549A Charge 76A The Battery is capable of supplying 11KA So its just the on resistance of the fets and the impedance through the primary & choke that limits the peek current. So as long as it is lower than the max pulse current of the fets & you have enough copper to balance the load across the fets you are ok |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Are you sure the peak discharge current of over 1270A is correct? If it is, then at 48v that's over 60kW peak discharge. . It's all too hard. Mike. |
||||
| pd-- Senior Member Joined: 11/12/2020 Location: AustraliaPosts: 122 |
Im not shore I did not build it . its a commercial current shunt with inbuilt electronics . i have a multimeter with a recording function , i could connect it across a shunt and i may get lucky but its shortest record interval is one second Other than that i would need to build sumthing or get a DSO witch would be handy for other projects my analog scope is getting on a bit. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
The problem with high power resistive current shunts is the very low voltage produced for the Current through them, commonly anywhere from 10mv to 75mv per 200A, so trying to read that with a multimeter is virtually impossible due to the massive transient currents and inverter noise on the shunt and power cable, 100mv to 500mv or more will swamp anything you try to read. They can work with balanced filtering but take more effort then it's worth IMHO, especially since a good isolated Current clamp is under $30 and can read to over 700A depending on which one you choose. A DSO of some type will still be needed with an isolated Current Clamp if you want to to catch peak transients, but a multimeter is fine for everything else, a unit like the one I posted a link to earlier, it's low cost and very low noise. . It's all too hard. Mike. |
||||
| Page 8 of 26 |
|||||
 Print this page |
