 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 11 of 26   |
|||||
| Author | Message | ||||
poida Guru  Joined: 02/02/2017 Location: AustraliaPosts: 1418 |
I support KeepIS views My inverters have rather smaller cables too. The chokes get warm, the toroid primary gets hot with extended high power. 4 x 4.5m2 for the primary. 3 x 4.5m2 for the choke. This is enough for the 12kW air compressor start surge into 3 x 4 FET power board. wronger than a phone book full of wrong phone numbers |
||||
| noneyabussiness Guru Joined: 31/07/2017 Location: AustraliaPosts: 512 |
If your worried about bugs etc. spray some of This on the board when you are finished.. seals everything mine has been in open air (covered) for many years with bugs and spider webs and it doesn't matter... i even use it to seal connections etc. now as it even stops copper from oxidizing.. I think it works !! |
||||
KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Yes, use that a lot, and the PCB cleaner. Although the cleaner always seems to make the board dull, and hard to remove all left over residue from between the pins. Perhaps you're supposed to use the whole can to do one small board  . It's all too hard. Mike. |
||||
phil99 Guru Joined: 11/02/2018 Location: AustraliaPosts: 2135 |
"Perhaps you're supposed to use the whole can to do one small board" Yes, unless you do that all that happens is the gunk gets redistributed. As the solvent dries the localized gunk is spread in a uniform film. You need enough to flush it off the board completely and it is too expensive for that. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Thanks for that, everyone used isopropyl many (many) years ago but it's not the best as it's hygroscopic for starters. Anyway, as it's an important board I guess I can waste a can of Cleaner. . It's all too hard. Mike. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
And so it begins at last. I built Wiseguy designed power board with isolated supply's and opto-isolated drive. I tested the boards 3 isolated supply's and the 4 Opto drivers - all worked first go. The picture shows the mock up with the Screw in Caps boards that Mike designed for me. I don't know if I will use all 8 caps yet. I have the older boards to swap in for testing, they are populated with 5uF 480v mains caps for testing. That should remove some HF spikes, and not have as much splat capacity as 80,000uF if something goes wrong as I work up to full voltage, and gradually increase the current limit from the DC supply, and I always use a swamping resistor between the Toriod and FETs for initial low power testing, especially at full voltage. Tomorrow I plan to start mounting the FETS and Drilling the three Heat sink Bars. I'm mounting the FETs directly to 38mm x 22mm heavy aluminum bars, the FETS are sandwiched between the board and these Heat Bars which become part of the Board assembly. The Heat bars in turn will be isolated when bolted to a huge heatsink after all testing is completed. I also started on the Controller, but taking my time to get everything right.  . Edited 2023-05-02 18:38 by KeepIS It's all too hard. Mike. |
||||
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
That looks rather familiar, very neat ! Those are some serious looking caps. The little square parts on the ends of the capacitor boards were intended to be cut off & separated to make very useful 3mm insulating washers. I assume that you remembered that the 8 screws mounting all the FETs under the + supply rail plane need insulating bushes that fit into the PCB holes so the screws dont short the + planes to the heatsink. I also used the square fibreglass washers on top of the insulating bushes. To tension the Mosfets onto the silicon pads, on the 3mm screws, I use rectangular section 3mm spring washers that are ~ 0.7 of a mm thick and about 1.5mm wide. I tighten the screws so the spring washer has just become compressed enough to resemble a flat washer and no further. Has worked fine for me for many many hundreds of FETs. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Thanks Mike, yes just a mock up with the cap boards to check clearance and correct orientation of the caps, I have to cut the washers off after a band saw blade change. No problem with shorting anything, I've done this many times building high power 1kW amplifiers. Just getting old means checking everything 20 times on different days in case one was a seniors day. I'm screwed if I get two in a row  . It's all too hard. Mike. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Thought I'd post a photo of the next stage of the build, 16 HY5608W mounted, ready for the board to be bolted to 3 intermediate heatsinks bars.  . It's all too hard. Mike. |
||||
Murphy's friend Guru Joined: 04/10/2019 Location: AustraliaPosts: 648 |
Cutting fibreglass PCB's with a bandsaw is *really* tough on the blade, blunts them in no time. Here is a little project for you: make a PCB cutter using one of those 1mm abrasive disks. How? Easy, I used an old mini angle grinder gearbox and adapted the motor shaft to fit a drill chuck. Make a little table to fit it under and if you use a slow turning drill it will cut PCB's perfectly. I think I posted a picture of it a few years back. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
It's an old blade kept just for this kind of work, I use it to cutout sections of old Grid tie inverters boards as well, yes it does get blunt but its a big blade, cost me $15.00 to get resharpened. Biggest problem I have with the abrasive cutters is the very fine invisible dust that drifts off, for some reason I'm really allergic to that type of dust. Yes I can wear a mask but I don't need to on the BS due to massive dust extraction. Still a great idea though, I might revisit it with some modified dust extraction when I get the time if it saves me changing a blade for small jobs. It's all too hard. Mike. |
||||
| analog8484 Senior Member  Joined: 11/11/2021 Location: United StatesPosts: 108 |
Makes a lot of sense. The thing that I find a bit unsettling is the trial and error nature of finding the optimal choke inductance. Given the collective experience of the many great inverter builders on this forum, it's a bit surprising that a simpler method hasn't been found. I also see many of the Aliexpress inverter boards recommend 48uH primary choke so it's similar to your findings. Another related thing I wonder about is the relationship between the primary choke inductance and the secondary filter cap size. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
I think most of us feel the Choke is a bit of black art, I was frustrated with this initially. There were so many things being discussed in building inverters and what has been shown to work, but not why. The initial reason for the primary choke was to reduce FET switching current / transients in the transformer, and thereby reduce the idle current. There was little concern about Saturation as FET switching speeds are so fast. I could see a pattern emerging as I repeatedly went over every build and failure for the past 7 years. It slowly became apparent the stronger and bigger we built the Toriod, winding it with railway track, and winding the Choke with the same railway track, that there was in my mind, a disaster waiting to happen, and all it needed was one thing to go wrong with a combination of loads, or a glitch anywhere in the FET drive, and bang. Anyway, early builds suggested around 40uH on an ferrite E-Core worked well to reduce idle current. As a bonus, the choke also removes some of the HI side 25kHz noise. However I also fit a choke of similar inductance to the LO side. A 50Hz square wave generates RF into the MHz range and the LO side choke made a subtle difference under complex loads. I also wanted some extra non saturating inductance in the Primary along with a little R-Loss as a buffer between the toriod with Railway line windings and the FETs. I have finally settled on the Black toriod rings, low cost, easy to wind. Stack a number of rings to give you the inductance you want, and the saturation you need for for your cable size and your Inverter usage. Problem is the Saturation level chosen is usually way to low. I found out in testing that Transient DC input Currents drawn by the FETS was in the hundreds of Amperes when seemingly small devices with low running loads cycle on and off and interact. However: 1: The inductance must be verified. 2: The Saturation level must be measured. I feel a lot of failures attributed to the poor old choke were not really from the choke itself, more likely there was a hidden weakness there all along, that weakness was pushed over the edge when playing with choke values that reduced the accidental R-Loss that had been initially in place. BTW the R-Loss (resistive loss) is very small in the scheme of things, and in no way undermines the reason we make big stacked toriods wound with Railway line. I now use a slightly smaller gauge of Railway line for the choke. I don believe that the Secondary capacitance has much to do with the final Choke value, maybe more to do with the reactance the FETS see driving into the primary under certian conditions, and that could be enough to send the FETS over the edge in combination with other build shortcomings or errors if the choke is wanting. It will be interesting when I get this new design driver running, it's Drive scheme is different, but I really feel it won't change the Choke requirements as they are basically the same, there is still switching and HF noise and the FETS still need the little safety margin that the chokes give them, especially with transient DC current peaks over 530A. But these are just my thoughts and findings in my inverter, for whatever they are worth. . . Edited 2023-05-04 12:21 by KeepIS It's all too hard. Mike. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Cut down the old scavenged Heat sink bars, tapped 16 x 3mm threads, everyone of them perfectly aligned, gotta love a small drill press to start the threads "manually" in seconds with perfectly vertical alignment on the aluminum block, then simply finish using the hand tap. I just finished fitting them to the power board. A job I was dreading that turned out to be simple and pretty dam quick for a change. I know others have their way of doing this but this is perfect, I guess whatever floats your boat.   . Edited 2023-05-04 19:16 by KeepIS It's all too hard. Mike. |
||||
| analog8484 Senior Member Joined: 11/11/2021 Location: United StatesPosts: 108 |
That's exactly my point. For sure. I look forward to seeing the results. Edited 2023-05-05 03:04 by analog8484 |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
The following discussion is generally based on the nano code at 19kHz, whether generating the SPWM regardless of unipolar bipolar or alternating mono/uni polar (I made that last one up to better? describe the way it works in its standard issued form) and an attempt to de-mistify the inverter choke. Consider a resistive loaded inverter at 240V, assume efficiency of 90% and a max modulation depth of 90% will have the following properties. Also assume the inverter maintains 240VAC output over the DC input range 45V to 52V. For a 1kW output: DC battery input current at 45V is 25A, at 52V is 21A Primary AC input voltage Peak is 41V, its AC RMS voltage is 29V FETs Peak current with 45V input is 54A or 38A RMS At 52V DC in, the FETs must still supply the same energy but they are now operating at a lower duty cycle, else with the higher input voltage for the same on period the energy would be greater so with 52V in the currents are 62A Peak and 44A RMS For a 2.5kW output: DC battery input current at 45V is 62A, at 52V is 53A Primary AC input voltage Peak is 41V, its AC RMS voltage is 29V FETs Peak current is 135A or 96A RMS At 52V DC in, the FET currents are 156A Peak and 111A RMS. For a 5kW output: DC battery input current at 45V is 123A Primary AC input voltage Peak is 41V, its AC RMS voltage is 29V FETs Peak current with 45V input is 271A, RMS is 192A. At 52V DC in the FET currents are 313A Peak and 222A RMS Assuming a turns ratio of 1:8. A toroidal transformer with a 10u capacitor on the secondary is reflected back through the 1:8 turns ratio to appear as a 640uF cap across the primary (I am accepting professor warps input that impedance is reflected back multiplied by the turns ratio and as a squared function), the impedance of that capacitor at 20kHz is 12 milliohms. With no choke the peak current with no load on the inverter at 45V would be in theory 3750Amps. With fully charged 52VDC that figure is 4,333A. Now consider adding a 47uH choke in series with this circuit. The 47uH choke has an XL impedance of ~ 5.9 Ohms and the current figures becomes 7.6A & 8.8A. Note I am ignoring winding resistances power factor and phase shift etc for the purposes of simplicity and the results should be considered as rough ball-park estimates. So it is easy to see that the introduction of a choke has a very beneficial effect. The choke has another beneficial function, it passes the sinusoidal average current required by the inverter but also stores extra energy during every single HF on period and releases it during every off period. It is easy to see that a saturating inductor can quickly raise the peak currents in an exponential manner, so it is not just about the micro henry value that has to be considered but the energy storage capability of the choke. To model the choke requirement the circuit can be considered as a buck regulator. Of course a buck regulator usually has a fixed voltage output, but if the reference voltage was varied in a sinusoidal manner the output voltage would also likewise vary. I use the peak current and voltage requirement of the buck converter model to calculate the choke worst case inductance requirement. The worst case occurs at max input voltage and the peak of the transformer primary voltage when regulating at the desired output. Please note I used 240V and 1:8 for the following calculation results, which yields a starting point to begin choke design from and empirically we can further tweak for a balance between idling current and max load efficiencies and waveform distortion. Lets consider a 1kW inverter and 45 - 52v input range: With 45V in, 41V out and 54A peak the inductor should be ~ 19uH using an inductor current swing of 10A P-P. With 52V in, 41V out and 62A peak the inductor should be ~ 46uH using an inductor current swing of 10A P-P. Now consider the 5kW inverter: With 45V in, 41V out and 271A peak the inductor should be ~ 8uH using an inductor current swing 25A P-P. With 52V in, 41V out and 313A peak the inductor should be ~ 19uH using an inductor current swing 25A P-P. Also note that the Nano uses I think 19.2kHz switching fundamental but the EG8010 uses 23.4kHz. I have run out of time to discuss further, re ampere turns, saturation etc which has all been discussed in previous posts. I have tried not to make errors here in my assumptions or calculations but am happy to receive criticisms arguments suggestions and highlighting of any errors. It would appear that Mads 2 x 10uH chokes were a reasonable starting point after all. My go to and recommended material is the sendust/HiFlux/KoolMu type moly/iron powder cores due to their graceful and gradual knee when approaching saturation. next best is the gapped C or double C core inductors. I consider that Ferrites are probably best avoided due to their sharp knee & rapid onset of heavy saturation. Edited 2023-05-05 15:42 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Murphy's friend Guru Joined: 04/10/2019 Location: AustraliaPosts: 648 |
Thanks Mike, that explanation is very helpful. So it appears that my chokes are way too big in the uH department, now I have more experimenting to look forward to  .Lithium battery voltages go from 52v to 55V in my case. Agree about the ferrite cores, the one Oztules suggested worked with the Powerjack inverter but it never was any good in any of the other inverters I built (EG8010 or nano). |
||||
| pd-- Senior Member Joined: 11/12/2020 Location: AustraliaPosts: 122 |
Below is a photo from a 5kw 48v commercial inverter you can see the sticker with 24.8 uh between the transformer & choke The unit switches @20khz and uses 6 x IFRB4310Z fets in each leg of the bridge There rated at 120A continuous & 560A Pulse each I recon your on the money there Mike  |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1679 |
Mike, I like effort and technical detail that you have put into that description, not only from the point of Choke parameters, but a general insight into the current values in this part of the inverter under load conditions. One thing I feel should always be verified in an inverter. Although I have a very stiff battery system with respect to the voltage it will hold under a high running load, you still need to put a DSO right at the inverter PCB (board) DC supply connection points, and capture the DC supply sag under a heavy startup load. The DC voltage on a 500A surge in mine is 47.8 volts, a big workshop VAC can surge to over 250A at start, and we've been over the other startup current winners. This low voltage sag is what I try to work from when calculating conditions the FETS may have to deal with in the inverter, and SFAIK, only a DSO will capture that. I forgot to add, like Murphy, my high voltage is usually 55V, similar batteries to Murphy friend. Looks like a good idea. . Edited 2023-05-05 19:36 by KeepIS It's all too hard. Mike. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
For the battery to fall from 55 to 47.8 with 500A draw is a drop of 7.2V. 500A & 7.2V yields an internal impedance for the battery + wiring of ~ 0.014 milliohms. I would be cautious of using the sag voltage to get the minimum value of the choke though. If the batteries or and wiring are replaced and perform better, the voltage could rise by quite a few extra volts. I am not warranting any of these calculations or theory with a FET back guarantee just describing a method that has worked for me so far. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Page 11 of 26 |
|||||
 Print this page |
