PLEASE NOTE: If you had an account with the previous forum, it has been ported to the new Genetry website!
You will need to reset the password to access the new forum. Click Log In → Forgot Password → enter your username or forum email address → click Email Reset Link.
There is always a middle ground. I actually tidied up the fan wires this morning in the lid with wire management...if it doesn't help with EMI and such, I don't know what will!
Looks nice man. Yeah doesn't help with EMI, you could try the heat shrink shielding or the wire wrap shielding.
Check out Zippertubing, guessing out of the price range, but maybe handy for development to know if shielding would help. They make good stuff and well been around forever so one of those names people trust to know its good stuff.
The main point in the routing for electrical cabinets is to ensure that everything stays in is place mechanically. The looser the wires are the more issues develop the terminal screws loosen up wires short out its crazy all the stuff that happens.
September 10, 2022:
-
All design files updated to reflect the various corrections determined by the test units:
- rounded off some (very) sharp corners on the lids. Laser-cut stainless steel is incredibly sharp--and very strong when compared to skin
- redesigned FET boards with an active clamp circuit for both 6kw & 12kw--as per test results, this should significantly help reliability & possibly even improve charge efficiency (the latter remains to be actually tested)
- slightly narrowed the FET boards on the 12kw to make it feasible to get the middle mainboard mounting screws in!
- adjusted positive busbar on the 12kw to fit better
- Updated files sent to the factory. Saturday EST is Sunday for China--their Monday starts Sunday evening EST. Factory has been greenlighted for production of the preorder run. About a month later than we'd hoped--but it's finally happened.
September 15, 2022:
- Very little in the way of updates...no progress reports from the factory as of yet. I am assuming (hoping!) that they have sent the PCB files out for manufacture.
-
Looks like I have to fire a connector company I asked to make the custom new MOSFET signal cables for the GS inverters. The biggest wire they stocked was 24AWG, but I wanted no less than 22AWG.
- Their 5pcs samples in 24AWG looked great. If we could change the cables to 20AWG, we'd be golden....
-
But as we wouldn't be ordering 1,000 connector sets, there would be a considerable amount of wire left on a (for them a special-order) roll of 20AWG shielded cable. With this in mind, I suggested that we could pay for the entire roll (that way they wouldn't be taking a loss!) and that way only buy 50-100pcs connector sets. If we never came back...they'd have the roll to themselves.
- after literally over a month of bugging them for a quote on a roll of 20AWG cable, I've finally concluded that despite their answers, they don't want to do business with us. I'm not comfortable with 24AWG cable on those FET gates...resistance is too high for my liking (not to mention being too close to the 28AWG ribbon cable that PJ uses...and which I know is far too thin). Yes, things look good with low-level tests on my bench--but I'd feel much more comfortable if things were a tad beefier than "the absolute minimum possible".
- Now I have to find sources for the connector headers, pins, and 20AWG wire for the inverter factory to procure and manufacture in-house.
- Headers are basic Dupont-style--but finding one with a polarity "key" is not easy. Heck, I have Amphenol header part numbers--but they're (seemingly purposely) just so slightly different than Dupont headers so that the easily-procurable Dupont pins...don't fit in them. And it goes without mentioning that the Amphenol headers alone cost 3x more than the entire cable assembly from the (now fired) connector company...
Manufacturing. Gotta love it.
On 9/15/2022 at 5:17 PM, Sid Genetry Solar said:finding one with a polarity "key" is not easy
Just an idea, but what some manufactures do is leave one pin out of the header and put a blank into one of the holes on the plug. That ensures that it can't be plugged in backwards also it can be used to prevent one cable from fitting into another location if the key pin is put into other positions.
2 hours ago, AquaticsLive said:Just an idea, but what some manufactures do is leave one pin out of the header and put a blank into one of the holes on the plug. That ensures that it can't be plugged in backwards also it can be used to prevent one cable from fitting into another location if the key pin is put into other positions.
Yah, a bit difficult when I'm basically using all the pins. And also harder to find the other side of the connector minus one pin.
We did find the connector from a connector manufacturer, so that's good. Takes a week to manufacture...not bad at all, I guess.
Yah, a bit difficult when I'm basically using all the pins. And also harder to find the other side of the connector minus one pin.
That part is normal you just get a fully populated header and during the install they just pull the pin out of the header. That is normal for the board manufacturer to do that, just have to let them know in the notes. A bit wasteful pulling the pin, but for the price of those they don't care. Interesting is they fill the hole in the board with solder, not sure why they do that, guessing its part of the board pick and place settings. Then on the plug side too you just pop the wire out and install the polarizing key. I get it though that the board design would need to be modded to accomplish this though if it wasn't planned for so not an easy fix for this version at least. Googled a bit and found the part number from Molex 15-04-0292.
We did find the connector from a connector manufacturer, so that's good. Takes a week to manufacture...not bad at all, I guess.
Not a worry then got this solved already. Just figured I would toss over a low cost solution that has been used. Most now have swapped over to a different style plug entirely since we were required to go back and glue them in place in industrial environments.
September 18, 2022:
- PCBs will be ordered this week. We will have the factory order extras, so we have spares on hand if needed.
- Transformer cores have already been ordered
- That connector company...after close to 50 days of waiting for a quote on 20AWG wire for the FET cables...FINALLY responded with a quote. Now, do I accept their quote and still use 'em...or tell them that they're far too slow?? Not sure I trust the inverter factory to get all the connectors pinned correctly--there have been notable mistakes on that front before. Hmm.....
- Still trying to sort out terminology to get those cases bent correctly...
September 21, 2022:
- Finally got the case bending corrected. And they got it tighter than I expected...or should I say, tighter than FreeCAD's Sheet Metal Workbench would allow me to!
<fileStore.core_Attachment>/monthly_2022_09/image.png.762370cd57db0e9bf9391d9302d00f45.png
- You'll probably note that this is a partial case. They're trying to reduce wasted material while getting this right...though somewhat comically, the lower case frame was bent backwards here, resulting in the front panel connections getting mirrored. I've already pointed this out.
- Funny how well all the screws line up when the case is bent closer to spec....
- Just below the red square, you may notice the corner on the "lid"...I rounded this over, as it was an extremely sharp point on the prototype units.
- With the case fitting so well now that it's bent as designed, I gave the factory the green light to manufacture all of the 12kw & 6kw cases.
- Still going in circles on the FET connectors. Now the company tells me that 20AWG is not possible (which I guess I'm not surprised by, as the connector datasheet specifies 22-24AWG for the particular pins).
- worth noting that 22AWG has 37% less resistance than 24AWG, so I can't complain too loudly. 20AWG would have been much better, but it does not look to be feasible.
- they're also saying that it would take 2-3 weeks for the run of connectors. Still not sure I want to trust them on it.
- It comes down to whether they give a good price...because if it's a bad price on TOP of being slow, then the factory will make the connectors.
- but then I'm concerned that the factory will mess up random connectors. We discovered that the power button on the test 12kw unit had a mirrored pinout--that's never happened before in over 100 inverters we've ordered from them!
- manufacturing: it isn't a perfect world!
September 29, 2022:
- Got some updated "test cases" from the factory with the proper bending specification...and they actually look like the cases I designed in FreeCAD now 😉. Here's the 12kw case; they are fully aware that the case screws aren't the spec type.
- yes, I do notice the little gap on the front at the bend. The bending location here looks correct on the other views, so not sure what's going on there.
- Underside view shows that the feet are bent nicely--there's actually LESS gap on the corners than FreeCAD would let me do!
- Asked the factory for an update, turns out that the inverter PCBs have not yet been ordered, nor will production of those start until after the Chinese holiday starting on Oct. 1st, and ending on Oct. 7th. Yeah, I thought PCB manufacture had already started....
- but there's a silver lining here. I think Sean is wanting to do a collab livestream this weekend about some updates on the 12kw project. Let's just say that I have made some major discoveries--and revising the MOS boards will very likely be in order. (Not a lengthy task at all.) As they aren't in production already, that means I can adjust the design without delaying things (or losing money by junking boards).
- I can say that it's almost funny how significant improvement in one area result in unexpected issues in a different area--and how design assumptions don't coexist very well with Murphy's Law.
- Connector company is back to not responding again. I'm done with them--we need the factory to find a local company to them that can make the GS MOS connectors.
Ready to update the pictures on the preorder listings?
October 7, 2022:
- After a round of bench tests, I finalized updated FET board and mainboard designs. Sent to the factory. It's after the latest Chinese holiday, so they should go into production hopefully within a day or 2.
- PCBs have a 2-week finish time. This is going way slower than anticipated--but in this case, the delays did work out for the better...because the significant improvements to the FET boards and mainboards not only reduce no-load, they also have huge reliability gains. Which is kinda critical....
Does this update effect the 6K as well?
redesigned FET boards with an active clamp circuit for both 6kw & 12kw--as per test results, this should significantly help reliability & possibly even improve charge efficiency (the latter remains to be actually tested)
Sean youtube say the voltage spike need to be fix on 12kw inverter . He also talk about Miller capacitance . The voltage spike is fix with an active clamp circuit and the Miller effect is fix with the one mosfet design spec . These two item being fix will stop the FETs from blowing up on inverter 10kw and higher . Powerjack do not have such protection and will blow up FETs at load over 10kw . What if that mosfets that take care of MIller effect is no longer available ?
Does this update effect the 6K as well?
All necessary fixes/improvements to the 12kw inverter I have also applied to the 6kw "whether needed or not." It's going to be worth the increased parts cost 😉.
All necessary fixes/improvements to the 12kw inverter I have also applied to the 6kw "whether needed or not." It's going to be worth the increased parts cost
These fix should be done to all Powerjack inverter . I done an active clamp circuit with the EG 8010 and it works with the ASL9.0 transformer . Saving the FETs is worth all the extra cost of labor and returned shipping .