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.
9 hours ago, dochubert said:To get a cool running, continuous 12kw inverter trans, have you tried using two of your 6kw trans in parallel? (I'm having Visions of my older, multiple trans pj's) I realize it wouldn't be cost effective for production, but it might be a learning experiment.
No, the total size is much larger, requiring a larger chassis. It otherwise should theoretically work.
9 hours ago, dochubert said:Are your GS transformers using copper or aluminum wire? Your above comments to Dickson caused me to wonder.
Aluminum. The manufacturer insisted on trying a copper-wound tranny with the same specs--and it barely reached 120F at the 14kw load test (in other words, more than sufficient). BUT...the tranny was well past DOUBLE the cost (as copper is TRIPLE the cost of aluminum), and will weigh significantly more (as copper weighs almost double as the same sized aluminum)--which increases shipping weight. Double whammy for trying to make a cost-effective inverter.
9 hours ago, dochubert said:What transformer voltages did you end up using for the 48v/6kw GS inverter? 32/240?
Closer to 30/240...the sharper ratio to allow for more voltage drop in the tranny / cabling and still maintain a pure sine wave. (The ideal transformer core voltage also significantly affects the available options for transformer voltages...the next step up from ~30v would be 33v.)
9 hours ago, dochubert said:Do you still think 12kw is a realistic reachable goal (for a production inverter)?
Yes. We were quite close with the first prototype tranny...reaching ~10kw before temps started to go higher than we wanted (~180F) with a very typical 88% total inverter efficiency. I'm thinkin' those other Chinese manufacturers run their inverters well past 212F, if not 250F...which we aren't comfortable doing.
Adding another 30% wire to the primary on the next 12k tranny (as well as using a "high efficiency" core--higher core voltage) should hopefully push the total 12kw inverter efficiency up to 90% (or maybe even a tad higher) and allow it to run <180F at full 12kw load.
I'm thinkin' those other Chinese manufacturers run their inverters well past 212F, if not 250F...which we aren't comfortable doing.
I read somewhere that the transformer winding insulation will melt at 195 degree F . PJ inverter max should be 170 degree F where in one of Sean youtube say the transformer was on fire because the temp sensor did not shut down the PJ inverter .
I'm thinkin' those other Chinese manufacturers run their inverters well past 212F, if not 250F...which we aren't comfortable doing.
Yes I remember watching that video, very scary! If I remember it was a little 3kW PJ unit with an ASL1 transformer which would be capable of 1.2kW continuous at best!
I read somewhere that the transformer winding insulation will melt at 195 degree F . PJ inverter max should be 170 degree F where in one of Sean youtube say the transformer was on fire because the temp sensor did not shut down the PJ inverter .
Maximum temperature depends on the actual enamel on the wire, as well as the actual way the transformer is wound. PJ winding spec has too little wire on the high voltage AC output side--and that's the windings that are wound on the core first. This means that those windings' temperature far exceeds the temperature measured on the outside of the transformer, likely far in excess of 212F.
My rule of thumb is: if you smell the enamel baking while the tranny is running a test, it's too hot!
Going back to my earlier point, we don't actually need to measure flux density (Gauss) as it is already incorporated in the formalue in my spreadsheet. All that matters is that we stay under the maximum gauss of our chosen core, and 10,000 is a good compromise but higher is possible with good quality toroidal cores.
<a contenteditable="false" data-ipshover="" data-ipshover-target="/profile/2-sid-genetry-solar/?do=hovercard" data-mentionid="2" href="/profile/2-sid-genetry-solar/" rel="">@Sid Genetry Solar I got confused in my previous post. You were quite right that the turns/volt is not a fixed constant for a given core. However it varies according to the maximum flux density we want our core to see. This can be seen by altering the 'target gauss' parameter in my spreadsheet. As long as we stay well below 14,000 gauss all should be well.
Assuming a fixed pri-sec turns ratio, the more turns we add to both the primary and secondary, the lower the maximum gauss will be at full load, allowing for a smaller core. However then we start to encounter other problems such as wire resistance and difficulty getting enough turns on the undersized core.
As the old saying in automotive engineering goes, there's no substitute for cubic inches! The same applies to transformer CSAs. However just like automotive engines, larger transformers use more power (fuel) even under light load!
It's going to be a very difficult compromise to make an efficient 12kVA transformer, but I know you will get there in the end. It will never be as efficient at low-load as the 6kVA but I expect people buying the 12kVA unit will have more generating capacity than those buying the 6kVA and so can tolerate the higher low-load consumption.
Here's the spreadsheet I've been referring to.
The 'max kVA' calculation for any given core dimensions assumes a maximum gauss of 10,000, although the simplified formula doesn't take this as in input parameter.
The other parameters you can play with, and I've included links to some reference sources for those who are interested.
It's going to be a very difficult compromise to make an efficient 12kVA transformer, but I know you will get there in the end. It will never be as efficient at low-load as the 6kVA but I expect people buying the 12kVA unit will have more generating capacity than those buying the 6kVA and so can tolerate the higher low-load consumption.
Would you believe <0.8A @ 53v no load consumption on the current 12kw prototype?
GS6kw production inverters run ~0.5A @ 53v at no load. In other words, a single 12kw idle consumption should be slightly LESS than 2 GS6 inverters.
The updated GS12 tranny uses a different core than the current prototype...which is going to have the most impact on the no-load current. We will discover what the no-load power usage is once we get the tranny and actually test it on a GS setup.
I got confused in my previous post. You were quite right that the turns/volt is not a fixed constant for a given core. However it varies according to the maximum flux density we want our core to see. This can be seen by altering the 'target gauss' parameter in my spreadsheet. As long as we stay well below 14,000 gauss all should be well.
Can you explain how load affects the "gauss"? (I shall also link this [very knowledgeable] fellow who has a very strong opinion about the use of such terminology in first 3 paragraphs: https://ludens.cl/Electron/Magnet.html )
I would fully expect "max load determined by 'max gauss' of core" terminology in power amplifier circles--where the louder the racket is blasted, the higher the output voltage required to get more wattage through the (fixed resistance) speakers. This will definitely put a "cap" on the "maximum load" from the transformer core perspective--as once the core voltage starts to reach the saturation level, efficiency goes through the floor, heat goes through the roof, and the whole party comes to an end.
Conversely, in power inverters, the output voltage is (or should be!) constant from no load through full load. As a result, the core voltage will only increase a very small amount from no load to full load--actually, only as much as is necessary to overcome losses in the secondary winding for maintaining the desired output voltage. (This core voltage increase can very easily be calculated if the secondary winding resistance is known.)
If we're inadvertently mixing up "transformers for power amplifiers" with "transformers for DC-AC power inverters", this would explain a lot of the confusion--as the 2 worlds may only share similar power (wattage) ranges as common ground. The math and limitations for the separate fields are considerably different.
For inverters, I detailed above measuring the "peak efficiency" core voltage...using only (relatively) simple tools/equipment, not laboratory equipment costing thousands. (Funny when the expensive equipment comes to the same conclusion as the backyard experimenter!)
@dochubert I'm too young for the excuse to say that I couldn't read something--but I definitely misread what you wrote, thinking you were referring to the 48v/12kw GS inverter.
Yes, the 6kw 48v GS inverters run ~32vAC -> 240vAC. I'd kinda like to reduce that spec a tad, as due to the thinner windings, etc., losses make it start to flat-top the AC wave pretty early on. Issue there being the turns/volt spec, and the next step down being 28v--which is a good bit lower than I'd like to go.
I'm impressed that the prototype 12kVA GS inverter only consumes 0.8A at 48V idle. That's so much better than I expected!
Like you say, backyard testing (as I did) can come out with just the same results as expensive test equipment. I just used those formula as a guide for building my little transformer. as thought they might make in interesting read for the less experienced here. I don't pretend to understand it all to the extent the GS guys do.
Apologies if I came across as a 'know it all'. I can assure you that was not my intention.
It's just a very interesting discussion that I'm enjoying being involved in.
Apologies if I came across as a 'know it all'. I can assure you that was not my intention.
Not at all, I am just trying to understand how (or if!) "gauss" has any significant meaning for winding inverter power transformers. I'm very much learning about how transformers work at the moment anyway, figuring how to calculate winding specs for desired end goal, etc.
It seems to me that "gauss" and other terms are "fancy schmantzy" readings that then have to be calculated into something actually useful--such as core voltage (from which then the number of turns required on the coil can be determined.) But someone who knows everything about transformers (and all of their usecases/fields) will probably say that "gauss" is the best measure of a core. And they'd be right...because a power inverter transformer is usually running 50/60Hz and a fixed output voltage. Power amplifier is running the frequency spectrum and variable output voltage. Switching power supply transformers run extremely high frequencies, and different duty cycles, etc., etc.
P.S. Split the discussion to another thread 'cause it was quite off-topic on that U-Power thread 😉
Yes it had gone rather off topic, thanks for moving it!
Like you say, the formulas used may be ideal for a fixed power transformer where both input and output are expected to be a pure sine wave and the load is usually within a closely controlled range. But this is not the case for a DC-AC inverter, where the primary is driven by an SPWM signal (even if cleaned up a bit by a choke) and the output load can vary from nothing to several kW and we need to maintain reasonable efficiency throughout this entire range. This makes designing an inverter transformer much more complicated than my simple spreadsheet suggests.
Somehow I think that Sid's method of testing, experimenting, and calculating from the test results is far better approach. Mine was just a first stab in the dark being new to all this. I've yet to measure it's efficiency across it's entire load range and may find efficiency it is utter pants despite the fact it copes fine with the 3kW load during a 20 minute test run. Also I don't have the full tool set that Sid has (Variac etc), just a simple scope to check the cleanliness of the sine wave, AC/DC current clamp and multimeter to measure voltage and current.
Like you say, the formulas used may be ideal for a fixed power transformer where both input and output are expected to be a pure sine wave and the load is usually within a closely controlled range. But this is not the case for a DC-AC inverter, where the primary is driven by an SPWM signal (even if cleaned up a bit by a choke) and the output load can vary from nothing to several kW and we need to maintain reasonable efficiency throughout this entire range. This makes designing an inverter transformer much more complicated than my simple spreadsheet suggests.
An inverter transformer is actually quite simple to design...the #1 challenge is jockeying the specifications (core types and resulting winding specifications) to get wire losses low enough that the transformer temperature is within the desired temp range at full load. There isn't anything else you really CAN change or do--efficiency at lower ranges is going to be what it is. If the no-load current is very low, then the inverter SHOULD be quite efficient across the entire range.
What's complicated to design is a power amplifier transformer...those have a huge frequency range to deal with, as well as significant core voltage undulations, etc.
Mine was just a first stab in the dark being new to all this. I've yet to measure it's efficiency across it's entire load range and may find efficiency it is utter pants despite the fact it copes fine with the 3kW load during a 20 minute test run.
It does seem that the higher the desired wattage of the transformer, the more important the specifications become. At lower wattages (3kw for example), if you're running 80% efficiency, that's 600W of loss. Still rather high--but nothing compared to 80% efficiency at 12kw -> that's 2,400W. Bit harder to dissipate, especially if the transformer isn't 4x the physical size!
For starters, you can ohm out the coils on your transformer (if you want to!) I use a 50A Juntek adjustable power supply (yeah, not the cheapest thing, but I'm pretty impressed with them so far). Set the voltage to 2-5v, current to 50A...and connect it across the transformer coil in question. Use a DMM to measure DC voltage across the transformer coil (don't read the Juntek display, you'll get all your cabling losses added to the mess!)...then shut it off and use Ohm's Law to calculate resistance from the measured voltage and limited current.
Once you've established the resistance of all the coils on your transformer, you can then calculate wattage losses from the estimated full load amperage (i.e. 3kw / 220v on the output = 13.6A, and assuming 88% efficiency at 3kw is 3,360W / 32v input coil voltage = 105A) and the measured coil resistance.
This doesn't give you 100% of the losses, but it'll pretty quickly tell you if the coil sizes are mismatched! Again, I don't know how to calculate maximum transformer heat dissipation, but that means there's still more to learn 😉.
Also I don't have the full tool set that Sid has (Variac etc), just a simple scope to check the cleanliness of the sine wave, AC/DC current clamp and multimeter to measure voltage and current.
Compared to other test tools, variacs aren't terribly expensive. Bonus points if you can find a cheap one laying around a spare parts shop or something...but if not, they aren't prohibitively expensive online: https://www.ebay.co.uk/itm/272778720447
15 hours ago, Sid Genetry Solar said:It does seem that the higher the desired wattage of the transformer, the more important the specifications become. At lower wattages (3kw for example), if you're running 80% efficiency, that's 600W of loss. Still rather high--but nothing compared to 80% efficiency at 12kw -> that's 2,400W. Bit harder to dissipate, especially if the transformer isn't 4x the physical size!
I tried explaining the implications of the efficiency figures to a group some time ago but I don't think most of them understood. 2.4kW is more instantaneous power than a domestic electric oven uses but the oven cycles the elements once the temperature is reached too, so the typical average power draw is much less. A continuous 2.4kW in an oven wouldn't so much cook dinner as incinerate it. Those alleged 25+ kW PJ inverters would be having to get rid of something like 4kW - 5kW of continuous heating ... if they could actually produce that sort of output of course.
A continuous 2.4kW in an oven wouldn't so much cook dinner as incinerate it.
Also known as the "clean" cycle with a locked door 😉
Those alleged 25+ kW PJ inverters would be having to get rid of something like 4kW - 5kW of continuous heating ... if they could actually produce that sort of output of course.
I did actually ohm out a "20kw" PJ tranny, and came up with the following results based on the nameplate rating:
primary (36v): 10A @ 0.033v [20A @ 0.066v] = 3.3 milliohms resistance -> 22,000W / 36v = 611A = 1,232W of heat
secondary L1: 10A @ 0.762v = 76.2 milliohms +
secondary L2: 10A @ 1.36v = 136 milliohms = 212.2 milliohms resistance -> 20,000W / 240v = 83.3A = 1,472W of heat
-> 2,704W of heat, that's not possible to dissipate 😉
It's worth noting that PJ winds the 2 output phases with different wire thickness...that's why I measured them separately.
I'm also probably not accounting for ALL of the losses in the tranny--but the coils are a big enough loss...
-> 2,704W of heat, that's not possible to dissipate 😉
It's worth noting that PJ winds the 2 output phases with different wire thickness...that's why I measured them separately
That explain why my ASL 9 transformer overheat with 7 delta fans running a 4 ton heat pump and sound like a 747 in my back yard . Now only use 2 fans running less than 4000 watts . A real engineer would not design a transformer like that .