Been making 'improvements' to the PJ/Upower 8k

Hi Paul!

Looks like some very nice work there!  I thought I was the only one who added a resistor to the resistor ladder to increase output voltage.  It does work.  I used that control board for several years before it became unrepairable.  I added one resistor the same value as the others in the ladder and got an almost perfect 240v output (what I wanted) instead of the rated 230v it previously supplied.  That was on a 15kw, 48v powerjack.

As for fans, I added independent control in parallel with the existing setup.  Cheap ebay temp control.  One control for heatsink temp and another for transformer temp.  That way I could set at exactly what temp I wanted the fans to turn on, while still allowing powerjack's algorythm to turn them on.  And I can see at a glance what temp my equipment is running at.  My settings always had the fans on sooner anyway, but the redundancy was reassuring.  I have more recently gone to a two speed fan setup using a high speed fan with a resistor for low speed and shorting the resistor for high speed. 

You might compare the rating of the shaping capacitor(large brown cap on control board) on your two inverters.  Powerjack got cheap (surprise!) and went with a lower rated cap on some newer units.  Your older unit probably has a 475j 630v cap across L1 and L2.  If the upower unit's cap isn't that size, your output sine wave  might improve by replacing with the 475j 630v.  Worth a try anyway.

I never thought much of running my entire house through those crappy little fuse holders either.  Getting rid of them is definitely a good idea.  I eliminated mine in favor of breakers also.

Your self-wound transformer looks great!  Looking forward to hearing how it does with your new v11 board.  Pic of the temp control I use below.

I forgot I was going to bring up the idle current issue.  From what I've learned over the years using powerjacks, about the best we can do (in my non-expert opinion) is put one or two turns of the low voltage side of each of the transformer leads through a ferrite core.  This is, of course, a compromise.  But, not much more is gained by a large custom made and expensive choke. Yes, that large high current choke will reduce the idle current more than the two turns thru a core, but how much more and is it worth doing?

That brings me to the point.  I run my inverter to power my house.  I never run it for any length of time with no load.  It always has a load of some amount.  Sometimes a large load.  Anywhere from 10a to 95a in a given day's use.  Probably showing my ignorance here, but what difference does it make what the no load current is if it never is run at no load?  It seems way too much emphasis is put on idle current. 

Maybe I'm wrong.  If so, maybe someone will explain.

Always fun to see someone trying to improve a PJ 🤣.

6 hours ago, Paul said:

and experimented on the 'Upower' unit with a new home-wound transformer based on a 52cm2 toroidal core which I obtained via ebay and stripped, then rewound.

What voltage specifications did you wind your transformer to?  Just curious 😉

39 minutes ago, dochubert said:

Probably showing my ignorance here, but what difference does it make what the no load current is if it never is run at no load?  It seems way too much emphasis is put on idle current. 

Maybe I'm wrong.  If so, maybe someone will explain.

So my current conclusion is that the ferrite chokes are for absorbing/"filtering" the 24KHz SPWM carrier before it reaches the main transformer.  The main transformer will appear more or less of a short circuit to a 24KHz signal--which would explain why without the chokes, the idle current is so extremely high. 

My guess is that yes, while the no load current is a very easy "measurement point", I would expect that a high-idle inverter will also run at a measurably lower efficiency (and higher temperature) due to the non-useable 24KHz carrier getting dissipated as heat in the main transformer.  If said carrier can be filtered out via the chokes, then there should be a higher efficiency end result for the main transformer.

44 minutes ago, dochubert said:

From what I've learned over the years using powerjacks, about the best we can do (in my non-expert opinion) is put one or two turns of the low voltage side of each of the transformer leads through a ferrite core.  This is, of course, a compromise.  But, not much more is gained by a large custom made and expensive choke. Yes, that large high current choke will reduce the idle current more than the two turns thru a core, but how much more and is it worth doing?

My personal experience (having tried multiple PJ chokes, a large E-core, lots of inverter design testing, etc., etc.) is that 2 full turns around a PJ ferrite is the maximum it can do.  More turns don't seem to help no load.  Had someone try gluing 4 ferrites together and running 2 turns through the whole stack--same no-load as a single PJ ferrite with 2 turns.  Ditto for the E-core--same as 2 turns around a PJ ferrite.

Where it gets interesting is if when running extreme high voltages on an inverter design (referring to 36v on a 24v inverter).  While the no-load current remains relatively constant from 24-30v, somewhere beyond that, the no-load current starts to rise again.  Adding a second choke (2 turns) on the other leg of the transformer reduces the no-load current back to the former number.  Obviously, with the increased voltage and same no load current, the efficiency is going down...but extra ferrite chokes do help at higher voltages on a lower voltage transformer.

...and don't try running a 24v PJ inverter at 36v (or a 48v at 65v).  It's going to blow up 😉.  Been there, done that.

Funny how

21 hours ago, Sid Genetry Solar said:

Always fun to see someone trying to improve a PJ 🤣.

 

What voltage specifications did you wind your transformer to?  Just curious 😉

 

So my current conclusion is that the ferrite chokes are for absorbing/"filtering" the 24KHz SPWM carrier before it reaches the main transformer.  The main transformer will appear more or less of a short circuit to a 24KHz signal--which would explain why without the chokes, the idle current is so extremely high. 

My guess is that yes, while the no load current is a very easy "measurement point", I would expect that a high-idle inverter will also run at a measurably lower efficiency (and higher temperature) due to the non-useable 24KHz carrier getting dissipated as heat in the main transformer.  If said carrier can be filtered out via the chokes, then there should be a higher efficiency end result for the main transformer.

 

My personal experience (having tried multiple PJ chokes, a large E-core, lots of inverter design testing, etc., etc.) is that 2 full turns around a PJ ferrite is the maximum it can do.  More turns don't seem to help no load.  Had someone try gluing 4 ferrites together and running 2 turns through the whole stack--same no-load as a single PJ ferrite with 2 turns.  Ditto for the E-core--same as 2 turns around a PJ ferrite.

Where it gets interesting is if when running extreme high voltages on an inverter design (referring to 36v on a 24v inverter).  While the no-load current remains relatively constant from 24-30v, somewhere beyond that, the no-load current starts to rise again.  Adding a second choke (2 turns) on the other leg of the transformer reduces the no-load current back to the former number.  Obviously, with the increased voltage and same no load current, the efficiency is going down...but extra ferrite chokes do help at higher voltages on a lower voltage transformer.

...and don't try running a 24v PJ inverter at 36v (or a 48v at 65v).  It's going to blow up 😉.  Been there, done that.

Funny how we all seem to be good at blowing up PJ products.

With the 4 glued cores, would it have been able to take more turns to lower the no load? Or would it just as likely become a heating element?  What about two cores in series on the same wires?

No, the limit to the number of turns has to do with the physical limitation of the core hole diameter and the size of the wires.  Stacking 4 cores together would (I think) increase the inductance of the choke, allowing for higher voltage operation. 

Once the 24KHz carrier has been reasonably filtered out, no further reduction of no-load current seems possible.  (At least in my experience.) 

Speaking of a heating element: with a GS inverter cleanly and equally driving all four quadrants of the H-bridge, the chokes get HOT at no load.  This in part is why my (unscientific) hypothesis is that the chokes are filtering out the 24KHz carrier--because if the FETs are being firmly driven, there'll be sharper corners on the 24KHz carrier as compared to a PJ.  (Transformer also runs nearly silent at no load with the GS design.  Tests on the GS 12kw inverter with 11 FETs...the transformer is SO quiet, I can't hear a hum even putting my ear on the transformer.  Of course, under load, it does start to hum/buzz.)

On 10/17/2021 at 12:15 AM, dochubert said:

Hi Paul!

Looks like some very nice work there!  I thought I was the only one who added a resistor to the resistor ladder to increase output voltage.  It does work.  I used that control board for several years before it became unrepairable.  I added one resistor the same value as the others in the ladder and got an almost perfect 240v output (what I wanted) instead of the rated 230v it previously supplied.  That was on a 15kw, 48v powerjack.

As for fans, I added independent control in parallel with the existing setup.  Cheap ebay temp control.  One control for heatsink temp and another for transformer temp.  That way I could set at exactly what temp I wanted the fans to turn on, while still allowing powerjack's algorythm to turn them on.  And I can see at a glance what temp my equipment is running at.  My settings always had the fans on sooner anyway, but the redundancy was reassuring.  I have more recently gone to a two speed fan setup using a high speed fan with a resistor for low speed and shorting the resistor for high speed. 

You might compare the rating of the shaping capacitor(large brown cap on control board) on your two inverters.  Powerjack got cheap (surprise!) and went with a lower rated cap on some newer units.  Your older unit probably has a 475j 630v cap across L1 and L2.  If the upower unit's cap isn't that size, your output sine wave  might improve by replacing with the 475j 630v.  Worth a try anyway.

I never thought much of running my entire house through those crappy little fuse holders either.  Getting rid of them is definitely a good idea.  I eliminated mine in favor of breakers also.

Your self-wound transformer looks great!  Looking forward to hearing how it does with your new v11 board.  Pic of the temp control I use below.

//content.invisioncic.com/g308908/monthly_2021_10/1453947873_tempcontrol.thumb.jpg.a80379e91efe4d84e1a95df036593bb2.jpg

Thanks. I added a 470k resistor which seemed to increase the output voltage to ~245V. The newer boards that I have (10.3c and 11.1) have a pot on the control panel to adjust the output voltage but this seems to max out at 236v. That's a moot point anyway with the ASL3 transformer in the Upower inverter which is too small and has the wrong winding ratio to achieve that RMS voltage without clipping of the peaks when run above about 800w anyway...

However when I finally get around to assembling the new unit with the rev11.1 boards and my home-wound transformer I would like to find out how to raise this above 236v as I know from testing that my transformer doesn't suffer any clipping even with a 3.5kW load. However I am a little confused as the newer boards seem to have 3 resistor ladders - one on the control board, and another two on a double-sided sub-board of the separate charging board.

I like your fan control board - much more flexible than my setup but nevertheless my setup seems to work ok.

I started off with a 52cm2 CSA toroid. Target min input  = 18v DC, Target avg output = 240v RMS 50Hz.

I have put 210 turns of 2mm diameter enamelled wire for the secondary, and 12 turns of 16mm2 CSA (two in-hand as seen in photo so total 32mm2 CSA)for the primary. I found some simplified formulae on the internet and made a spreadsheet to calculate approximate parameters (see attachment).

I've not blown up any PJ's yet and I don't intend trying to, so I won't be running it at more than the limit set on my charge controller (29.4V) 😉