testing the egs002 inverter board

[oztules]

The circuit you alluded to is a HF complete circuit.

If you look at it you will see a 12v to 320v step up push pull converter to do the voltage lifting, and then they use the 002 to do the 50hz sine wave.
The problem with these, is that all your energy storage is in the hv capacitors @ 320v... so if you have a surge requirement that can't be met from that tiny energy storage, you sag.

The LF uses the battery as the energy storage, so basically unlimited surge, for as long as the tranny and fets can stand it.

It also suffers form driving highly capacitive loads, and will probably blow the fets in that case, as they see the caps as a very low impedance.

It is always useful to see how things work, as the more you learn, the more you can do.


Clockman, the protagonist has self immolated, and the admins have given you their blessings.  The politics of envy have been played out.


..............oztules

[lighthunter]

Hi guys, great work on evaluating this inverter option. I ordered 3, just to play, no plans yet but was curious if the oscillator connections on the 8010 could be paralleled between multiple eg002 boards with one crystal feeding all producing synchronized output waveforms? If that answer is yes, could the same be done to parallel two pj inverters ...kind of like 2 grid ties working together of course voltage feedback would have to be tuned so they shared loads at (=) or desired ratio.

Quote
 . "I was thinking that it gave me some insight into how the egs002 board was functioning. "

Heres another schematic of similar setup that.... might? Be closer to example you are looking for. Not sure on attachment/posting link dos and donts. Please yell if i need to remove.

 http://www.lz2gl.com/power-inverter-3kw/

Again, am very thankful for all the thoughts and test runs!

Lighthunter

[ClockmanFrance]

Thanks Oz, yes the world is full of strange folk........

Sorry, the concept of me walking behind a man with a red flag, "the protagonist", and doing what he tells me to do, especially without hard proof  .........  nope, the World moves constantly forwards..........

Okay......Still on the New Power Board, tracks masks finished, quadruple checked. But will check again.

Doing the mask for the component outlines, values & numbering and heatsink position etc. I will do this mask in a light colour so you can scan from, nothing to see if I do it in white......

I will put the resistor values on each, it helps with assembly I think. Also I will tin all the copper as might need to do some extra stuff.
 Track pads are a size that gives a good solder joint both sides, but not to big or to close to cause problems.

Nothing worse than drilling a board and finding the pads coming away, because the pad is small and the adhesion is weak. The drill always pushes if blunt, and tears if too sharp. Copper/ brass... reduce the rake angle on the cutting faces of the drill.

The board is now finished at 282mm long.

Heat sinks for this board.
 http://www.aliexpress.com/store/product/160-80-26-9MM-high-quality-aluminum-heat-board-electronic-radiator/1522105_32482586868.html

These seem a reasonable price, and several folk stock these at this size. You will need 4off, 2 will need joining end to end for a single 266mm long. Each of the 4 is 160mm long, 80mm high, and 26mm wide, so cut a bit off the 6mm thick excess and use that, only needs a 20mm strip and 4 small bolts/screws drilled and tapped.

 The mounting bit for the FET's is 6mm thick, but the fins are more than PJ, but thinner, so the mounting holes on the New Power Board will probably need moving closer to the thicker bit, but I will asses when I get them here and alter the Board accordingly.

I note MaryB mentioning that more copper can alter Inductance, so I have kept track ways etc, pretty much the width of the original within reason, although some PJ's are scary.

That big muddy red cap in the middle is ordered at 4.7uf, although on the original it was 1uf.

Give me a few days yet, (I have other jobs on now the weather is getting better, ha ha), and I will post on Fieldines, I can achieve 512k image size there.
Unless...... some Mod here can give me an email address, then they can have the 4off masks, each at 600dpi 2meg, Jpg or PDF, and post them for us on this site.?

 

[frackers]

As well as still waiting for the egs002 boards to arrive I've now ordered some IR2110 chips having discovered that the $3 stm32f103 dev boards I have been playing with the last few months not only have complimentary outputs from the PWM channels but have programmable dead-time from 125ns to 125us. Basically all the hard work is done in the hardware.

Having brushed up on my sine wave synthesis its just a matter of scaling the maximum on time to the voltage feedback and using the current feedback to decide when to shut down!!

Does anyone know the best algorithm(s) for soft start and stop so as to avoid spikes with a toroid output device? When I get the egs002 board I'll knock up a test jig with a 15V 20VA transformer backwards to I can run everything from my bench PSU and get a logic analyser on the 4 pwm outputs from the eg8010.

The way I see it, the worst that can happen if the CPU is doing extra stuff (e.g. updating a display) is that the response times to transients could be compromised. Having said that, with a 72MHz clock and DMA capable SPI and I2C, I'm sure I can squeeze all sorts of things into 64k of code space

And of course it will get put up on github under a GPL license like most of my stuff https://github.com/g8ecj

Guess I've got too many projects on the go again and the walnut harvest is about to start!!

[oztules]

The test board as shown is the one i am using, but forgot to put in 2 changes to it

I still advise NOT to use this pattern, as it is very sloppy in track clearances and placement is difficult compared to the real estate it uses.. but some won't be told, so here are blow ups of the board as it was.. but with two changes.

When I first tested it, I blew up the voltage regulators... as mentioned somewhere back there.. it was because the zener went to the wrong track by 2mm. Changed that, and the voltage feedback didn't work because of no ground link.... fixed both in a few minutes, and it has since been running the house till now... more than a week I think... works flawlessly for a botch board.

It was not until I took it apart that I realised the 2 changes I forgot about in the excitement.... as I had not touched it for all that time.

So here are the blow ups of the changed board.... no other improvements... just the two changes... I would prefer you not to use this until we get a nicer more robust version ( physically), but some folks can't wait apparently.












It really is a crap board on reflection, but it works so well... I havent though about it any further.
I will tidy it up and make it easier to build, it is cramped and silly at the moment.

.................oztules

[oztules]

Like the heat sinks.

 Mary is correct, but at these frequencies, I have found no difference with ribbon length and performance... which tells me that we need to strengthen the tracks, rather than be frightened... remember the 15kw boards are twice as long as the 8kw boards, so tracks are much longer and different... still work fine... so get the engineering right, and I suspect the differences will not matter.

The reason I get things done, is I don't procrastinate about things I don't understand..... I do it......... then see what happens, and learn what works from there.... modeling some stuff is near impossible, but suck and see works every time... you will get results, and they will tell you which direction to head in.

The idiot monetary policy in the world today done by genius modelers, give you some idea how wrong models can be. .... facts don't lie.

If your screen printing, then populating the board is much easier if you label the bits.

"That big muddy red cap in the middle is ordered at 4.7uf, although on the original it was 1uf."..... will have to check mine, but memory says 4u7... now I'm worried..... My 002 boards are running 2x2uf as well.


...............oztules

[oztules]

Frackers.
"Does anyone know the best algorithm(s) for soft start and stop so as to avoid spikes with a toroid output device? When I get the egs002 board I'll knock up a test jig with a 15V 20VA transformer backwards to I can run everything from my bench PSU and get a logic analyser on the 4 pwm outputs from the eg8010."

When you get the 002 in your hot little hands, the scope will give you the answer on how to do it , and how not to do it.

The start up is a simple3 second increase in the amplitude, ie the fundamental is held constant 50hz square wave, and the spwm is increased from zero to the final pulse width for the load at turn on at the voltage we need.

So by cutting out the 393, we get smooth 3 second start, and instant stop with no trialing pulses... and all is well now.

Provided you keep the max amplitude within a reasonable range, it is next to impossible to kill the thing  from shorts or current o/l if you have the fets to cover it.... and 6 x 4110 seems to do that.

If you run the pj or the 002 without feedback, it kills nothing... so does not go over the top when an under voltage is recieved by the ac feedback.... but limits itself and stops after a second or two... it runs without current feedback very well too, but short may push it too far... not tested yet... been having too much fun with it working good... will destry it for real if I can next week... but reasonably confidant the AC short will be survived just fine.

The cap omn the AC feedback tells me you will have plenty of time to do house keeping, and the 4uf on the CT feedback tells me the same thing... just keep it short.

I would like to see you do just a chip that does the spwm, and leave the fancy stuff to add on components, so that the very best performance is assured.... of course that will not be a challenge for you, so you will do both, but I would have my hands full just with the spwm, simple jump table that it is.

The next step would be varying the frequency as the load increased, ie insert more table values as the load came on, to keep the waveform cleaner.
I would not bother unless I had some valid reason to do it... cant think of any other than bragging rights from a practical perspective.

Will follow this carefully Frackers.



.......oztules

[oztules]

Lighthunter,

That circuit is a better version of what I am doing.

I say better as the driver is the difference, but seems not to be warranted for run of the mill use.
Finally measured the idle power at 30 watts for this one, so losses are very small, and I can't justify the extra complexity for no material gain.

Folks are more likely to do my style of board as there are hardly any real parts, has low idle power, and massive output power..... what more can you ask form a simple through hole board with a dozen or more parts.


...............oztules

[oztules]

The weakness in this version of experimental board is the rudimentary power supply. The 7805 is prone to failure in further bench testing.... and that takes the chip out.
It didn't happen in the first board as I must have been lucky at turn on. In the pj board, the caps must have slowed the inrush currents enough to protect it, but a second one on the bench into a small transformer with little filtering failed a few times and took a few boards with it.... so I have run out of boards for the time being, and won't risk the one on the big transformer, so bench testing will stop until new victims arrive, and a new board with better power supplies and other changes are done.

I'll keep the big one running until they arrive, just to continue the power testing.. but it looks strong once up and running so there is much hope.

in short don't do that board unless you better the power supply.

The next board will have plug in chips so I can be more gruesome in the testing of it.


................oztules

[frwainscott]

Oz, Speaking of few parts, in the early 70s popular electronics came out with an article on building your own 120 volt inverter. it used a 120 to 24 ct ei xfmr which was very easy to come by. and the drivers were two 3055 darlington transistors the oscilator for the 60 hz was an ne555 ic. I bought the parts and built one, and used it for years to run my electric razor in my car on the way to work. It would'nt power much But it was very stable surprisingly. So, much can be done with little. Back then an inverter was beyond most peoples means. Why not use a few parts and build an integrated 12 and 5 volt power supply that can't be killed?

 Heres To Later Days and Better Lays....  Frank

[RFburns]

Some of the boards use a different driver chip (mine have 2110 and 2113 drivers on two different boards) same chip as far as pin outs but different ratings. I am not at all suggesting that this is/will cause any problems and only wish this to be noted in case someone gets a board and thinks that they may have got the incorrect version.

 A few places have these boards very cheap  ($7.85 free shipping) ;also if ordering these boards they are supplied default with under voltage protection but you may ask for other options depending on the seller (some do not offer this and the board is supplied with default).

I am thinking stiffening up with a couple of 100uF caps (overkill) maybe beneficial to both the 12v and 5v supplies and a limiting resistor on the input so they dont see the spike at turn on. Any way looks like its working great.RF

[frackers]

also if ordering these boards they are supplied default with under voltage protection but you may ask for other options depending on the seller (some do not offer this and the board is supplied with default).

I am thinking stiffening up with a couple of 100uF caps (overkill) maybe beneficial to both the 12v and 5v supplies and a limiting resistor on the input so they dont see the spike at turn on. Any way looks like its working great.RF

Looking at all the published circuits for the eg8010, they all seem to monitor primary current and secondary voltage so no way of shutting down on a flat battery. Is this really the case or am I missing something? I would prefer to measure current and voltage on both sides of the power transformer...

I've been thinking about the specs for the bootstrap capacitor between Vs and Vb on the IR2011 and I reckon there is the potential for a lot of current there since it will be supplying the current to charge the FET gate capacitance. Any thoughts on a low ESR type here? Also steering diodes on the FET gates to improve turnoff - can't think that 1N4148 are going to help much - maybe something more like a fast 1A schottsky (e.g. 1N5819) to keep the voltage drop down.

[oztules]

Frackers
Schottky is good... when I get some will try it.
I relented, and rebuilt a 002 out of three dead ones... found the fault that caused all the heart ache... nothing like hair line cracks to cause chaos in this type of thing

I also put 120r and 100uf in the voltage line, and that stopped the voltage reg dying prematurely.... and then used 2 tip35c for the voltage reg instead of the 7805 and 12.

So the second board is now fully functional, but  will terrorize it some more, then redraw a much better board without such thin tracks.. less chance of hairline cracks for a start.

When I do the 8010 board I will make the 20khz on the low sides only, and 50hz on the high side.... that should make the switching much easier.... these ones use high side and low side 20khz.

Frackers I have no use for shut down, so have not considered it.
The current control I use is not cycle by cycle,  I monitor secondary current only, as I want big surge.... have not been able to stop it thus far.... the way I like it. The voltage sense is filtered as well, to stop any too rapid responses it may decide on... there's 4u7 on each sense to temper any behavior.... all secondary.

If it didn't work so damn well I would go with my pwm tripple supply, and that would solve everything..... but I'm not convinced it needs solving....

It is working stupidly well... still a bit amazed....


...............oztules

[oztules]

some pics of the board as it is now.

This is the a4 paper with patten after spray with crc.



and offending board



Pic of the wavform



You can see the tip35's for the voltage reg and resistor tacked on.



..............oztules

[frwainscott]

Somehow I missed my main point.... Oz... YOU have done an amazing feat going through all of the extra BS that is embedded in their circuitry and cutting it down to the very essentials.... Truly amazing!

 For all the time and parts you have put into this project. Is there somewhere I can donate to the cause? If so.... please let me know.

 Thanks ... Frank