PIP inverter repairs and hardware modifications

[weber]

I see perhaps one potentual issue when rewiring it with litz wire that will take up more space.
...
On on the late 2016 model ... the inductor sits very close to the inverters heat sink.

Could this be a issue?

Yes it could. Well spotted.

Fortunately, Coulomb convinced me last night by email, that my original suggestion of 100 strands of 0.2 mm will be sufficient to carry the maximum continuous current of the inverter without overheating (dissipating 10 W at 21.7 A), and that any greater number of strands is likely to completely fill the hole in the middle of the toroid before achieving the required number of turns.

But even with the 100 strand wire, the inductor will probably end up a little larger than it was.

I just measured the clearance that you mention, as 4 mm. There is a clearance of 10 mm above the toroid, so I think we should offset it upward to maintain that 4 mm clearance if possible, before we epoxy it in place.

[Northland]

Could one of you explain slowly the theory here? I understand the inductor filtering out the frequency part. But why is the factory way wrong / inefficient? Is it cost issue? Space?

I have a spare inductor so I could replicate this

[andys]

I'm curious what you will discover!

Every inverter I've seen (that can do >4kW) wastes alot of power when idle, except the fancy models which seem to be able to do dynamic power down of a high power circuit until it is needed.

[weber]

Could one of you explain slowly the theory here? I understand the inductor filtering out the frequency part. But why is the factory way wrong / inefficient? Is it cost issue? Space?

We might well be missing something, and the factory method might turn out to be almost as good or better. But the theory we're basing this on can be read here:
http://en.wikipedia.org/wiki/Skin_effect
http://en.wikipedia.org/wiki/Proximity_effect_(electromagnetism)
http://en.wikipedia.org/wiki/Litz_wire
Litz wire does cost more and take up more space.

[offgridQLD]

After looking at my two pip4048 inverters side by side today. I am sure cost is playing big part in the design and material choice of the inverter.

This new design with the little heat sink for the SCC jammed between the two inverter heatsinks has cost cutting written all over it. I see no other reason for the redesign.


Allthough the pip works well as a standalone (offgrid) inverter perhaps when it was first designed they were thinking the pip would be used more as a grid conected battery backup power. If you have the grid feeding the pip overnight that 50w vs say 25w idle consumption isn't so critical.

The results will be interesting.

[coulomb]

[Moderator note: This is a follow-up to a post by Coulomb in the PIP-4048MS and PIP-5048MS inverters topic.]

Here is the power flow topology for the PIP-4048, from "Axpert MKS-4000/KS-5000 Service manual", which I found on the web:



As actually implmented:



The battery is at the left, at the terminals marked P1. The AC loads appear at the right, at the terminals marked L and N (Line and Neutral). Note that one of the "switches" is actualy a diode, and the buck switch is actually in the bus negative rail, rather than the positive rail.

The two full bridge converters at the left convert the ~50 V of the battery to ~400 V. This flows through inductor 15 and S12 (S10 is off when inverting) to Vbus. The "output capacitor" I mentioned in my previous post is marked Vbus. The full bridge on the right converts this ~400 VDC to 230 VAC. The full bridge appears to be a generic diagram, not intended to represent specifics of the PIP inverter (which uses IGBTs not MOSFETs, and one not two large inductors L1/L2).

So the question remains: what happens if a grid interactive inverter is connected across the PIP's loads, and generates more power than the loads absorb. Again, Vbus will rise in voltage, but now there is a power flow path from Vbus to the battery. I don't know what will happen. My guess is that the PIP will prevent power flow, even unintended power flow, from Vbus to the battery (in battery mode). If it did allow this power flow, then all of the excess power from the grid interactive inverter (and there could be many kilowatts of this at times) will flow to the battery, with no way of regulating the power flow or the battery voltage.

So my answer remains much the same: I can't see a way of making this work unless there is a good system for ensuring zero power is "exported" to the PIP inverter. Even then, I'd be dubious that it could be made to work, but there may be a way I haven't thought of.

[ Edit March 2021: see also this post on another forum for a more high-level, more complete block diagram. ]

[ Edit: Inductors 13 and 15 seem to be combined into a single inductor L1 in place of 15. Switch S10 does not exist, and S12 is a paralleled pair of either MOSFETs or IGBTs (Q31/Q32) in the negative side of the power supply, not the positive side as shown. Inductors L1 and L2 of the diagram are combined into a single inductor L4 on the actual hardware. ]
[ Edit: The large (470 μF 500 V) electrolytic capacitors (C40, C41) are on the high voltage side of the DC/DC converter, in parallel with two 0.47 μF film capacitors (C32, C34) near the associated IGBTs. The 230 V inverter IGBTs have two film capacitors, 0.68 μF and 0.47 μF, across their DC bus, which is separated from the actual DC bus by the buck transistor(s). The large film capacitor C33 (20 uF) is part of the 230 VAC line LC filter, along with L4. ]
[ Edit: Added second diagram for comparison. Removed the "suspect buck converter is back to front"comment. ]

[paulvk]

More on the SCC:
I like semiconductors in my own designs to not get hot and definitely not so hot that I can not touch them, the heat sink on the SCC gets very hot with no forced air cooling so I sat fans on top of two of my units even at low speeds the fans keep the heatsink to warm.
So not wanting to run them 24/7 as they do at present and only when the SCC is active I have been looking at the SCC board to see where I can get a signal or power.
I have found 2 ways:
1. connect to relay coils and run 12v fan from there but involves removing from unit!
2. there is a 6 pin header on the board for CPU debug with + & - volts on it so use this to run an opto coupler (for isolation) then switch a transistor on to run the fan when SCC has power.

The connection from the inverter to the SCC is isolated with 3 opto couplers and is serial rs232 should not be to hard to record what is going on!
Note there is an unused serial & panel socket on the board.

[coulomb]

When looking at the power flow topology diagram I posted (five posts ago now), it struck me that it looked rather like a diagram you'd find in a patent application. I did a quick search, and it looks like the diagrams come from this patent, though I'm not sure that it applies to the PIP-4048:

https://www.google.com/patents/US20140268892



I find it confusing. They seem to be claiming a buck/boost capability with S12 and S10, but I only see boost (taking power flow from left to right). [ Edit March 2018: It's actually buck only, only used when power flow is right to left (charging battery from AC input). ] They mention a resonant circuit, but I only see Ls and Cs that are not meant to resonate. Still they might have something novel there; I'm no expert on these converters.

[weber]

What's that you say? You think I might be voiding my warranty?

[Adverse Effects]

ROFLMAO

[offgridQLD]

Weber, I just read through the PIP4048 user manual cover to cover and no were did it say not to cut the inductor off the main board with a power saw.

so your warranty should be fine.

Kurt

[weber]

Hee hee ha ha. Well I tried to desolder it. But it has 6 pins! 4 are mechanical only, but that makes it as difficult as an IC. You have to melt the solder on all the pins at once. At least with an IC they are close together. The inductor's two electrical connections, and the tracks they are soldered to, just sucked up whatever heat I threw at them. So I got out the micro butane torch. I almost set fire to the PCB. I certainly caused it to char and bubble. The inductor never budged. I think its base board is glued to the top side of the PCB by the conformal coating, and the black stuff that glues the winding to its base board is a thermoset, as hard as a rock. I tried poking it with a hot soldering iron. It didn't notice.

But I was determined not to let the bastard beat me.

Hence the sabre saw.

I won!

[paulvk]

Could you have a look at the FETs on the small heat sink near the DC connection to confirm they are really 60V working.

[weber]

It's a pair of T225-size cores glued together. POCO part number NPF225040. Overall dimensions 57.2 mm OD, 34.8 mm ID (difference/2 = 11 .2 mm), 29 mm wide (14.5 mm per core). 50 nH/turn². 64 turns, in 3 layers. Approximately 6 m of trifilar 1.45 mm dia wire, 1.65 mm² (15 AWG). [I earlier thought it was 1.29 mm dia, 1.31 mm² (16 AWG)]. Datasheet

Could you have a look at the FETs on the small heat sink near the DC connection to confirm they are really 60V working.

Yes. IRFB3006. I understand they are only used to protect the inverter from a reverse polarity battery.

[paulvk]

When I get the main board out of this inverter I will have a look and see how the FETs are connected with only a 60v rating if they are always across the battery then they a flying very close to the sun!

[coulomb]

50 nH/turn². 64 turns...

The datasheet says 50 nH/turn² for each core. From this page, the formula for inductance of a toroid is

L ~= μN²A/(2πr)

where A is the cross sectional area. So with twice the area, for the same permeability μ, number of turns N, and radius r, you'd expect twice the inductance from two cores. That's 100 nH/turn², or 100 x 64² = 409,600 nH = 410 μH. That's reasonably close to my measured 390 μH.

I note that I was way off with my distance estimation, coming up with 4.4 m. But that was based on 1.31 mm² wire, and Weber measured it at 1.65 mm². That would increase my estimation by 1.65/1.31 to 5.5 m, not too far from his approximately 6 m. I recall the first time I measured the wire, I found a higher diameter than the 1.3 mm Weber claimed, but did it again and found 1.3 mm, so I took his figure.

The wire may not be exactly 15 AWG; China is a metric country, and odd AWGs are uncommon. So it might be 1.5 mm diameter, or 1.4 mm.

While poking about on my spare main board, I noted that the battery side converter's transformer and inductor are both wound with litz wire.

[ Edit 19/Jan/2017: rewrote the first section calculating expected inductance, as I had made such a hash of it. ]

[Northland]

My pip died again. A minor short circuit elsewhere in the system caused this on the scc board:


Do you think it's just a case of running a wire from one side of the board to the other?

[offgridQLD]

As long as nothing more was damaged.

I have never liked the ridiculously tall standoffs used to transmit power down to the main board.

What kind of protection down/up stream did you have? Protection was rated to high to do it's job or the connection is underrated. Probably the latter.

That said go over all connections when you get a new pip as both my units has several that needed tightening.

[coulomb]

My pip died again. A minor short circuit elsewhere in the system caused this on the scc board...

That black carbonated mess needs to come completely off. It can be conductive, which would reduce the isolation between copper pads that may have 450 V on them.

Do you think it's just a case of running a wire from one side of the board to the other?

You've answered this yourself; no it won't. There are already many vias connecting the top and bottom copper areas already, so adding a wire won't help.

[coulomb]

When I get the main board out of this inverter I will have a look and see how the FETs are connected

I traced that part of the schematic a while back [ Edit: this may not be the MOSFETs you are talking about; sorry for the confusion if that's the case. ] ; I reproduce the left corner of that diagram here:

Reverse protection sch.png (5.79 KiB) Viewed 3790 times

The four reverse protection MOSFETs are in parallel. So they see battery potential, but only when you connect the battery the wrong way. At that point, it's unlikely that the battery is near 60 V, so I guess that's not too bad. As soon as the battery is present and the right way around, it's a low impedance, with only millivolts from drain to source.

Edit: but as I look at the schematic above, it seems to me it would never start, so there must be a bit more to it. But the essentials would be correct.
Edit 2: It starts due to the integral diodes in the MOSFETs. Changed schematic to show these.

[paulvk]

So they are always across the battery does that not mean that their working voltage will be exceeded if the battery goes above 60v!

On another note I came across an interesting power meter on e-bay
100a 120v wireless with color lcd display just search on
"DC 120V 100A Wireless Volt Ammeter Power Meter"

[offgridQLD]

Guys if you haven't already ordered the wire from that eBay seller listed I would perhaps consider not doing so.

I could have have just annoyed him by asking him to change my order to a different spec. He was Ok with that at first (so he should be as it was listed for less money)

Then asked him to also change it to 30ft instead of 10 offering to pay in full the extra necessary. He insisted that now the cost would now be $30 for very 10ft of wire and I would need to pay an extra $60 on top of the $22.70 I had already payed him total $82.70

Real cost should have been $51.86 an extra $29.16 not $60

On top of that I then get a message saying he has sent just 10ft of the original order 35/32 size.



Kurt

[coulomb]

So they are always across the battery does that not mean that their working voltage will be exceeded if the battery goes above 60v!

No, the reverse protection MOSFETs are in series with the negative lead of the battery.

[weber]

Hi Kurt. I didn't see your post before I placed my order. But you can't really fault them for having promptly sent off your original order, can you?

I just ordered 6 m (20 ft) of the litz wire consisting of 100 strands of 32 AWG (0.2 mm). I ordered Qty: 2, Length: 10FT, and included a message saying "Please supply as one 20 ft length". I confirmed with them beforehand, that they will honour this request. Total AU$47.28. It will take 3 to 5 weeks to arrive (free postage).

[offgridQLD]

That sounds like it should begin fine.

I think I will just give up on emending that order and just add the 10ft of 32/35 the the parts bin in the shed for another project one day.

If your results are sucsefull then I will reorder the same under my wife's eBay acount as he has it in for me now .

I reversed the direction of the two stock fans In my 2013 pip4048 today. Sitting idle most of the day it actualy feels like there is more hot/warm air flowing out of the case now.

As in placing your hand near the top left side vent you can feel heated air on your hand where befor with the fans where blowing down I don't remember there being much not air exiting the bottom. Vents.

Perhaps this is just because the air is flowing over the hot Inductor then directly out the top side vents.

To me the black heat sink on top feels warmer to than when the fans where blowing out the bottom.

Kurt