PIP inverter repairs and hardware modifications

non-EV Solar, Wind and other renewable power sources
non-EV batteries and other energy storage stuff
Forum rules
Important!
This forum is for discussion of Non-EV matters.
Post Reply
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

The litz wire just arrived for the PIP PWM-blocking inductor.

It gives me some satisfaction that although I was forced to order it in cave-man units, as 20 feet of 32AWG x 100, it arrived on a spool with the only writing being "0.2x100, 7m", hand written.

Fortunately its overall diameter is about 2.9 mm, as I estimated it would be, from a table given by another manufacturer. See PDF page 7 of
http://edge.rit.edu/edge/P10022/public/team_docs/Wires_Casing/NEWENGLANDlitzSpecs.pdf
I allowed for 3.0 mm dia in my winding diagram and experiments, so hopefully all will go to plan.

It appears to be made as 5 bundles of 20, which is OK, so long as there is some randomisation of the 20 in each bundle, so that no strand spends too long in the middle.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

I still haven't rewound the inductor, but I've planned out how to do it. One problem is that the insulating enamel on the litz wire is only half the thickness of the original. So although I need to have three layers of winding in the middle of the toroid to fit all 64 turns, I don't want turns with a large voltage difference to touch one another. And in particular, I want a decent gap between the two ends of the winding. This has the added benefit of minimising inter-winding capacitance which might otherwise allow some PWM to bypass the inductance.

I came up with the following scheme that you might call "3 steps forward 2 steps back". The turns are all parallel to each other as they pass thru the inside of the donut, but they criss-cross like crazy on the outside, as you can see in the "unrolled" view on the right. No wires that touch are more than 5 turns apart.

Only one half of the winding is shown. The two halves are identical. It makes sense to start at the middle of the winding and hence the middle of the length of wire, so that you only have to pull a maximum of 3.5 metres of wire through the hole each time, instead of 7 metres.

Image

I thought I'd post this now, so after I try to put it into practice, and post a photo of the result, you can all have a good laugh about "the best laid schemes o' mice an' men".
Last edited by weber on Fri, 10 Feb 2017, 19:34, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Re: PIP inverter repairs and hardware modifications

Post by offgridQLD »

350 attempts later .......Image

Perhaps it just looks visually tricky in the drawing. Hopefully a slow and methodical approach gives you the outcome your looking for.

The results will be interesting.

ps, I did get my wire in the post, though just 3m of it.
paulvk
Senior Member
Posts: 457
Joined: Fri, 23 Oct 2015, 23:45
Real Name: Paul
Location: Sydney

Re: PIP inverter repairs and hardware modifications

Post by paulvk »

Why not put some high temp tape between layers to improve insulation, it also makes winding the next layer much easier I have found.

On another note with all this hot weather I have been running my one inverter system with 61 amps shown on the solar input even had three AC units running at once on Friday for about 4 hours 3000watts, ambient temp was 32C and 45C on the heat sink with both 100mm fans (xtra cooling) on the sides at the top going.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

offgridqld wrote:350 attempts later .......
Ha ha. Good one.
ps, I did get my wire in the post, though just 3m of it.
Sorry I messed you around with this, Kurt. But it turned out to only need 5.9 metres. And you got the 0.2 mm x 100 strand litz wire, right? So if it works, and you decide to do it, you can just buy another 3 metres and solder them together and heatshrink the join, and start the winding with the join on top.
paulvk wrote:Why not put some high temp tape between layers to improve insulation, it also makes winding the next layer much easier I have found.
Sounds like a good idea, Paul. But I enjoyed the challenge of designing a winding that would also minimise interwinding capacitance. And I'd already wound it by the time I read your post.

You can see the result below. I think it turned out quite well. That's a piece of 3 mm polycarbonate epoxied in at the bottom as a separator. The biggest problem while winding it was the roll-curl. It made the wire want to kink as I was pulling it through, for the first 15 turns or so on each side. So next time I'd try to overbend it in reverse as I unwind it from the roll, to try to straighten it before starting winding.

ImageImageImage
Last edited by weber on Sun, 12 Feb 2017, 09:17, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Re: PIP inverter repairs and hardware modifications

Post by offgridQLD »

Looking good,
             Not as jammed packed as I was expecting. Good to know that joining two rolls of wire is OK.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

Coulomb and I installed the rewound inductor and reassembled the PIP today. Here are some photos.

ImageImage ImageImage Image

Sadly, we have to say it really wasn't worth the trouble. It only runs about 10 degrees cooler. We measured the hottest spot at 89°C after running with no load for over 2 hours. The previous version ran at more than 100°C.

We also measured the power consumed from the battery with the inverter running, but no AC load, at 69 W. Unfortunately we didn't measure the consumption of this specific inverter before rewinding the inductor with the litz wire. But the received wisdom was that it was, in general, 50 W. However, it seems very unlikely that we've increased the no-load power consumption at the same time as making the inductor run a little cooler.
Last edited by weber on Mon, 13 Feb 2017, 19:19, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Re: PIP inverter repairs and hardware modifications

Post by offgridQLD »

Thanks for giving it a go guys!

I appreciate the effort and expense involved. Bugger looks like there is more to the PIPs 50w baseline metabolism Image

I have measured two pips as just under 50w (48w) with a little shunt based dc power meter. 69w is 40% more consumption.

Perhaps the new inductor runs cooler as it dissipates the heat better due to better air circulation around the wire surface.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

I think the exposed surface area of the winding is much the same as it was before. It might even be less than it was. So although the high no-load power consumption of the inverter as a whole may be a mystery, I'm pretty sure the lower temperature of the inductor is due to lower power dissipation in it.

It seems likely that most of the remaining losses in the inductor are hysteresis losses in its core. If it happens to be going a little too far towards saturation, it might benefit from removing a few turns, otherwise such losses may be unavoidable. By the way, the inductance measured the same as the original, using Coulomb's meter.

I expect the other major loss in the inverter will be hysteresis loss in the core of the high frequency transformer used in the 50 V to 400 V bidirectional DC-DC converter. But its temperature can't be easily measured with the SCC in place over it. Nothing can be done about that loss either, except such tricks as running a lower PWM frequency at low loads, which may result in annoying audible tones coming from the inverter.

It's not inconceivable that by reducing the equivalent series resistance of that inductor at PWM frequencies, I've increased the losses somewhere else. But I can't think of where that might be. Nor can I think of a mechanism for it.

It's possible the Voltronic engineers knew exactly what they were doing in not using litz wire for this one inductor, or perhaps they tried the same experiment I just did, and got the same result.

[Edit: Axpert engineers -> Voltronic engineers]
Last edited by weber on Tue, 14 Feb 2017, 07:25, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

I have a bigger problem than inverter power losses right now. I have the loss of an inverter. Sigh.

When I turned on the 1200 W vacuum cleaner for about the fifth time, to remove the dust from drilling the holes for the rewound inductor in the spare PIP, I heard a pop from the PIP that was powering the vacuum cleaner, and another pop a second later. So I hit the big red button that disconnects it from all energy sources (battery, solar and utility). And after establishing that there was no smoke or flames I powered it up again briefly and saw the familiar fault code 9 which is "Bus soft-start failed".

So after reassembling and testing the spare PIP (which is now the main PIP), we disassembled the blown PIP and found the usual failed MOSFETs on the battery side of the DC-DC converter. All eight MOSFETs in one half-bridge (in parallel groups of 4) had their gates shorted to their drains, and had consequently blown several components in their gate drive circuitry as well.

It's nice that they have upgraded from the earlier IRFB3307 to IRFB3077 that have nearly half the on-resistance, and a pulsed current rating of 850 A instead of 510 A, but they are still only rated for 75 V! I really don't think their problem was ever their current rating. They need to use 100 V MOSFETs.

I earlier recommended the 100 V IRFB4310Z and installed them pre-emptively in one of Kurt's PIPs, along with higher-voltage-and-longer-life capacitors. But they only had the same on-resistance and pulsed-current ratings as the IRFB3307. I feel obliged to match, or nearly match, the ratings of the IRFB3077 now, in a 100 V device. I'll let you know what I come up with.
Last edited by weber on Tue, 14 Feb 2017, 07:18, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Re: PIP inverter repairs and hardware modifications

Post by offgridQLD »

Hopefully some good luck from now on.

That upgraded pip of mine has to deal with a lot surge loads as its main duty now is powering the shed. Particularly the car hoist the spike is significant more so when started under a few ton load. The 16cfm air compressor and a few other things give it a real workout.

Will probably go pop now I have talked it up Image
Last edited by offgridQLD on Tue, 14 Feb 2017, 07:32, edited 1 time in total.
andys
Groupie
Posts: 78
Joined: Wed, 20 Jul 2016, 19:44
Real Name: Andrew Snow
Location: Sydney

Re: PIP inverter repairs and hardware modifications

Post by andys »

ouch Weber. how old was this unit?

Any idea how exactly would they be seeing anywhere near 75V?
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

andys wrote: ouch Weber. how old was this unit?

Any idea how exactly would they be seeing anywhere near 75V?
Hi andys. It was built in October 2015, but it has only been in use for about 6 months. Admittedly a hot 6 months. I suppose that could have aged the capacitors (Jamicon WL-series 3300 uF 63 V 105°C 2000 hour) sufficiently that they were no longer able to adequately control voltage spikes on the MOSFETs.

I don't know for sure that they are failing on overvoltage, but when you look at their current specs and the fact that there are 4 in parallel, it's very hard to see how 4 MOSFETs capable of 150 A each could be over-current on the start surge of a 1200 W vacuum cleaner, even if the start surge was 10 times that power. Whereas a MOSFET would only need to have more than 75 volts on it for 10 microseconds to take it out. So I am imagining a voltage spike caused by the sudden change in current through stray inductance between the MOSFET and its DC bus capacitors, or possibly the voltage control loop undershooting followed by overshooting.
Last edited by weber on Tue, 14 Feb 2017, 11:54, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

Here's my post from 2 years ago, about upgrading battery-side MOSFETs and caps.
http://forums.aeva.asn.au/viewtopic.php?title=pip4048ms-inverter&p=56103&t=4332#p56103

Here are five low-cost 100 V MOSFETs that may be suitable for replacing the existing IRFB3077s. None are more than $5.00 AU or $3.50 US. Most are available from more than one of Digikey, Mouser, Element 14 or RS Components.
MDP1921, TK100E10N1, AOT290L, CSD19535KCS, SUP70040E.

My search has been far from exhaustive. I haven't looked at Mouser at all, and my efforts with Element14 and RS Components were cursory. But I need to get back to work on other things, so I'd be pleased if anyone else can suggest other options, or tell me why any of those above are not suitable, or which they think is the best choice.
Last edited by weber on Sun, 19 Mar 2017, 09:48, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
paulvk
Senior Member
Posts: 457
Joined: Fri, 23 Oct 2015, 23:45
Real Name: Paul
Location: Sydney

Re: PIP inverter repairs and hardware modifications

Post by paulvk »

I found these caps 3300uF SLPX332M080A9P3 @20KhZ ESR-0.076 Ripplel-4.04A 80V 100V surge size 22 x 50 3000h at 85C & full ripple element14 1604359 I am thinking of using them.
I have had 3 AC units running off my one PIP system with SCC at 61amps during this time a 1hp pump ran twice as well with no problems so its strange that a 1200w load caused a problem.
I wonder if the lead acid wet cells 225Ah T105s also soak up spikes have less than 1 meter of 50mm welding cable in each leg.
Would lithium with BMS have the ability to absorb spikes?
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

Hi Paul, I agree a 22 mm diameter cap will just fit, although the existing are 18 mm diameter. And height isn't a problem. And they have the right lead spacing (10 mm). And 80 V is a good rating. Ripple current close to the original. I note that every rated hour at 105°C is worth about 4 hours at 85°C. But the real killer is that their ESR is about 5 times greater than the original. It is the ESR, or more precisely the high frequency impedance (which includes the effect of ESL), that is important in this application. The capacitance doesn't really matter. The 80 V caps I put in Kurt's PIP were only 1800 uF, but they had the required low HF impedance.

It's good that we have the option of up to 22 mm diameter now. When I did Kurt's the pairs of caps were so close together there was no option other than 18 mm. So it should be easier to find an 80 V cap with a HF impedance of 15 milliohms or less, now.

The LiFePO4 (160 Ah) battery has a low internal resistance (about 16 milliohms), and we use double 50 mm² cable to connect it, with the battery in shelves directly below the inverter.
Last edited by weber on Tue, 14 Feb 2017, 16:45, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

Re: PIP inverter repairs and hardware modifications

Post by offgridQLD »

Could there be something particular about that vacume cleaner load that caused the voltage spike. The few vacumes I have pulled apart look to have very crude motors.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

offgridQLD wrote: Could there be something particular about that vacume cleaner load that caused the voltage spike. The few vacumes I have pulled apart look to have very crude motors.

Yes Kurt, it could have been brush arcing. That inverter ran that vacuum cleaner many times over the past 4 months. Coulomb also pointed out to me yesterday that there is an element of Russian-roulette about where the 50 Hz sine wave is at, when the motor is turned on.

I should also mention that we were running firmware in it that we had patched. We've ported our bug-fix and LiFePO4 changes to version 72.70 and were testing it. But our patches have nothing to do with any low-level operations such as gate drive or fast voltage-control loops, and the patched firmware had been operating successfully for two days during which the vacuum cleaner was used several times, not to mention the microwave, kettle and fridge. But no other major appliances were in use at the time of the failure. The load just before the vacuum cleaner was turned on would have been 200 to 400 watts.

[Edit: A possible cause of the failure was found and described at the end of this post.]
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
andys
Groupie
Posts: 78
Joined: Wed, 20 Jul 2016, 19:44
Real Name: Andrew Snow
Location: Sydney

Re: PIP inverter repairs and hardware modifications

Post by andys »

Now pondering installing "soft start" circuits to my big loads, seems easier than replacing blown components.
paulvk
Senior Member
Posts: 457
Joined: Fri, 23 Oct 2015, 23:45
Real Name: Paul
Location: Sydney

Re: PIP inverter repairs and hardware modifications

Post by paulvk »

I built a TL494 PWM speed control for my truck radiator fans many years ago and had failures of fets (100v rated), current was around 50amps with 2 fans I put it on my CRO and saw spikes over 100v, I had some 16v 5watt zeners I put one across the fets now over 10years later under the bonnet in the heat and still going strong.
Maybe a 60-70v zener will work here?
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

paulvk wrote: ... Maybe a 60-70v zener will work here?

Hi Paul. Unfortunately the existing MOSFETs (and my proposed 100 V replacements) have zeners built in to them. They are referred to as "avalanche" diodes in the datasheets because technically the "zener" effect only operates at voltages below about 10 volts. But clearly this hasn't saved them.

They can only absorb a certain amount of energy per spike; 200 millijoules in the case of the existing MOSFETs, assuming the spikes don't come too close together. 200 mJ might not sound like much, but if the spike only lasts for 10 microseconds, that's 20 kilowatts or 267 amps at 75 volts. Which argues that any overvoltage had to last significantly longer than 10 us. Which argues for the unstable voltage-control-loop theory and against the voltage-spike theory, now that I think about it.
Last edited by weber on Wed, 15 Feb 2017, 07:25, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
andys
Groupie
Posts: 78
Joined: Wed, 20 Jul 2016, 19:44
Real Name: Andrew Snow
Location: Sydney

Re: PIP inverter repairs and hardware modifications

Post by andys »

Given the terribly slow reaction times of the rest of the system (eg. turn off appliance and it takes 10 seconds to ramp down solar input when the battery is already full and supposed to be on float charge)...

It sounds highly plausible to me that it doesn't react fast enough with sudden loads applied with unfortunate timing.
paulvk
Senior Member
Posts: 457
Joined: Fri, 23 Oct 2015, 23:45
Real Name: Paul
Location: Sydney

Re: PIP inverter repairs and hardware modifications

Post by paulvk »

Had diodes in my mosfets but they still died I even tried transorb diodes these also died but the metal stud zener lives on doing the job.
Note the zener diode is across the fet not on the gates.
The spikes may be back emf from the transformer.

"takes 10 seconds to ramp down solar input" I do not see this on my systems.
Last edited by paulvk on Wed, 15 Feb 2017, 11:01, edited 1 time in total.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by weber »

paulvk wrote:... I had some 16v 5watt zeners I put one across the fets now over 10years later under the bonnet in the heat and still going strong.

I think you'll find your stud mount zener is rated at considerably more than 5 watts. All MOSFETs have diodes built in, but not all have avalanche (aka "zener") diodes built in. And yes, I meant drain to source, as I assumed you did.
Last edited by weber on Wed, 15 Feb 2017, 11:21, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
coulomb
Site Admin
Posts: 6357
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Re: PIP inverter repairs and hardware modifications

Post by coulomb »

I'm taking apart my PIP-4048 to upgrade the MOSFETs and capacitors. It's a slog getting to the main board. The mains connectors to the main board are either very tightly pushed on, or are in fact soldered. So it's necessary to disconnect the mains wiring at the terminal block ends. To get to the lower (AC output) wires, it's necessary to remove the top terminal block.

It took me a while to find the screws allowing me to do that, and I've taken PIPs apart before. So I thought I'd post a photo showing the screws here, in case it helps someone else ,and so I can find it myself in the future.

Image     Image

After removing these two screws, the top terminal block can slide out to reveal similar screws to the above, for the lower terminal block. The lower terminal block also has to be removed.

Edit: There is also the "last screw" (nine on the main PCB) holding down the main board. It's hidden underneath one of the date stickers, as shown in the second photo. (Before removal, it's far less obvious that there is even a screw there at all). There is also one between the battery terminals.

[ Edit: lower terminal block has to be removed as well. Also added second photo and last paragraph. ]
Last edited by coulomb on Fri, 24 Mar 2017, 16:57, edited 1 time in total.
MG ZS EV 2021 April 2021. Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
Patching PIP-4048/5048 inverter-chargers.
If you appreciate my work, you can buy me a coffee.
Post Reply