REPAIR : 4 x RCT MKS2  Inverters Received from Malawi

Went to receive the boxes from Alrode destination  as fast as my feet could fly.  I will refer to the inverters as MR1,MR2,MR3 and MR4 . MR1 and MR2 had been paired as parallel units  in a single location , MR2 and MR3 also paired but different location.

 

The initial bringing up of the inverters done by current limited dc source on battery.

MR4 :  error code 09 . That's nasty , indicates IGBT and/or MOSFETS blown , but great opportunity for me  to show my experienced gained in this regard

 

Below is short video showing the symptoms of MR1, MR2. 

 

MR1,MR2 :  You cannot believe this . Exact same problem as the previous Cape Town machine. Frequent clicking sound . Couldn't resist the temptation . Immediately disassembled , and powered the main board and control board standalone. Exact same  problem . The input electrolytic capacitor on  U5 (5V regulator) has given up on life. Temporary soldered on a 470uF and problem solved.  Two exact machines in same location , same symptons , same remedy , crazy is it not !  Main boards marked 2015 .

MR3 :  This one has a major problem as far as I am concerned : it has no problem at all. So here comes the start of an issue regarding quantative  vs qualitative  measures.  So if we cannot quantify an error , we will have to use qualitative measures to ensure this inverter is good for a number of years ahead, and that we can dispatch back to Malawi with a calculated measure of confidence . This also holds true really for MR1,MR2 and MR4. 

So all in all all four inverters appears to be serviceable and I am looking forward to start with MR4 and document the findings and repair.

  

 

Edited January 28, 20232 yr by BritishRacingGreen

The New and the Old

Below on the left is the mainboard for an Axpert MKS4 5.6kw and to its right is the mainboard of Axpert MKS2 5KW , two generations earlier. Not much difference really . The MKS4 is dated 2021 and the MKS2 2015 . After cleaning the MKS2 with compressed air , it looks as new if not even better than the MKS4 .  But the MKS2 has been in operation for probable >5 years , while the MKS4 only a number months. So this is one of the qualitative issues i am reffering to : how can we ensure that the MKS2  be serviced to ensure another couple of years of performance.  Example , what can we measure to verify that the two large bus capacitors will hold , or should we proactively replace them . This is where I will be consulting Coulomb and others .

 

 

One thing I have noticed on the MR1-MR4 machines from Malawi is a massive heat sink mounted on the top of the machine. Remember I am only a 18 month rookie in solar , so I have not seen to many of the older machines. 

Turns out this is the heat sink of the MPPT module, totally different to that of other MKS2,MKS3 and MKS4 machines.

 

When I opened the first machine I went like - WOW! - this is some engineered MPPT module. Its got a presence and the quality and PCB layout is something !  Wonder if this is tooo expensive , why discontinue it ?

Edited January 28, 20232 yr by BritishRacingGreen

Hey Guys

 

I got a 3kW Axpert with blown components. I was about to order parts on Aliexpress (The WWW doesn't show me the exact part number in RSA) and wait a lifetime when I stumbled on your thread.

Would it be possible to swop out the GP4063D for a GP4066D instead?

 

11 hours ago, WindGat said:

Would it be possible to swop out the GP4063D for a GP4066D instead?

I would say so. The only problem would be if the gate drivers can't handle the extra gate capacitance of the larger GP4066D part (140 A versus 100 A for the GP4063D part). [ Edit: And if the power supply can handle the slight extra drive current; I imagine it would. ]

Does the 3000W model use the same gate drivers (ACPL-T350, marked T350) as the 5 kVA models? If so, you should be fine using the stronger GP4066D chips.

And you can take that poor GP4063D off the chopping block 😉🪓

 

 

Edited January 30, 20232 yr by Coulomb

5 hours ago, Coulomb said:

And you can take that poor GP4063D off the chopping block 😉🪓

😂

UPDATES : REPAIR : 4 x RCT MKS2  Inverters Received from Malawi

Machine MR3 : 

When I unboxed machine MR3 , it appeared to have no defect . However I decided to let it burn-in for a while , and eventually it started to click inadvertently the same as MR1 and MR 2. This is good news for me under the circumstances. So it turned out that I have received 3 machines with identical problem as the one I repaired for Cape Town . I believe the 5V supply filter capacitor had still some last breathe left , because initially the clicking cycle was slow , and gradually increased in frequency , so the last drop of electrolyte  had literally evaporated. While the fix is straight forward by now , it raises a new issue in general : How do we asses Electrolytic  and measure the capacitor's degradation , in order to asses whether we can still  revive the machine to lets say operate for 3-5 years. I will cover this in  a next post. 

So out of 5 MKS2 machines I have received so far , no less than 4  has this failure mode. How many more is out there in the wild that is failed like this , or in waiting . Interesting.

 

Machine MR4 :

Axpert Error Code  F09 is the one that's gets me going. It mostly throws everything but the kitchen sink at you . IGBT failure , IGBT driver failure , IGBT driver transformer failure , MOSFET failure etc. So I opened the lid hastily , and after going thru the standard procedure of disassembly of the inner , the main board indeed exposed destructed/exploded  IGBT's . In fact , all 9 IGBTS have blown and damaged, including the 4 on the HF DC-DC full bridge  , the 4 on the DC-AC Full bridge , as well as the Buck IGBT . I have never seen this level of destruction , typically the DC-AC full bridge but not the others . So that makes one wonder , what was the initial cause of this big-bang.?     In order to assess whether this machine is still serviceable , I will have to remove the devices , and check the quality  of IGBT driver signals. However , this machine will stay on the back burner until such time I have successfully repaired and refurbished the other three.

 

 

On 2022/10/25 at 9:13 PM, BritishRacingGreen said:

I redraw the sps power supply as shown below :

I think you'll find that C79 (at the input to the 5 V regulator) is upside down, and it probably has a voltage rating less than 50 V. I'd actually be interested to know the real voltage rating.

7 hours ago, BritishRacingGreen said:

In fact , all 9 IGBTS have blown and damaged, including the 4 on the HF DC-DC full bridge  , the 4 on the DC-AC Full bridge , as well as the Buck IGBT . I have never seen this level of destruction...

These are RCT branded, and we know that RCT is a genuine reseller. But the MKS II is a fairly old model, though now favoured by clone makers. Is it possible that RCT, desperate to get stock in difficult times, resorted to importing clones? Does the sticker look genuine?

Or perhaps these are just old. What is the manufacturing date?

It may be just the battery-side capacitors, in conjunction with 75 V MOSFETs in those days, causing the whole power train to blow. Can you see any obvious MOSFET destruction?

This is a great repair story in the making; thanks for sharing it.

Inspired by your capacitor stories. I did some pre-emptive re-capping in one of my inverters that failed recently. It seems that D53 (in series with the main power supply's battery connection) just went high resistance and open circuited. The day before, I got a fault code 80 (paralleling comms fault) which appeared to go away by resetting, but I strongly suspect that the diode was giving up the ghost at that point.

I don't see why D53 is needed at all, but I'm nowhere near confident enough to just short circuit it. Besides, it seems to have acted as a reasonable (if messy) fuse. It got hot enough to char the PCB and although it's hard to see in the photo, discoloured two tracks and even the "silk screen" D53 designator. The below is after I cleaned up most of the charring:

The diode itself was white and crumbly. It's a UF202 ultra fast 200 V 2 A part, uncommon enough that I had nothing like it in stock. They are painful enough to extract from a spare board that I thought it was worth buying a replacement, and of course a stock of brand-name decent spec capacitors to go with. These were the caps replaced:

I foolishly forgot to order a replacement for C69, a 47μF 100 V part. I thought I'd just buy a half decent one from the local hobby electronics store, but they don't have many 100 V electrolytic capacitors. Another store on the other side of town also didn't stock them. Ugh! Fortunately, when I mentioned this to Weber (I was using his desoldering station), he rustled through stock and found a 68μF 250 V Rubicon, and decided that would do, and save me paying 24x the price of the capacitor in shipping.

Edit: Bill Of Material for a typical recapping:

C78, C79: 1000μF 16V, e.g. Rubycon 16ZLJ1000MT810X16 Digi-Key 1189-4158-1-ND

C132: 220μF 25V, e.g. Panasonic EEU-FR1E221B Digi-Key P124224CT-ND

C95, C116: 100μF 25V, e.g. Nichicon UHE1E101MED1TA Digi-Key 493-17397-1-ND

C69: 47μF 100V, e.g. United Chemicon EKZN101ELL470MH15D (now obsolete) Digi-Key 565-4133-ND

C75: 47μF 25V, e.g. Panasonic EEU-FR1E470B Digi-Key P15360CT-ND

Not a smoothing part, but in the power supply circuit, for those  that want  to be very thorough:

    C7 10μF 160V, e.g. Nichicon UPW2C100MPD1TD Digi-Key 493-11888-1-ND

The above links are all for Digi-Key Australia; adjust the URLs as appropriate for your location if needed.

For completeness, in case ever needed:

C40, C41 470μF 500 V (bus capacitors), e.g. EPCOS - TDK Electronics B43545A6477M060 Digi-Key 495-B43545A6477M060-ND

C8, C12 and C9, C13: Best are still 1800μF 80V United Chemi-con EKZN800ELL 182MM40S, e.g. Digikey 565-4129-ND[/url]. Later models use cheaper 2200μF 80V in C9, C13 position. For higher power (>5kVA) models: these might not be suitable. See this 2015 post for the selection process (thanks, Weber!). [ Edit: These have 7.5 mm lead spacing; later models have 10 mm lead spacing, e.g. PIP-5048MK (Axpert King). But  they can be made to fit well; see this post. ]

Edited July 3, 2024Jul 3 by Coulomb
Added C75, C7, "for completeness" section. One cap now obsolete.

On 2023/01/31 at 2:50 AM, Coulomb said:

I think you'll find that C79 (at the input to the 5 V regulator) is upside down, and it probably has a voltage rating less than 50 V. I'd actually be interested to know the real voltage rating.

Noted thank you sir, the value is 1000uF 16V. 

19 hours ago, Coulomb said:

Inspired by your capacitor stories. I did some pre-emptive re-capping in one of my inverters that failed recently. It seems that D53 (in series with the main power supply's battery connection) just went high resistance and open circuited. The day before, I got a fault code 80 (paralleling comms fault) which appeared to go away by resetting, but I strongly suspect that the diode was giving up the ghost at that point.

I don't see why D53 is needed at all, but I'm nowhere near confident enough to just short circuit it. Besides, it seems to have acted as a reasonable (if messy) fuse. It got hot enough to char the PCB and although it's hard to see in the photo, discoloured two tracks and even the "silk screen" D53 designator. The below is after I cleaned up most of the charring:

The diode itself was white and crumbly. It's a UF202 ultra fast 200 V 2 A part, uncommon enough that I had nothing like it in stock. They are painful enough to extract from a spare board that I thought it was worth buying a replacement, and of course a stock of brand-name decent spec capacitors to go with. These were the caps replaced:

I foolishly forgot to order a replacement for C69, a 47μF 100 V part. I thought I'd just buy a half decent one from the local hobby electronics store, but they don't have many 100 V electrolytic capacitors. Another store on the other side of town also didn't stock them. Ugh! Fortunately, when I mentioned this to Weber (I was using his desoldering station), he rustled through stock and found a 68μF 250 V Rubicon, and decided that would do, and save me paying 24x the price of the capacitor in shipping.

 

Thanks for share, this is about the same locations I am going to use to replace caps. 

You will want to keep the old ones in order to maybe measure the ESR  at some time. Would be an intetesting excersize. 

Edited February 1, 20232 yr by BritishRacingGreen

@CoulombI have a somewhat offtopic question here. 

If we have a 5kw inverter, a flat battery, no PV  and  a load of near capacity. If we had set the max grid charge current to a high value, will the dsp throttle this down as to limit the overall machine usage within limits. I know that the grid will bypass to the load while charging, but surely the machine's  grid input plumbing  (wires , connections, relay contacts, pcb tracks, inductors, etc) has its limits. Or will it just give as much as it can and rely on the internal grid fuse to protect.? 

 

Edited February 1, 20232 yr by BritishRacingGreen

24 minutes ago, BritishRacingGreen said:

Thanks for share, this is about the same locations I am going to use to replace caps. 

You will want to keep the old ones in order to maybe measure the ESR  at some time. Would be an intetesting excersize. 

I dont have an ESR meter yet, so today i am busy practising one with a 100khz 50 ohm generator  and an oscilloscope, with manual calculations the hard way.🙊

30 minutes ago, BritishRacingGreen said:

I dont have an ESR meter yet, so today i am busy practising one with a 100khz 50 ohm generator  and an oscilloscope, with manual calculations the hard way

Have you checked out the link posted by Holmoe on the AEVA forum for a basic ESR meter that can test in circuit? I haven't tested it myself yet but when I have enough time I'm going to build it.

https://forums.aeva.asn.au/viewtopic.php?p=91859#p91859

 

13 minutes ago, Shadders said:

Have you checked out the link posted by Holmoe on the AEVA forum for a basic ESR meter that can test in circuit? I haven't tested it myself yet but when I have enough time I'm going to build it.

https://forums.aeva.asn.au/viewtopic.php?p=91859#p91859

 

Hi @Shadders, thanks for the pointer. I do like the idea of the impedance matching transformer design, as opposed to a solid state drive, because the latter is a challenge, so it may end up well worth to wind up a transformer, for which one can also introduce taps for impedance ranges.

The article has some interesting comments as well. I look forward as to test ESR in circuit as drscribed. 

 

For anyone else that may be interested, here is a link:

https://ludens.cl/Electron/esr/esr.html

3 minutes ago, BritishRacingGreen said:

Hi @Shadders, thanks for the pointer. I do like the idea of the impedance matching transformer design, as opposed to a solid state drive, because the latter is a challenge, so it may end up well worth to wind up a transformer, for which one can also introduce taps for impedance ranges.

The article has some interesting comments as well. I look forward as to test ESR in circuit as drscribed. 

 

For anyone else that may be interested, here is a link:

https://ludens.cl/Electron/esr/esr.html

Its intriguing that the author specify 50khz as opposed to 100Khz which seems to be be standard by which cap manufactures specify ESR. 

But at high value electrolytics that probably ok, as the reactance of the cap under test is near zero at even lower frequencies.

18 hours ago, BritishRacingGreen said:

If we had set the max grid charge current to a high value, will the dsp throttle this down as to limit the overall machine usage within limits.

I don't believe so. The hardware is rated for its rated load plus maximum possible utility charge power. It's up to the installer to rate the input AC cables appropriately. In my case for example, that bumped the AC-in cables to 6mm², while the output cables are 4mm².

4 hours ago, Coulomb said:

I don't believe so. The hardware is rated for its rated load plus maximum possible utility charge power. It's up to the installer to rate the input AC cables appropriately. In my case for example, that bumped the AC-in cables to 6mm², while the output cables are 4mm².

Thanks, the 6mm on grid side is a good call for reason, something I have not accounted for in my system design . Fortunately I typically only allow 2A grid charge, and only increase to 20A  when there is heavy loadshedding, but then I rationalise loads. 

On 2023/01/31 at 8:51 PM, Coulomb said:

Inspired by your capacitor stories. I did some pre-emptive re-capping in one of my inverters that failed recently.

I just got the repaired inverter back into position today; there is so much gear packed into the metal case that it's a pig of a job. There is still an hour's work to wire it in. It was late, so I decided I'd just turn on the already-wired inverter that has been running the house on its own for the past week, and blow me down it did the ker-clunk routine. Maybe something got dislodged, but it sure looks like the symptoms that BritishRacingGreen has been running into with the 1000 μF capacitors drying up.

I'm wondering if we'll see a even more of a spate of these as these cheap capacitors reach their end of life. Though I have to wonder about the other dozens of electrolytics. Maybe they don't dry out as fast because they don't suffer the same repetitive surge currents that the high frequency power supply capacitors do.

At least it had the decency to hold on until the other inverter was repaired. I'll just have one night of utility power, and it was a poor enough solar day that the battery would not have lasted the whole night anyway. I wonder if powering it up causes a bigger surge than running continuously.

24 minutes ago, Coulomb said:

At least it had the decency to hold on until the other inverter was repaired

😁 Most Intriguing. Yep, so this seems like the wrong end of the socalled bathtub curve of lifecycle. 

 

MicroPython : a very handy tool for testing, scripting and prototyping

Two days ago I had to generate a 100khz square wave for repairing/testing purposes. No I think I am a battle hardened C/Cpp developer for a number of decades. But you so many times the need to rapidly script something, and C is a pain, you need to create project, CMake dependancies, compile, link etc. 

Heard so many times of MicroPython for embedded devices, so i gave a try.downloaded Thonny, flashed MicroPython onto Raspberry PI Pico, and got a example PWM snippet downloaded to pico, 2 hours to setup, i kid you not 😎

Whats more you tweak the PWM frequency variable on the fly and download within seconds. Wow. 

So for those C/Cpp Arduino die hards out there, there is place for Micropython for rapid protytyping. 

Maybe when time allows we must create a little thread on this subject. 

Edited February 2, 20232 yr by BritishRacingGreen

Capacitors :  Why ESR Matters

If we look at the  inverter on its components level ,  we find mostly solid state  devices , electromechanical devices like the relays , electromagnetics eg. transformers , film capacitors and electrolytic capacitors.  Of all these the electrolytic has a deterministic lifecycle of degradation , even gracefully , and this has to do with its construction . The electrolyte is embedded in paste between the layers of aluminium foil , and it is fluid .  The capacitance and performance of the device depends on the quality and quantity of the fluid.  unfortunately , in the ambient environment of the high power  inverter , temperatures get high , conditions gets dry , ripple currents are high , working voltage is close to max voltage and with it the electrolyte dries out over time. In fact the manufacturer will state as a spec the number of hours that the capacitor can operate. This is typically 5000 hours , but you do get very high quality at > 10000 hours. This 5000 hours may seem very short , but it only 'counts' when the operating temperature reaches above certain high temperature thresholds  .So that means if the capacitor runs relative cool , it can last many years , although other factors also influence the degradation as previously stated.

So if we receive a machine that has been running for 4 years already , do we need to change the caps as a qualitative measure ?  Not necessarily  , as the caps could have been subjected to good operating condition as described above. So the question arises , can we measure the level of degradation . Fortunately we can , by measuring its Effective Series Resistance  (ESR)   . While this is never a silver bullet , it does provide a good measure . The ESR of a capacitor is a resistance that can be seen as being in series with the capacitance . So the higher this resistance is , the less is the device capable of filtering , and under ac circuits it can influence the reactance. This ESR value is typically in the order of tens / hundred of  milliohms when the capacitor is good , but as the electrolyte dries out , this resistance increases , and values approaches 1 ohm or more will be a telltale that the capacitor has degraded to a level that it needs replacement.

I am pretty new to the  subject of ESR and it is difficult just to measure ESR as an absolute value and  qualify it as good or bad . Fortunately I have some good new capacitors to expermiment with as to provide a basis of comparison.

So how is ESR measured ?  The capacitor within this model consist of capacitance in series with the internal resistance. An ESR meter employs a low voltages ac oscillator to generate a 100hz square wave signal . if the Capacitor Under Test (CUT) is exposed to this ac circuit its own reactance will be very close to zero ohms , given a 100khz frequency and a high value of capacitance. If the reactance of the CUT  is near zero all is left is the ESR , and this we can measure using voltage divider circuit.

Unfortunately I have not yet layed my hands on a commercial meter , and the one that is available in local store is in the order of R6000-00 . So I decided to cook up a homebrewed circuit , based on DIY designs on the web .  Below is a circuit diagram of my crude ESR meter:

 

I am using a Raspberry PI Pico to generate the 100khz square have at 3.3V . This signal drives 6 logic inverters in parallel , each inverter feeds a 270R resistor. So the effective  source resistance of the generator is about 45 ohms.

The output is low pass filtered by C3 and feeds R7 , a 10 ohm resistor.  This provide about 450 - 500 mV at point A , and this depends on the drive capability of the generator , which is not optimum in my case , but good enough.

The CUT is terminated by R8 , 10R to ground. So you can see that the voltage at B will get lower as the ESR of CUT gets higher. Currently I am using an oscilloscope across points across points A and B to get a measure of ESR magnitude.  When A and B is short circuited , I read about 6mV or lower . When I connect a 1R resistor between A and B I read something in the order of 30mV across A and B . This is some crude calibration.. So anything between 7mV and 30mV provides an ESR of 0 to 1 ohm. That's the range i am interested in. 330 milliohm is about 10.6 mV.

Below is an image of the ESR voltage level for a new 470uF 63V electrolytic. You can see that at 7mV the ESR is very low.

 

Here is a 1000uF 16V one removed from inverter . 

 

At nearly 750 milliohm (22mV ) its bad . I say its bad because I have tested good 1000uF 16V devices and I get far better results than this. Now here is the interesting thing about this CUT : the capacitance reading on cap meter is 1023uF !!!!  , which implies its ok , but BEWARE , the ESR shows its bad , and its probably going to fail in not so distant future !!!!

 

The next sample is also a 1000uF 16V , also removed from inverter , this is one is even worst :

 

And next is what the ESR looks like when the capacitor has ran dry : its virtually open circuit !

 

I will continue this post in a next one , cover in-circuit test of ESR .

 

Capacitors :  Why ESR Matters  continued 

There is a promise that we can measure ESR in-circuit , although along with it a whole lot of ifs and buts . The basic theory behind the possibility of performing  in-circuit  tests , is based on the fact that the drive voltage is low , and it cannot turn on solid state devices , or at least most of them. The other reason is that it measures very low impedance and such impedances are typically not dramatically  affected by the circuit surrounding the CUT capacitor under test.  It is of course wise to know the circuit , and be the judge whether your readings will make sense . This is not a problem for me because my ESR measurement effort is driven application specific , and that is the Axpert family of inverters. However  , there are some things to watch out for . If you have two or more caps in parallel , we will be reading the ESR of the parallel.  A good example is the bus capacitors on the dc bus , there are 470uF 500V in parallel. 

So really the first price will be to measure as many caps in circuit when you receive a faulty main board. So I gave this a try and  my results were very positive. Fortunately I have a faulty MKS4 main board that is not brand new , but its close enough as it only had been in service for a couple of months. SO I took this board and measured the SMPS related caps in-circuit. As  expected , they gave me readings of close to short circuit .  I then took the Malawi MR2 board and performed the same tests. The C79 (5V filter cap) showed near open circuit , C78 (12v filter cap) showed a bad reading of 1 ohm , and the others indicated ok , although not as good there MKS4 counterparts. This gave me a lot confidence , as it manifested the same level of degradation I had with the Cape Town machine I repaired (5V cap very bad , 12V cap  bad , others ok.

This is great because no initial de-soldering is required .The reading also prompted me to investigate why the 5V rail caps will deteriorate  so badly in relation to the others . Lets have a look :

Below is a simplified image of the SMPS output  stages along with the filter caps for 12V,5V and -12V :

Notice that C79 and C78 are both identical 1000uF 16V electrolytic . Its always C79 failing , causing the clicking oscillation on the display. C78 is also bad in terms of ESR , but things are somewhat more forgiving on this rail . Let find out why . 

Firstly it appears that the control board (DSP)  is thirsty on the 5V , I cannot measure the load current , this is my assumption. Secondly , C79 is a kingpin filter cap, because  its the only cap on the input of the 5V linear regulator. This is in contrast to the 12V rail , for which the filter cap C78 is affectively in parallel with other capacitors downstream.  Then we must also take in consideration that the MAIN SMPS regulation is performed via this 12V rail , and it can compensate for some sagging to an extent. Lastly , I assume that the 12V rail has less load current  than the 5V rail.

So it would appear that ripple current is the difference between rate of the 5V cap and the 12V cap degrading. There are now just enough field data to prove C79 will fail first. But make no mistake C78 is also on its way . Replacing only C79 results in the entire inverter working again , but be aware : replace them all , and in particular C79 and C78.

The rule of thumb is replace all 5 the filter caps as shown in the image above , and for the good measure replace  C69 and C75 as @Coulomb has shown  his image below :

 

Its of course worth to test other caps as well , but the important ones are the bus capacitors . The two large bus capacitors will read near short circuit . But lets be aware , their ESR will be very low because of the 500V size . I have at least removed one of them , tested both capacitance and ESR out of circuit . I also made sure that the left over capacitance in-circuit capacitance of the remaining capacitor is also in the 470u region. I don't have much confidence yet whether my ESR readings are good because these caps are huge . But I decided that they will remain , as they have a close to zero ESR anyway .They are also very expensive to remove.

As far as the battery bus caps are concerned , I am not to worried , their capacitances are good and overall ESR good . Because they are connected to battery (the perfect storage / filter) i am under the opinion that they will not be subjected to  ungraceful degradation. (Might be proven wrong , but I am confident, I assume ripple currents are low on average) .

 

And of course I am focusing on the main board only , of course there is also MPPT and SPS  boards etc. But I will cover those once my main boards are good

 

 

 

 

 

Edited February 6, 20232 yr by BritishRacingGreen

3 hours ago, BritishRacingGreen said:

As far as the battery bus caps are concerned , I am not to worried , their capacitances are good and overall ESR good . Because they are connected to battery (the perfect storage / filter) i am under the opinion that they will not be subjected to  ungraceful degradation. (Might be proven wrong , but I am confident, I assume ripple currents are low on average) .

Err, these are the first ones I replace, pre-emptively, with the longest life ones I can find (10,000 h 105°C parts if possible). That's because these protect the MOSFET, and they have to absorb the spikes of energy transferred from the MOSFETs via the inductance of the MOSFET battery-side circuit. Yes, the battery is there effectively across those capacitors, but there are long, inductive battery cables in series, limiting the battery's usefulness in the tens of kilohertz region where the action is happening. And if those capacitors stop doing their job, it doesn't just loop the inverter on power-up, the entire array of MOSFETs and probably more are damaged in spectacular fashion. Also, the battery capacitors are overseeing a multi-kilowatt circuit, not just a multi-tens of watts circuit. So they see very heavy ripple current.

3 hours ago, Coulomb said:

Err, these are the first ones I replace, pre-emptively, with the longest life ones I can find (10,000 h 105°C parts if possible). That's because these protect the MOSFET, and they have to absorb the spikes of energy transferred from the MOSFETs via the inductance of the MOSFET battery-side circuit. Yes, the battery is there effectively across those capacitors, but there are long, inductive battery cables in series, limiting the battery's usefulness in the tens of kilohertz region where the action is happening. And if those capacitors stop doing their job, it doesn't just loop the inverter on power-up, the entire array of MOSFETs and probably more are damaged in spectacular fashion. Also, the battery capacitors are overseeing a multi-kilowatt circuit, not just a multi-tens of watts circuit. So they see very heavy ripple current.

Thank you, yes the cable length can have a major influence. Noted.