Maynuo M9712 Electronic Load - New unit faulty with bang and smoke - Now repair

[Oli.Hall]

Hi,

I have a Maynuo M9712 which has sadly failed with fireworks within the first few uses.

I purchased it new from Shenzhen Measuring Instruments Store on AliExpress and it was delivered to me in Aug 2022. It worked fine for the first few weeks but then failed earlier this week during normal use accompanied by a loud bang and a cloud of expensive smoke! 

I had been using this unit to test the capacity of 52V e-bike batteries (58V fully charged). These discharge tests were being run at 3A / 180W, well within the Maynuo specs. of 150V / 30A / 300W max. I can confirm that until it failed, this unit was excellent for this purpose and I found it easy to control with HKJ's TestController to datalog and make graphs on the PC.

• The battery on test had been tested three times previously with the same protocol and same test leads. All previous test runs had completed successfully without incident.
• The test leads in use were constructed especially for this purpose and used fully keyed connectors throughout to guard against reverse polarity connection.
• The connections had all remained unchanged from the previous test run, except from unplugging one connector to plug in the battery charger between each run.

I was preparing for another test run. Load mains power on, 150V voltage range selected, 3A current range selected, PC serial mode set, 3A CC discharge set. Load on/off toggle switched off. Upon plugging the battery into the load, a large spark occurred at the connectors accompanied by a loud bang and smoke from the load. I immediately disconnected the battery and powered off the load. After determining that everything was safe and checking all the connections for polarity (they were all correct), I powered on the load did some testing with a regulated power supply.

Autopsy

• The unit appears to still pass its power on self test! :lol
• However, Its input is around 7 Ohm when the load toggle is off.
• With no connection to the load terminals, the display always shows REVERSE and beeps like a reverse polarity connection.
• Applying a low voltage (2.0V) to the input, the load displays input voltage of 1.935V and a current of around 400mA flows into while the load toggle is off.
• Programming the load to 500mA CC and toggling the load on results in the load displaying a current flow of around 499mA, while the measured current flowing from the PSU is around 697mA.

Warranty Claim
On contacting the seller through AliExpress, he initially refused to honor the 12 month warranty. He stated that burned components on the motherboard made the load unrepairable. When I asked for a refund he refused saying burning of components on the motherboard was not covered by the warranty. He claimed this could only be caused by user-error or non-compliant use. After I sent him photos proving that all connections were made correctly and that the use was well within the specifications and being used in a compliant way, he stopped replying to me so I opened a PayPal claim. He has since asked me to send the load back to him in China for a refund. As the seller has already stated that the unit is unrepairable, I am hoping that PayPal might not require me to return it as £100 to send it back to China will eat up 25% of any PayPal refund.

Damage
A peek inside shows at least four of the IRFP264N are burnt out along with cracking visible on their accompanying ballast resistors. What I presume are the associated gate resistors are also burned. I have not done any more than a quick visual inspection, so I am unsure whether further damage has been done to other components.

Cause
I can see many marks on various areas of the PCB that look like water residue. What looks like corrosion is also visible to the leads of several components in the photos below, including discoloration on and around the legs of the burnt MOSFETS, the legs of the ballast resistors, and around the TLE2064 Op-Amp and nearby SMT resistors.

I can say with confidence that this unit has not been subject to water ingress since I have owned it.

What do you think?
Do you think this is water damage?
Or is it water-based flux that has been uncleaned after production and has caused localised corrosion in these areas?

[temperance]

You've got what you payed for. Toys.

An IRFP264N should not be used in linear applications. Check the SOA of this device and you will see it has no DC specification.

[The Doktor]

temperance, the fact that there is no DC SOA specified does not mean the device cannot be used in that application. That information may be available from one of the manufacturer of the device but not included in the datasheet, or the folks at Maynuo may have determined the DC SOA by their own testing. I've got the M9712B myself, and I'm often and run it right to the Limit with absolutely no trouble. Granted it did have a problem with the current sense when I bought it, and had I tested with a battery instead of my bench Supply, it would have gone up in flames. But after the repair, it is fine.

 That circuit board looks like it came from the radio room of the Titanic. it was either contaminated somewhere between the manufacturers facility and you, or it got out the door without being properly finished. I can't be sure that's why it failed, but I would certainly be suspicious. I personally really like my Maynuo load for testing smaller power supplies and such. If it worked well for you, I would not be afraid to buy another one, perhaps from a different distributor.

If you want something a little bit more robust and don't mind not having some of the features like battery capacity measurement, check out some of the units by Transistor Devices Incorporated (TDI). They call their units Dynaload oh, and you can often find pretty big ones at decent price on eBay. May need some fixing up. They are large, heavy, and kind of plain-looking, but they seem to be put together pretty well. Manuals are available online for many of the models, absolutely free. I've got a 50V 400A 3000W model myself, and you'll see other EEVBlog members have various models as well, everyone seems to like them.

[Oli.Hall]

Thank you temperance & Doktor,

All of the teardowns that I can find of similar Maynuo models, including the M9710, the M9711, and the M9812 all show those units are usually fitted with the IRFP250 or IRFP250N.

It would appear that Maynuo has for some reason substituted IRFP264N in my unit for the usually fitted IRFP250N.

Why I don't know. Possibly they are unable to get the IRFP250 due to chip shortages, or cost?

Has anyone else seen the IRFP264N used in their Maynuo load?

[Edit]
temperance - re. your comment "You've got what you payed for. Toys.".
I did hours of research on here when deciding which load to buy and this one was far and away the most highly recommended load on EEVblog.
Nevertheless, I do thank you for helping me to spot that the MOSFETs are different to usual.

[temperance]

Some reading on the subject:
https://www.infineon.com/dgdl/Infineon-ApplicationNote_Linear_Mode_Operation_Safe_Operation_Diagram_MOSFETs-AN-v01_00-EN.pdf?fileId=db3a30433e30e4bf013e3646e9381200

The ZTC point of for the IRFP264N is around Vgs= 6.2V with Id being 100A.

The DC SOA data published by manufacturers is also not valid. An explanation from Little fuse, IXYS, themself.
https://www.ixys.com/Documents/AppNotes/IXAN0068.pdf

MOSFET's for linear applications do exist from Little fuse and Micro linear. Expensive, but they never fail in linear mode.

Contamination: where do you think the contamination came from if all power semiconductors and power resistors went up in smoke? Even the connectors in that area seem molten.

[The Doktor]

temperance , why do you keep focusing your attention on the transistors? I don't believe the model of transistor had anything to do with this failure. Remember what he said, the power to the load was switched on, but the load was switched off. This means the transistors should have been biased 100% to cut off, which is about as firmly in the SOA as you can possibly get, regardless of what transistor is used.. The problem here is that for some reason those transistors were not Switched Off, they were trying to draw far too much current, which is why they exploded.



 As far as the contamination, if you are saying it is all caused by the failure, I believe you are wrong. Certainly much of the dirty appearance of the circuit board is caused by burned-up circuit components settling on the board, but I don't believe that is the cause of the things that look like big water marks. I believe that those are exactly what they look like, deposits leftover when contaminated water evaporated. I kind of suspect this unit may not have been new, it may have been refurbished by somebody and sold as new, but of course that is just a guess. I'm thinking that perhaps those deposits could be conductive when they absorb some humidity, maybe that caused the transistors to be turned on when they shouldn't be?

[temperance]

The battery pack has build in protection which might have saved the load from exploding previously.

Despite the evidence provided you can keep on using MOSFET's not made for linear applications if you believe those to work fine.

[Oli.Hall]

The Doktor and temperance, you both make good points and I think you are both correct.

You are right to say that no current should have been flowing through the MOSFETs at the time it went pop and I would suggest that the exploding MOSFETs were likely to be caused by the corrosion having made a conductive path, causing a small gate voltage to be present at the moment the battery was connected.

This small gate voltage would have put the MOSFETs into their linear region which is where temperence's suggestion about unreliable linear operation in the IRFP264N comes into play.

The IR spec for the cells suggests that the battery has a short circuit current in the region of 300-400A. The BMS is set to cut the output off at around 60A but this would take a few ms. So it is quite probable that somewhere in the region of 300-400A flowed through the MOSFETs for a few ms. Being in the linear region at that time caused them to pop. Maybe the IRFP250 would have taken this ok, then again maybe not.

Ultimately, the cause was the corrosion, as without it, the MOSFETs would not be conducting any current at the time the battery was connected.

So you are both right! 

[Oli.Hall]

I am interested to hear from other Maynuo M9712 and M9812 owners.

What MOSFETs does your unit contain, and what date did you buy it?

IRFP264N same as mine? or IRFP250/IRFP250N like some of the tearwdowns show?

Do you have clean PCBs or does yours also show signs of water / flux residue?

[manicdoc]

Regarding the water damage, either it got there via a direct mechanism (ingress or 'animals') or it condensed into that place, i.e. warm damp air hitting a cold surface.

I've had small snakes, lizards, mice, etc be a problem before, a few mL of pee pee in the wrong place...

[bastl_r]

Hello

Look for "spirito effect"
In the fllowing link is an exlpanation to the effect.
Because of this a BSM112 broke at only 1 - 2A and 40V which should have been good up to 100V and max(peak) 625W.
5V and 50A were previously possible without problems.

regards

[CaptDon]

I had a rack mount Sorenson supply simply sitting on my desk and used for circuit boards requiring from 5 to 15 volts. The supply could go to 40. One day while troubleshooting a board on the desk I heard the Sorenson go full tilt and my board promptly blow up. Turns out the Sorenson control board showed signs of water damage. Guess what, the second shift guy had been taking his break at my desk and sitting his mug of iced tea over on the power supply!!!! Dripping condensation daily!!!!

[Oli.Hall]

Hi,

I have decided to repair this load so I requested a circuit diagram from Maynuo.

Maynuo provided me with the partial schematic of the power stage, attached. The same schematic was posted by another eevblog user a few years ago and relates to the M9711. My M9712 PCB has the code M9711-2011081 witten in the corner and the majoriy of the component values match the schematic. However, in addition to the changed MOSFETs detailed earlier in the thread, my version of M9712 a has different op-amp chip and different gate resistors driving the MOSFETs.

To aid my decision regarding which components to replace them with, I would like to know opinions on whether the gate resistor change relates to the op-amp change? Or whether it relates to the change from IRFP250N to IRFP264N? Does the resistor change relate exclusively to only one of these other changes? Or are they all related?

MOSFET gate resistors x8:              R8, R21, R43, R58, R73, R88, R122, R109
Maynuo schematic:                        100 Ohms
My M9712 unit:                             220 Ohms

Gate drive Op-Amps x2:                U3, U5
Maynuo schematic:                       TL084
My M9712 unit:                            TLE2064CN

Before I noticed these differences, I had asked Maynuo about whether there were other differences between loads fitted IRFP250N and those fitted with IRFP264N. They stated that there were no other component differences, or firmware differences apart from the different MOSFETS. They said IRFP250N was only substituted with IRFP264N due to component availability.

I have gone back to Maynuo to ask for clarification if the gate resistor change was made prior to the IRFP250N to IRFP264N change for any other reasons relating to performance or reliability, or whether the gate resistor 220R change is linked exclusively to the IRFP264N change.

I've got the M9712B myself

In the mean-time, I am keen to hear from other M9711/M9712 what MOSFETs you have? Whether you have TL074/TL084/TLE2064CN op-amps? Whether you have 220R or 100R gate resistors?
(Resistor locations are shown in the attached photo, the op-amps are the two 14-pin DIP packages in the centre).

At this point I have two repair options available:

1. Restore it to 'as-received' spec. i.e. Replace the 4x IRFP264N, the 4x 220R gate resistors, the 1x TLE2064CN op-amp like for like.
2. Replace all 8x MOSFETS back to IRFP250N, replace the op-amp with another TLE2064CN, and possibly change the 8x gate resistors back to 100R depending on whether this value is linked to the op-amp or the MOSFET change.
(there are other components which also need replacing including Q7 S8050, Q8 S8050)

Given what has been said before about the IRFP264N and uncertainty around SoA for linear operation. I would like to get some opinions on whether people think changing back to the originally specced IRFP250N would result in a more reliable unit?

Kind regards,
Oli.

[Oli.Hall]

I think I've been able to answer my questions above...

Looking at the photos other users have posted of their teardowns of the M9711/M9712/M9812 units in the past, such as:

So here are some quick photos...

So, this is what MY M9712 looks like inside.

It is clear to see that the gate resistors in the two units linked above are both 220R (Red, Red, Black, Black, Brown) the same as in my unit. As the two units above are both fitted with IRFP250N MOSFETs and TL084 op-amps, this means that the 220R gate resistor change stands alone and is not linked either to the MOSFET change, or to the op-amp change (changes I presume were made in later units due to component shortages).

As the IRFP250N appears to be the original spec for the majority of this unit's production run, I am leaning towards removing all of the IRFP264N in my unit and replacing them with IRFP250N and hoping it will be more reliable. Although, I still don't know whether the initial failure was caused by contamination of the main board, or the IRFP264 MOSFETs.

Kind regards,
Oli.