SU3000NET rattles on battery

[Anonymous user]

I have an older SU3000NET - well really old, 2002 model. Very good condition - spent its whole live in the AC room.

When on battery or self testing, the transformers rattle like there is a loose bolt shaking around. Of course all of the mounting screws are tight. The unit seems to work fine, but the fan runs all of the time. (maybe needs new batts).

The rattle is very loud. Has the transformer shaken itself loose inside?

[BillP]

It might be that the unit has electronics failure. The transformer should emit a somewhat gentle humming "50/60hz" noise when operating on battery. If the sound coming from it is harsh, then probably some of the switching mosfets on the heatsinks gone bad and the output waveform is getting distorted. The model SU2200NET, which is pretty much identical to SU3000NET (apart from the VA difference) is known to have problems with certain resistors R38 and R43. Eventually those resistors fail and that leads to the failure of the inverter circuit.

Note that this is just a speculation, without actually hearing and seeing the unit.

[Anonymous user]

Wow - you may have won the prize!! R43 and R38 are both bad, or out of spec. R43 is at 5K ohms and R38 is at 0.6K ohms. Both look "crusty". Diagram said both are supposed to be 1K 5% 1/4W.

Time to pry out the board, load some new 22uF caps and 1K resistors.

I'll take a Youtube vid before I dive in and post it here.  https://www.youtube.com/watch?v=BVCG0T2cvWQ

[Brad_C]

On 1/5/2017 9:25 AM, karl said:

Wow - you may have won the prize!! R43 and R38 are both bad, or out of spec. R43 is at 5K ohms and R38 is at 0.6K ohms. Both look "crusty". Diagram said both are supposed to be 1K 5% 1/4W.

Replace them with 1 or 2 watt resistors and stand them off the board to allow better cooling. Don't rely on the schematic. Depending on manufacture variations those resistors can be 470 ohm or 1K. Either way, replace them with 1K. While you are there, check the ESR on the high side bootstrap caps (C34 & 35). They are not critical, but every unit I've serviced has wanted new caps. Might as well replace them while you're in there.

I've never seen an individual mosfet fail in these. Any failure I've had cause to fix has been one failing short and causing a cascading failure taking out both high side banks and the associated driver transistors. Much magic smoke was released.

[Anonymous user]

Here's a pic of R43. Crispy and faded.

[Brad_C]

Looks familiar. The other two marked resistors are the 5V divider for the charger. On this unit they were set to charge at 54.1V from the factory.

Those cooked resistors were new when I first repaired it in 2009. I didn't know about the excess heat back then so I just replaced them with 1/4W. I just gave it a birthday so it now has 1W ceramics spaced off the board.

[Anonymous user]

I knew R38 sounded familiar. I read an article about voltage mods and blown F3 fuses about a year ago, and he worked with APC on why R38 cooks!!

http://techno-fandom.org/~hobbit/power/dumbups/res.html

Now I just hope that this refurb will settle the fan down in this unit. As of now, it always runs.

[BillP]

Let us know your repair results. That sound in the video is definitely not normal, lets hope that those resistors didn't cause widespread problems in the inverter circuit by now. 

[Anonymous user]

Which 1K ceramics did you use? Did the legs fit through the board ok? I need to purchase some ceramics, so looking for a source online.....

[Brad_C]

Faulty memory. They were 2W not 1W. I spent a _lot_ of time searching until I found resistors with a >0.8mm lead diameter, so they do fit in the board. They were a bit squeezy though as the charring around the holes on my board had deformed the holes a bit, but they did fit.

Vishay PR02 Series Axial Metal Film Fixed Resistor 1kΩ ±5% 2W ±250ppm/K

[Anonymous user]

Thanks - found the same Vishay 0.78mm legs PR02 on Mouser.

[Brad_C]

While you are in there, have a look at some caps.

• Previously mentioned bootstrap caps on high side drivers
• 22uF caps on -8V charge pump converter
• 330uF caps on -24V converter

I routinely replace these when giving these models a birthday. The bootstrap caps get worked hard and every one I've replaced has been at least an order of magnitude out of spec. They still work, but when they eventually fail they'll take out the inverter. Same with ESR on the 22uF and 330uF caps on the respective converters drifting significantly high (though less so and with fewer consequences than the bootstrap caps). If you are replacing bits, might as well do those too.

I always replace the MOVs also, but that's just belts and braces. Again, if I've got the board out and the desoldering gun hot and extra $5 of parts is no biggie, and I keep a tray of them around. They do progressively wear out as they clip spikes.

These are solid units, easy to work on and can be kept up to spec relatively easily.

I'm baffled by the sound yours is making, but asymetrical operation of the inverter could be causing all sorts of transformer saturation and magnetization. All bets are off when magnetics start saturating.

If you had a CRO, checking the current waveform on the transformer would be interesting.

[Anonymous user]

Thanks Brad - I need to admit that I'm not an engineer, but I play one on TV....  (my soldering skills are getting pretty good though!! Bought a good quality iron)

I'll check the 330 caps as well. What brand do you recommend? Some seem to like Panasonic or Nichicon....

Don't have a scope, but that noise is familiar. I've heard it before in another 2200 unit I was testing.

This unit is worth saving, so i am puling the board to have a good look at it. The unit is extremely clean for a 2002 model.

And which MOVs do you replace? Looks like there are a bunch of them.

[Brad_C]

On 3/5/2017 5:52 AM, karl said:

I'll check the 330 caps as well. What brand do you recommend?

I don't generally go by brand. I pick the best fit from a name brand, but checking my last order the caps are Panasonic and Nichicon (probably a coincidence).

On 3/5/2017 5:52 AM, karl said:

And which MOVs do you replace? Looks like there are a bunch of them.

All of them. The 230V units have 5 or 6. They are all the same, so I just do a job lot replacement.

On 3/5/2017 5:52 AM, karl said:

Don't have a scope, but that noise is familiar. I've heard it before in another 2200 unit I was testing.

If it persists after you've cleaned up the inverter section, use a bit of plastic hose as a stethoscope to narrow it down.

[Anonymous user]

Ok - got a scope. Kind of. Cheap USB, but it will display a waveform, sort of. Not much fine control. The waveform on battery is pretty ugly. I guess that is the banging noise I hear.

Utility power - pretty good.

UPS test - battery power. Scary

[Anonymous user]

Ok- board out of the case and more ugliness under close inspection.

Oily black gunk on top of R26 resistor near the MOSFETs, but it measures ok.

Dry black marks near the MOSFETs at R229. (looks like GSR)

I wonder of the MOSFETs blew? Tried doing a diode test, but must not be doing it correctly.......

And a tiny bit of oil on Q32 near R42 (labels that is)

[BillP]

Replace the burned resistors and preferably all of the smaller sized electrolytic capacitors (eg. all of the caps except the big battery buffer capacitors next to the heatsinks). Then put the whole thing back together and try it to see if the problem is gone. If the output waveform stays distorted, then likely one or more MOSFETs (or smaller, driver transistors usually marked 2n2222 and 2n2907 near the inverter circuit) are bad. Smaller transistors of interest: Q27-Q30, Q32-Q33, Q35-Q36

Hint for replacing capacitors: always use capacitors with better voltage specs than what is originally in there. Eg. if you see 22uf 16v caps, replace them with at least 35v rated variants (I prefer 63v low esr for those 22uf caps though when replacing). The uf capacitance value should be the same.

If the inverter is already shorted out, then it is very likely that the 20 ohm resistors are burned or out of tolerance as well (R14-R33, R221-R232). When this happens, most of the time the damage on the MOSFETs are severe and clearly visible, eg. blown, burned, split in half, etc...

[Anonymous user]

Some more juicy pics. From what I can see, looks like the MOSFET blew its top (or side), but getting to them is a PITA!! It's the middle heat sink. Only way I can figure to desolder the resistor and FET is to remove the first and second heat sinks first. No way to get to the screws that hold the FET to the heat sink.

[BillP]

Yeah, sadly it seems the inverter is shorted out. As a side note, I was never able to successfully repair a unit that already reached this stage, yet. It is as you said, too much PITA to replace the mosfets, and you cannot possibly know how much more widespread the problem is (nevertheless, I always try to repair because it is a sort of a passion for me, even when countless workhours are lost and I end up failing to repair it). It is also likely that when you try to desolder mosfets and the resistors, the PCB traces will pry off the board due to being severely burned. If you still going to try to repair, replace all of the mosfets, not only the ones which are dead. I think IRF3710 is a good replacement for 48V units, though I am not sure (I know that IRFZ48N works fine for 24V units, but 48V ones need higher voltage rating, so I'd choose IRF3710.). The resistors are rated for 0.125W, but if there is enough room for them, you can replace them with 0.25W or maybe even 0.5W ones too.

The screws holding the fets to the heatsinks are easily deformed when you try to remove them, so be sure to use the appropriate screwdriver to get them off, otherwise you might end up needing a drill.

[Anonymous user]

Ha!! So close, yet so far. At least this unit does work, to a point. I'll tear it apart and see how far I get. I was practicing on another dead unit this morning, desoldering the heat sinks. Not too bad really.

[Brad_C]

I bought an SU2200 that had that sort of damage. I ended up replacing both sets of high side FETs, all gate resistors and most of the driver transistors. 7 years later it's still trucking.

The screws holding the FETs to the heatsinks are allen key and back right out with no issues if you use the *right* sized keys. I ended up taking all the heatsinks off. Because the fets are in rows, once you knock out a heatsink the screws on the fets on the next heatsink are easily accessible *through* the screw holes on the fets in front if you have long enough allen keys (or you can just bend them out the way). You can also take the screws out without removing the heatsinks with long allen keys. Start with the first and then just do them one by one as you work from front to back (ie take out the first one, then use the hole to get to the one behind and so on). That's a bit more of a challenge, but if you only need to get to a couple of fets it's easier than removing the heatsinks. Given yours "works" you probably only need to replace a couple of damaged fets and resistors.

I'm surprised it is running at all to be honest. Mine failed shorted and blew all sorts of stuff up as collateral damage.

I don't now remember if those screws are metric or imperial allen keys. One will be perfect and the other will be marginally loose. You really want the right key.

Do *not* cross thread them when you put them in. Because the fet to heatsink connection is used for high current, make sure both the fet and heatsink are meticulously clean when you put it back together and the screw is as tight as you can get it without damaging it. It appears the aluminium thread in the heatsink is originally cut by the screw and it's very easy to cross-thread and cut a new thread totally destroying the hole, so be super careful not to cross thread them going back in.

[Anonymous user · Accepted solution]

Ah!! Great advice on pulling the FETs and heatsinks. I have the correct metric hex wrench, a long one. I just need enough room to get an iron in there to solder the FETs and resistors on the 3rd bank. I have some parts coming in today, so the fun will begin! I got the unit for fee, without batts, so no loss if it doesn't live. But I'm hopeful.