Online UPS silencing

[taydin]

The 3 KVA online UPS that was giving me good service in the electronics room and that I had previously silenced, wasn't able to last long from the batteries. Lately the local electric company cuts power for many hours on end, and this UPS, with internal batteries, was only good for about 45 minutes with only my Linux PC running and anything else turned off.

So I had bought a new 3 KVA UPS that has an external battery bank with high capacity batteries. This one was able to last many hours, but had the major disadvantage of being very loud. There were two sources of noise:

- Fan noise. Three high speed fans were forcing air into ducts that are around the first inverter and the second inverter. It was like the sound of a hair dryer running continuously.

- Chirping noise coming from the battery charger SMPS. Really annoying ...

[taydin]

I first removed all air ducts and the three high speed fans. Instead, I put two large, low RPM 20cm fans on the heat sinks of the inverters (one fan removed in the picture)



This completely eliminated the fan noise. But when the UPS was running at high load (above 50%), the second inverter heat sink was getting quite hot. If the load got even higher than 50%, the UPS was issuing a high temperature alert and entering bypass mode. But this was fine for now. Most of the time, the UPS is running at about 20% load.

[taydin]

 Lately I must have been running with a higher than usual load, because the UPS was doing bypass more often. So I decided to do something about it.

I first got 1 meter length, large heat sink from a local aluminum extrusion company

[taydin]

I then took lots of pictures of the PCB's and then disassembled them. As can be seen, the heat sink of the second inverter is really tiny, so the design relied completely on high speed, forced air cooling.

[taydin]

I removed this tiny heat sink, and made a much larger one



But unfortunately, when cutting the tab on one side, I cut it on the wrong side. And I didn't have the option to redo it, because the extrusion was not enough for the other heat sinks that I needed to make. So I cut the tab out and welded it to the correct side 

[taydin]

I also replaced the tiny heat sink of the battery charger with a larger one. More about that later.

[taydin]

So this completed the second inverter and the battery charger side

[taydin]

Now, here is the reason I have also increased the heat sink of the battery charger. I basically want to put a permanent load to the output of the charger, which will continuously draw 1 A at all times. I know, this is horribly inefficient. the output voltage is about 77V under 1A load, which means 77W wasted day in, day out. But I did consider other alternatives and they seem to be quite difficult or uncertain of yielding success:

1) Remove SMPS transformer and pot it into epoxy. I'm sure this will reduce the chirping, but my feeling is that it won't eliminate it entirely.

2) Put another SMPS in the place of the existing one that doesn't chirp. This will be hard to do, because the UPS main controller is tightly integrated with this. It monitors the output of the SMPS and it does the throttling of the SMPS when there is no load. There is also a number of other, extra signals coming and going to the UPS main controller. The new circuit would have to simulate the correct behavior to make the main controller happy.

So, I decided to put a resistive load at the output that will draw 1 A at all times. two 270  and two 330  resistors, each 50W, connected in parallel will do this. I'm hoping the heat sink will keep them reasonably cooled so that they stay close to 1 A current draw.

[taydin]

Next, I made the heat sink for the first inverter



This still needs to be drilled and tapped. And there is also a bridge rectifier that needs to be glued on this.

[taydin]

Here is some extra bits of information on how the chirping manifests itself:

The battery charger SMPS uses the UC3843 PWM controller chip. When the SMPS is loaded, it drives the output MOSFET continuously at about 94 KHz. This won't be audible



But when the SMPS is not loaded, the UPS controller card starts throttling the SMPS using the uc3843 "output compensation" pin



And the frequency of those bursts is around 100 Hz, quite audible and annoying!

[taydin]

Once I finish the first inverter heat sink and install it, I will still place a 20 cm fan that sucks air between the two heat sinks, just to get some circulation going. I hope this will be enough. But worst case, I can put an extra 20 cm fan on the other end to increase air flow without noticeable air noise.

One thing that bothers me is, when I was running the UPS with just two 20 cm fans, and with the air ducts removed, the large inductors in the second inverter are getting quite hot. I have measured about 105 degrees celsius for the copper windings and 120 degrees for the ferrite core. Based on my own research, this extra heat will just cause the inverter to run at a lower efficiency. What do people think? Do I need to do  something to cool those inductors?

[babysitter]

Did you locate the source of chirping?

If it is a coil, KONTAKT Chemie PLASTIK 70 or a similar laquer varnishing the windings and core might silence it by preventing movement. (Source: How i learned it during my TV repair stint.)

Ceramic capacitors can exhibit a piezoelectric effect, you might either isolate or exchange against non-hissing types.

[BradC]

Do I need to do  something to cool those inductors?

Probably putting the cooling back the way it was designed would be a good start

What you appear to have discovered is the device was engineered with a specific cooling system and by attacking it in a piecemeal manner you keep finding further issues. You might be in for a long road, but yes I'd cool the inductors.

[taydin]

Probably putting the cooling back the way it was designed would be a good start

Throwing it into the garbage is a much better idea than that 

What you appear to have discovered is the device was engineered with a specific cooling system and by attacking it in a piecemeal manner you keep finding further issues. You might be in for a long road, but yes I'd cool the inductors.

The fact that this is designed for forced air cooling doesn't mean it will be very long road to silence it. I already did silence my other UPS, which is also a 3 KVA model and also was using forced air cooling (it didn't have the chirping issue). It ran for 5 years with 20 cm silent fans. Then it got broken and I fixed it, and after that it has been running another year the same way.

And with this UPS, I don't "keep finding issues". The heat sink part and the chirping part is done, I don't see any issues there. My only concern is the hot inductors. If it's just an efficiently issue, then I am willing to spend the extra money on the electricity just to be able to work in a silent environment. But if there are other issues, I'd love to hear about them.

[taydin]

Drilled and screwed the 4 resistors onto their heat sink and mounted the heat sink to the side of the chassis that normally holds the batteries. I'm using an external battery chassis for this UPS, so this area is available.

[taydin]

Finally installed all three heat sinks to the PCB's. I used thermal tape instead of paste. Just cut one long strip of tape, punched the holes with a wearable belt puncher. I was a little concerned that the heavy heat sinks would bend the PCB's, but it turned out that the attachments I made to hold the fans also prevented the heat sinks from sagging.

Drilled and tapped holes to the top of the heat sinks so that the attachments for the fans can be mounted there.

[taydin]

With the large heat sinks in place, two cable assemblies ended up being too short, so I had to make longer versions of them:

[taydin]

Installed the fan mounting brackets. These brackets added sufficient rigidity to the heat sinks, which prevented them from sagging and bending the PCB's



Finally installed the 20 cm fans. Only have one 20 cm fan at the moment, so I will temporarily use the fan that came out of my old UPS, where it ran for more than 5 years. After all that time, it has a slight bearing noise, but isn't bothering me. Already ordered a new 20 cm fan.

[taydin]

Wired up the resistor pack that will be used to draw 1 A from the charger SMPS

[taydin]

I attached the batteries to the UPS, but didn't turn on the main power switch in the back. Confirmed that the resistor pack is drawing 1 A (was actually 1.2 A). Then I turned on the UPS main switch, which was uneventful.

[taydin]

The only sound I can hear is the bearing noise of the old 20 cm fan. There is no other sound 

[taydin]

I have let it run for a while and monitored the heat sinks. With about 20 % load, the heat sinks are practically cold.



The resistor pack, though, is getting really hot. The resistors are at about 100 ℃ right now. This is expected, because they are basically converting about 80W into heat. Their power rating is adequate, so I will just let them be this way. If they get damaged as a result of extreme heat, I will know it, because the charger circuit will start chirping again  If that happens, I might just install another 20 cm fan over these resistors to make them last longer.

[jbb]

I'm not sure I'd be so bold 

How did you pick the 1A dummy load for the charger?  Clearly if you can get away with less load current there's less heat to get rid of.  Also, I suggest you have a look near (and above!) the resistors to make sure the hot air plume won't age other components.

Now that it's working, how about adding some grille (earthed or nonconductive) so that you can't stick your fingers (or drop a wire) into it by accident?

[taydin]

I'm not sure I'd be so bold 

If you have to listen to a sound that's similar to somebody drying their hair next room, day in, day out, and there is also continuous cricket sound, I'm sure you would be as bold 

How did you pick the 1A dummy load for the charger?  Clearly if you can get away with less load current there's less heat to get rid of.

I attached my electronic load to the charger output and kept increasing the current. The cricket sound level got lower and lower, until it disappeared at around 1 A

Also, I suggest you have a look near (and above!) the resistors to make sure the hot air plume won't age other components.

The resistor heat sink is attached to the chassis of the UPS. After about 1 day of continuous use, the chassis is slightly warm at the edges. The heat isn't radiating to the circuits, which are at the other side of the chassis.

Now that it's working, how about adding some grille (earthed or nonconductive) so that you can't stick your fingers (or drop a wire) into it by accident?

Yes, probably will do that. But I will have to stick my fingers into it for a while to verify that the heat sinks stay cool, that the air flow is good, and that the second inverter inductor heat isn't increasing beyond what it is now.

With 20% overall loading, the large heat sinks are not even warm. They are pretty much at room temperature. The charger circuit MOSFET heat sink is also at room temperature.

[taydin]

There is one improvement that I want to do. As it is now, when power is lost, the power resistors will still be in the circuit, draining the batteries. This will somewhat reduce their capacity. So I will let the resistors be put into the circuit through a contactor, or relay, with a 220V AC coil. The coil will be powered by mains. So if mains power is on, the relay will put the resistors into the circuit, cutting the chirping. If mains power is off, the relay will remove the resistors from the circuit, thereby preserving battery capacity.

I could even make it smarter, where the resistors will only be put into the circuit if the charger is NOT charging the batteries, but it is sufficient the way it is.