SOUND ON SOUND

[wireman]

I have a Creative Soundworks DTT3500 surround system that has been used as the main speakers for my last couple of PCs (I use monitors via an interface for serious audio use).
These have developed a hum which I have been investigating as I have a few spare days at the moment.

I think the issue is the 15V 4A power supply that provides the DC input to the system.
The output is not grounded but I think there must be a serious problem with the supply causing the hum. The following waveforms come from an oscilloscope connected to the output with a 20Ω and then a 10Ω resistor load.




Note that the vertical scale is just wrong, the blue arrow is the zero point, the Rigol Ultrascope software is often inconsistent with the scope on the vertical scale. I think the horizonal is correct althought I don't understand the 100Hz waveform. The average voltage is around 12V.

On opening the box this looks like a bridge rectifier circuit, although with capacitors in parallel with the diodes which I was not expecting and have never heard of.
I'm strugging to work out what is wrong from that waveform, I could understand a faulty diode giving you nothing on one half cycle of the mains but the waveform falls before it gets to the point where it looks like we have capacitor decay.

Further images below of the supply internals:







Should I get on the right track I can buy the components and attempt a fix. I was hoping the waveform would reveal the issue and I would not have to dismantle parts of the circuit to find out what was wrong.
From what I can see we have:

Diodes: 6A2BL (this exact model does not seem to exist)
Parallel caps: 104nF (assuming from the 104 written on them)
Smoothing cap: 6800μF 25V 105°C
Fuse and a resistor (1kΩ?)

I tried to join an electronics forum to post this but unfortunately I don't get the verification email, but if anyone can suggest an alternative forum for this question that would be helpful if needed. I'm hoping someone here can help.

[Hugh Robjohns]

I think the little ceramic caps across the bridge diodes are 'snubbers', intended to kill high-frequency ringing from the switching transients.

As to the waveforms... I'm not quite sure what voltage of ripple you're looking at in comparison to the nominal output voltage... but it looks a lot.

It could be that one or more of the electrolytic smoothing capacitors have died. Or it could be that your measuring/loading arrangement is drawing more current that the supply was intended to provide...

[wireman]

The 'ripple' is of order 1/3 of the peak. I used 10 and 20 ohms thinking that was fair for a 4A-rated supply, it is quite beefy. If the electrolytic had gone would the output not just track the input but be rectified?

[jjlonbass]

This is a simple unregulated linear power supply. You see 100Hz ripple as a full wave rectifier is used - both the negative and positive cycles of the stepped-down 50Hz AC supply are rectified i.e. in effect made positive so you see a twice the number of peaks.
I would say that the 6800 uF smoothing capacitor is most likely duff. Even if your 'scope trace was to scale, around 1.5V of peak-to-peak ripple is excessive.
If we assume that your 10 and 20 ohm resistors are in series, then at 15V you would have 15 / (10 + 20) amps of current flowing = 0.5A.
The peak-to-peak ripple voltage you should expect with this setup is roughly 0.5 amps / (2 * 50 Hertz * 0.0068 Farads) = 0.735V.

John

[merlyn]

It might help to make sense of the oscilloscope. What happens if you connect the scope to a 1.5V battery? To see what's happening on the time axis you could connect a guitar -- open A is 110Hz. If you've got a multimeter you could see what readings you get with that.

[wireman]

Thanks for the thoughts so far.

I started with multimeters and a DC intelligent load to see how the voltage might drop with load, the problem is that these just average the voltage.

I will have another look and put something useful on CH1 to compare with. The scope itself is fine and I have never seen it inconsistent with a known or expected input, it is just the software that is messing this up but I thought taking a photo of the screen might be a pain.

I realise that testing with the actual amplifier part connected makes sense but I don't want to dismantle that.

[jjlonbass]

wireman - Are your 10 ohm and 20 ohm resistors in series i.e. you're applying a 30 ohm load to the power supply?

John

[The Red Bladder]

Them diodes look far too small for the trafo. Replace them with something meatier. Pishy diodes are 90% what happens to them things. It ought to have been designed with a cooled bridge and not those fiddly little things!

You can test them by lifting one leg and seeing which one has 0-Ohms both ways. It looks like the one at the back.

[wireman]

wireman - Are your 10 ohm and 20 ohm resistors in series i.e. you're applying a 30 ohm load to the power supply?

John

No, they are made up from multiple resistors in a metal box that I made up a while ago.
10 ohm at 500W
20 ohm at 100W

Not that you put in anything like that power for long as the box heats up.
The box has 4mm sockets.

[Martin Walker]

I would say that the 6800 uF smoothing capacitor is most likely duff. Even if your 'scope trace was to scale, around 1.5V of peak-to-peak ripple is excessive.

That was my first thought too.

Oh, and welcome to the SOS Forums jjlonbass! 8-)

Always good to have a new member who jumps in with practical advice from day one :thumbup:


Martin

[ef37a]

A VERY rough calculation for pk-pk ripple for a 4A load and 6800mfds would be 4V.
For your 10 Ohm test load it would be around 1.5V pk-pk.

That means the amp it feeds must have some excellent regulation in it or extremely sophisticated Power Supply Rejection.

Then, 6800muffs is actually a bit mean for a 4 amp supply? The rule of thumb is 2000 muffs per amp of current. Good kit these days would use 10,000 mfd (easier 2 x 4700)

Yes the ceramic caps snuff RF radiation when the diodes switch off.

Last thoughts, a 15V 4A SMPS line lump would not be that expensive, will have a quick varder.

Dave.

[DGL.]

Another thing to think about is the supply must be regulated somewhere else or accept range of input voltages as that supply is clearly un-regulated so the output voltage will change with the input (mains) voltage, given this a higher voltage nearer 20V might be acceptable and as such a laptop PSU of at least 60 watts (V x I =P, 15 x 4 = 60) might work, would have to more than likely use the cable from the existing PSU attached to teh one from the laptop PSU as the connector type will invariably be different.
Then again the PSU form my vintage Toshiba T8200 is 15V 4A so perfect (model no. PA3048U-1ACA) or the similar vintage IBM supply from an A21m is close at 16V 4.5A so would probably also be a contender (Part No. 02K6654/02K6666).

Looking at Toshiba PSU's on ebay (probably better/safer 5than a random ebay special PSU) there is this used for under a tenner buy it now https://www.ebay.co.uk/itm/Genuine-TOSH ... SwwZBfixXJ

[wireman]

Some more information.

Using a Brennenstuhl power meter, the kind with a 13A socket that you plug into (no idea how accurate it is) I find that the power brick draws 0.12A from 245V 50Hz on standby. At normal volume this rises to 0.15A and only 0.16A at the highest volume I would normally use. So this is just over half of the rated power of the supply.

Using a (good) multimeter the measured DC voltage output was:

18.94 V with no load.
13.89 with 20Ω load (power supply drawing 0.15A from mains)
12.3V with 10Ω load (power supply drawing 0.17A from mains)

So my loads are reasonable for testing.

Note that with no load the ripple is small in comparison to previous results shared.

Below are photos of the oscilloscope for loads of 20Ω and 10Ω respectively.




Ch1 is mains input taken from the same power strip supplying the power supply (yes I know how to do this safely.)
Ch2 is the power supply output.

I have thought about purchasing what must be a much smaller SMPS but at the moment I don't know what the DC connector is, it is not the same as I have on other devices. If I had one I could try one of my variable linear or SMPS to see if the hum goes away.

[wireman]

This is a simple unregulated linear power supply. You see 100Hz ripple as a full wave rectifier is used - both the negative and positive cycles of the stepped-down 50Hz AC supply are rectified i.e. in effect made positive so you see a twice the number of peaks.
I would say that the 6800 uF smoothing capacitor is most likely duff. Even if your 'scope trace was to scale, around 1.5V of peak-to-peak ripple is excessive.

Thanks, yes, I get the 100Hz now, I'm just amazed the 5.1 system still works given this input.

I have some laptop power bricks at home from old laptops, so will first see if any of them are an option, they are mostly Dell with weird DC plugs on them though.

And just to clarify I have separate loads, I didn't mean in combination.

[Hugh Robjohns]

There's a good explanation of simple linear unregulated power supplies here, complete with formula to calculate the ripple:

http://www.zen22142.zen.co.uk/Design/dcpsu.htm

As you'll see from the expected waveforms in that article, your supply appears to have very little smoothing going on. So it may well be that the big electrolytic has dried out and/or gone open circuit, or there's a dry solder joint in its connections.

It shouldn't be too difficult to replace that C with a good quality new one, and you could even increase the value if you want. Just make sure you get the right voltage rating and fit it the right way around.