Blocking DC out of AC line to protect toroids

[steppenwolf]

Hi!

I noticed that my toroid hums and runs hot, even though it is vastly overrated. Strangely it is not always the case and seems to depend on daytime...At night, for example, it seldom hums or gets hot.
A friend told me that could be due to small dc offset in the AC line of my home when somebody uses a hairdryer or something and that toroids don't take dc well and could run into saturation and that could cause the problem.
Is that true? Could that be the problem?

I looked around the www and found two different schems to block out DC. I don't really understand what the difference is between them, maybe you could tell me, which one to consider.

I found those ones:

http://www.diyparadise.com/dablok.html

and (sorry in German but the schem should be clear)

http://www.saque.de/html/dc_filter_fur_10eur.html

Which one would you prefer and why?

Thanks you a lot!!

Best, Stefan

 

[Svart]

I have other questions for you before we all the trouble to attempt to block "dc" from the line.

what device is the toroid in and is it rated for 50hz? If you built it yourself, where did you get the toroid?

A lot of toroids are built for 60hz but are oversized so that they can run on 50hz(although they will get hot). I am wondering if this is a possibility in your case.

Do you have an oscilloscope? If you do, then you can see if you actually have "DC" on the line. Items like hairdryers probably won't induce a "dc" offset or DC signals, they usually backfeed HF noise but I doubt that is too much of a problem for you.

 

[CJ]

"even though it is vastly overrated"

That might be the problem.

If you open the sec leads of a trans, the voltage will rise until the core saturates.
It will hum and get hot a little later.
Sometimes a long time later.
Like when you are at work.


Running a big trans is semi like running an open circuit trans.

Put more of a full time load on it, like a light bulb or something, and then track the temp.

The only way to make a toroid work with dc is either by careful winding of offset half sections, like WE metal cans, or by cutting a gap in the the core with a wet saw.

 

[Svart]

True.

I still don't buy "DC offset" or anything like that though. It's pretty hard to induce DC offset into mains branch with a small appliance.

 

[steppenwolf]

Hi!

Thank you guys for your answers!
The toroid used is a 50Hz type and rated 50VA. The application:
The G9 preamp. There are two of them back to back in the circuit. The "problem" comes from the first one, the one directly connected to the mains.

@CJ:
I see your point, I had that problem with the second one. I loaded it's secondary with a resistor and solved it that way... They seem to need load on the secondary!

I have the pre on my desk at the moment and sometimes I can hear the hum, sometimes I don't. When I touch the toroid when I can't hear hum from it, it just gets warm, when I hear it hum, it is really hot.

The unit is on, of course. I kept the unit on for a long time to verify and I couldn't hear anything. Now a few minutes ago, it hummed again suddenly and got hot, too...

Really strange. The unit works perfectly, though.

I have a scope and I looked at the sine of mains with a 100:1 probe.
But it's difficult to see dc offset as there might be only a few hundred mV if at all or am I measuring wrong.
Maybe something stupid as a hairdryer could "induce" dc as they mostly use diodes to switch it into half power mode. They just cut out one half of the sine and use the other one for power. So one side should be slightly lower than the other du to loading and what you could get is a slight dc., right?
I have the parts lying around to experiment with those dc blockers just to sort that possibility out.
For which one would you go?

Thanks again!
Stefan

 

[CJ]

50 VA is not overly big.

Only other thing the cooked cells can dredge up is maybe a center bolt shorting, a shorted turn inside the trans from a nick in the wire insulation, or your chassis somehow making illegal contact with the toroid from time to time.

 

[steppenwolf]

Checked it, nothing is touching and the centerbolt is only attached on one side of course!
I'll swap the transformer, maybe it's just a faulty one...
Best,
Stefan

 

[shabtek]

are you using a sensitive digital volt-meter to measure this?

it is not really possible for a dc offset on the mains--if so, it would be on everything plugged into that branch of the electrical grid..and after a rectifier/cap/filter, that would need to be huge in order to drive the low DCR of the transformer windings and provide a ripple-less dc@ 50-60hz.

why use a x10 probe when looking for smaller volts?
cheers
edit: I just looked @ links for dc blockers. does that look really strange to anybody else? It would not block dc but create it--or at least inject switching noise into your transformer

 

[steppenwolf]

Yeah, I'm still trying to wrap my brain around those schems that actually seem to work...
Has anybody an idea how they really work?

 

[Curtis]

In some states of Australia the power utilities used to apply HF carrier frequencies to the mains for signalling things like domestic off-peak hot water cylinders and heaters to switch on and off at various times of the day. Clutching at straws here, but maybe something to look in to if the power utility in your country uses a similar system? Scoping the mains at various times of the day would tell you if there's anything superimposed on the incoming mains, and if it coincides with the period your transformer runs hot..?

 

[Svart]

I've been thinking about those DC blockers and to tell you the truth, I don't understand them either. What I don't understand is the use of the diodes. You'd have switching losses and partial rectification from what I see.

DC blocking is as simple as getting some high voltage caps and putting them in series with the L and N of the input.

i wouldn't suggest you actually do it because there are safety risks involved.

I still don't think there is DC on your line though.

HF hash has a much better chance of being present on the line, switchers with PFC can cause line hash (I hate PFC...) which can cause strangeness such as core saturation if it is bad enough. Hair dryers can definately cause hash too.

 

[CJ]

Doh on me, measure the pri current of both the good and the bad, I bet the hot one drags more current.

Clamp On meters are really wonderful for this.
No dangerous Mains to mess with.


Usually the pri C will filter out the hash?

No wait, the C also moves stuff across the coil to the sharp end of the circuit?

 

[s2udio]

Has the toroid got split primarys (240/110) ?
Are they phased correctly ?

 

[steppenwolf]

Hi!
No, it only has a 230V primary winding...

 

[guavatone]

http://sound.westhost.com/articles/xfmr-dc.htm

 

[Gordy]

Do NOT use the diyparadise.com circuit, as the diodes are wired just plain wrong.


Many (most, all?) large electrolytics do not like a big reverse DC bias. The diodes are there to conduct when the DC offset on the line exceeds the reverse bias DC rating of the capacitor.

The circuit in figure 8 at http://sound.westhost.com/articles/xfmr-dc.htm is quite reasonable. (The diodes are drawn like that because he most likely started with a bridge rectifier at some point).

Personally I would use the back to back capacitors, and across them wire just one pair of diodes in anti / parallel formation. In this way you get to block 0.7V DC before they start to conduct, and the capacitors are never going to see an excess reverse DC bias (and hence are unlikley to explode!).

I suggest 35 Amp diodes to give them a shot at survival before the fuse / breaker blows if there is ever a circuit 'event'.

 

[steppenwolf]

Thank you for that article! It is much more detailled and seems right!
I have some big diodes and some 10.000uFV caps so I will try that before
buying a new transformer.
I'll report back!
Thank you!!

 

[steppenwolf]

Hi!

I build the circuit as mentioned by Gordy and installed it in a small aluminium box. So far, the transformer didn't hum anymore and I used it many hours in continous. I hope I'm not telling that to early, but the problem might be solved! The transformer also doesn't get hot anymore, just barely warm.
Thank you guys for your comments!
Best,
Stefan

 

[steppenwolf]

The problem is definitely gone!

I did some measurements on my mains. I put a 100k resistor in series with a 10uF cap across the mains and measured dc across the cap.
Normally, the dc offset just sits at 0V and fluctuates into + and - by only a few mV.

Then, just for fun, I plugged in my hair dryer into the same line and put in into half power.
The reading was some stunning 1,2V dc...
I let the dryer run downstairs, to see how bad it is and still the least I measured was around 400mV.

The next experiment was to hook my G9 up without the dc filer and monitor the mains with the circuit described above. I did some desk work in the evening and let the G9 run, cause I can hear it hum when it's silent around me.
After an hour or so it started to hum and the meter said -800mV dc on the mains, it was gone a few seconds later and so was the hum...

Maybe, at my home it's extraordinary bad but still it was a fun experiment...

Best,
Stefan