Hi,
if the brown wire is Live then it looks like you have fused the Neutral!!!!
Fuse only the Live.
It's OK to put the DC block in either the Live OR the Neutral.
if the brown wire is Live then it looks like you have fused the Neutral!!!!
Fuse only the Live.
It's OK to put the DC block in either the Live OR the Neutral.
I did a test run ( in a tupperware box!) yesterday - i have some nichicon 10.000 uf/80V handy so i used these with a big bridge.
This is in front of a felix mains conditioner that feeds a zhaolou dac with a zapfilter thus with something like 2x 30 VA attached.
Good news- nothing blew. The R core transformer is now dead quiet, but the toroidal for the zapfilter still buzzes - say 50% less than before, and without that high tone pitch that got on my nerves.
Soundwise, i am very pleased.My tannoy actives do not sound any more bloated in the bass, and finally seem to soundstage which was my big disappointment against my older pair of Genelecs. Treble's quite a bit improved too.
Anyway, will bigger caps be better at killing the last buzzing coming from the transfo? I still have several caps lying around or could buy some big 33.000 uf ones ( this is to drive the above 60/100Va max mentioned above)
last thing for my own safety sake: once disconnected, is there any potential left across the caps ( when i take them to their final box )? need to discharge or will it self discharge?
This is in front of a felix mains conditioner that feeds a zhaolou dac with a zapfilter thus with something like 2x 30 VA attached.
Good news- nothing blew. The R core transformer is now dead quiet, but the toroidal for the zapfilter still buzzes - say 50% less than before, and without that high tone pitch that got on my nerves.
Soundwise, i am very pleased.My tannoy actives do not sound any more bloated in the bass, and finally seem to soundstage which was my big disappointment against my older pair of Genelecs. Treble's quite a bit improved too.
Anyway, will bigger caps be better at killing the last buzzing coming from the transfo? I still have several caps lying around or could buy some big 33.000 uf ones ( this is to drive the above 60/100Va max mentioned above)
last thing for my own safety sake: once disconnected, is there any potential left across the caps ( when i take them to their final box )? need to discharge or will it self discharge?
the caps can discharge through the transformer primary and/or the bypass diodes, if the mains is broken when they still hold a charge.C37 said:
You do not need to use high voltage caps if the bypass is properly proportioned for the load you intend to connect.
I have seen recommendations of between 6.3V and 16V.
I have used 10V without any problem.
Keep in mind that the diodes have to pass fault current (>500Apk) until the fuse ruptures, if the mains side of the equipment ever shorts to Safety Earth.
Probably a wise idea to check your home distribution board. Just checked my board and it has no fuses, and instead uses RCCBs which trip at 30 mA. (i.e. When there is a 30 mA difference between the current flowing in on the Live and the current leaving on the Neutral, it trips open circuit). If your home has a fuse on the circuit that you use to supply your home constructed equipment it may be worthwhile changing the fuse system for a circuit breaker system. (Use an experienced professional Electrician).
For a good writeup and a concluding circuit, check out Rod Elliot's article on blocking dc from mains transformers on his website:
DC Offset and Transformers - Many people have found that toroidal transformers growl at times. This explains why, & how to fix it
DC Offset and Transformers - Many people have found that toroidal transformers growl at times. This explains why, & how to fix it
That's great news that it seems to work O.K. It's something I would seriously consider adding to any "new build" I do.
Now a thought or question. Although the DC volt drop is small over the caps would adding a high value (say 470 k) resistor across each cap help equalise any imbalance that may cause a small reverse bias to develop. As we know, the tolerance of electroylitics is poor (+100% -50% in some worst cases).
Trying to imagine it now, a constant DC voltage across two series caps of unequal value, yet you are pulling a "constant" AC current draw through them. Bit early in the morning for that 😉 .
It was something that was certainly done in the the old days for connecting caps in series to make one of higher working voltage, but that was something completely different than this. Thoughts anyone ?
Now a thought or question. Although the DC volt drop is small over the caps would adding a high value (say 470 k) resistor across each cap help equalise any imbalance that may cause a small reverse bias to develop. As we know, the tolerance of electroylitics is poor (+100% -50% in some worst cases).
Trying to imagine it now, a constant DC voltage across two series caps of unequal value, yet you are pulling a "constant" AC current draw through them. Bit early in the morning for that 😉 .
It was something that was certainly done in the the old days for connecting caps in series to make one of higher working voltage, but that was something completely different than this. Thoughts anyone ?
hi,
we are not blocking pure DC with these caps.
We are reducing or removing the unbalance between the upper half waveform and the lower half waveform that equates to a net current flow in one direction that the transformer sees as a DC current superimposed on an AC current.
The DC current brings the core flux up nearer saturation and that results in increased AC current trying to generate the AC flux changes that we need to give us our transformed AC output in the secondary.
Is this close to the real situation?
Comments welcome.
we are not blocking pure DC with these caps.
We are reducing or removing the unbalance between the upper half waveform and the lower half waveform that equates to a net current flow in one direction that the transformer sees as a DC current superimposed on an AC current.
The DC current brings the core flux up nearer saturation and that results in increased AC current trying to generate the AC flux changes that we need to give us our transformed AC output in the secondary.
Is this close to the real situation?
Comments welcome.
MCB still follow the I^2 T rule and are designed to be capable of passing many kA of fault current and yet still break that fault current without destroying the MCB. Look at the labeling on the MCB it will state something like 5kA breaking current for our domestic breakers.Gordy said:
The Merlin Gerin that I use state 10000 on the MCBs and on the RCBOs
Detection of leakage current is completely different from breaking fault current.
I could be facetious here and ask what impure DC would be like...
It is much simpler and entirely accurate to state that the intent is definitely to block DC. You are correct in yr next statement - the dc current causes the transformer core to saturate (or at least approach the region of saturation). This causes the permeability of the core material to decrease, which in turn reduces the inductance (the number of primary turns remains fixed, of course). When the inductance decreases, the impedance of the transformer primary decreases. The reduced primary impedance causes increased primary current and increased magnetostriction. It is the magnetostriction that causes the laminations that make up the transformer core to vibrate and so make that annoying hum.
Removing the primary dc current is one solution to quietening the mechanical vibration coming from the transformer. Another is to vacuum impregnate the core with a suitable low viscosity varnich/potting compound. Even better is to choose a power transformer that has a low flux density - this typically means taking a conservative view of the transformer VA rating when selecting it. Unless you are dealing with a very honest and capable transformer winder/manufacturer!
Hope this helps.
AndrewT said:
🙂 Is it close to the real situation. Probably, but the question remains. Do the unequal value caps ( unequal value due to tolerance ) see a different DC voltage across each, that could reverse bias one or other cap.
Come on you lot with your simulators. Answers please !! 😉
I think the back to back caps are assumed to be resilient to reverse DC.
For the sake of argument, the reverse biased cap has failed, momentarily, and the forward biased cap blocks the DC and passes the AC. 5mS later the reverse biased cap is being reformed and then blocks the DC and the other cap has become reverse biased.
The caps see the AC voltage across them, if they have not failed, and this is due to the cap impedance at the supply frequency times the AC current. We have seen that manufacturers claim their electrolytics can stand between 700mV and 1500mV of reverse bias. If this is the case then the caps should never fail if the bypass diodes are doing their job.
Similarly the caps should never have more than ~1400mV across the pair for the dual series diode bypass. At this low voltage, resistor balancing of the AC is not required. It will simply be a prorata voltage dependent on the actual capacitances installed.
For the sake of argument, the reverse biased cap has failed, momentarily, and the forward biased cap blocks the DC and passes the AC. 5mS later the reverse biased cap is being reformed and then blocks the DC and the other cap has become reverse biased.
The caps see the AC voltage across them, if they have not failed, and this is due to the cap impedance at the supply frequency times the AC current. We have seen that manufacturers claim their electrolytics can stand between 700mV and 1500mV of reverse bias. If this is the case then the caps should never fail if the bypass diodes are doing their job.
Similarly the caps should never have more than ~1400mV across the pair for the dual series diode bypass. At this low voltage, resistor balancing of the AC is not required. It will simply be a prorata voltage dependent on the actual capacitances installed.
Hello Andrew,
I wouldn't like to see reverse biased caps at all, even a couple of hundred millivolts. Relying on the reforming process to correct a short !! 😀 😀 That's not like you !
I guess there is only one thing for it 🙂 Now where is that bridge and my sweating peg (Soldering iron) to you lot 🙂
I wouldn't like to see reverse biased caps at all, even a couple of hundred millivolts. Relying on the reforming process to correct a short !! 😀 😀 That's not like you !
I guess there is only one thing for it 🙂 Now where is that bridge and my sweating peg (Soldering iron) to you lot 🙂
back to back caps passing AC will always have one of the caps reverse biased when the voltage is above or below 0Vpk. It cannot be avoided.Mooly said:
Line level voltages are generally <=2Vac. This is about 3Vpk.
We pass AC through back to back caps like this in many designs.
Some have said that the common should have a DC bias applied for best audio performance.
We even pass that 3Vpk signal through a single polarised cap in some designs.
I presume that since the transient peak only hits the 3Vpk level on very rare occasions, the caps are assumed to tolerate this reverse bias.
Our DC block should never hit the 3Vpk level. So what's worse?
O.K. This was very interesting. A little practical test.
I built a bridge from 4 separate IN5408 type diodes and wired it as in the above project. Connected in series with the 550 va toroid (actually it is a complete amp in it's own right D Selfs blameless class B) and swithed on.
I took some measurements but didn't use a DVM, I used an AVO 8 instead --- why, just a hunch it might show something ? DVM's are too twitchy on readings like this, the good old swinging brick shows trends as well.
Results. No caps at all fitted. AC across the bridge = 1.55 volts. DC across the bridge = 0 ( Meter twitchy around zero )
Fitted two 2200/35volt caps in series with negatives connected together. AC volts now 0.4 volts. The DC volts was interesting, I used the AVO on the 50 microamp range which is 100 mv FSD. The average was zero, but there were very sharp "kicks" in the needle in pos and neg directions. The needle only made it too perhaps 4 or 5 mv before it was "kicked" in the opposite direction. I would estimate an error of a few hundred percent due to the inertia of the meter movement. However the DC volts on both caps were equal as near as could be determined and hovered around 10 mv for each (The polarity was correct for both).
Next I added another 2200 Mfd across just one cap to give an imbalance. The AC voltage across the bridge was now 0.3 volts and the DC across the bridge seemed unchanged. The larger cap now seemed to have around 100 mv DC and the other 2200 mfd around 75 mv still with correct polarity.
Exact readings were very difficult as the mains constantly changes. The transformer was pretty silent however even with the caps removed altogether.
So it looks worthwhile as a mod and no worries over reverse biasing of the caps. It's good 🙂
I built a bridge from 4 separate IN5408 type diodes and wired it as in the above project. Connected in series with the 550 va toroid (actually it is a complete amp in it's own right D Selfs blameless class B) and swithed on.
I took some measurements but didn't use a DVM, I used an AVO 8 instead --- why, just a hunch it might show something ? DVM's are too twitchy on readings like this, the good old swinging brick shows trends as well.
Results. No caps at all fitted. AC across the bridge = 1.55 volts. DC across the bridge = 0 ( Meter twitchy around zero )
Fitted two 2200/35volt caps in series with negatives connected together. AC volts now 0.4 volts. The DC volts was interesting, I used the AVO on the 50 microamp range which is 100 mv FSD. The average was zero, but there were very sharp "kicks" in the needle in pos and neg directions. The needle only made it too perhaps 4 or 5 mv before it was "kicked" in the opposite direction. I would estimate an error of a few hundred percent due to the inertia of the meter movement. However the DC volts on both caps were equal as near as could be determined and hovered around 10 mv for each (The polarity was correct for both).
Next I added another 2200 Mfd across just one cap to give an imbalance. The AC voltage across the bridge was now 0.3 volts and the DC across the bridge seemed unchanged. The larger cap now seemed to have around 100 mv DC and the other 2200 mfd around 75 mv still with correct polarity.
Exact readings were very difficult as the mains constantly changes. The transformer was pretty silent however even with the caps removed altogether.
So it looks worthwhile as a mod and no worries over reverse biasing of the caps. It's good 🙂
Hi,
Eeasy to bias one of the series connected caps with a few volts from a simple floating supply, then no worries of reverse voltage over the caps.
Eeasy to bias one of the series connected caps with a few volts from a simple floating supply, then no worries of reverse voltage over the caps.
Really this sort of thing needs to be logged over time with a chart recorder/data logger. and it will differ depending on your mains supply and the different loads connected to it in your area.
perhaps try the test under fixed conditions and inject dc to simulate a worst case and see if the circuit works.
this could be done at much reduced voltages for safety.
-Dan
perhaps try the test under fixed conditions and inject dc to simulate a worst case and see if the circuit works.
this could be done at much reduced voltages for safety.
-Dan
Hi Dan & 4fun ,
They are are good points. My set up was just a litle experiment into how effective it all was. I tend to forget that we generally have pretty decent mains supplies in the U.K. and that many locations will be far worse, not just in voltage stability etc but the purity as well.
There are always ways and means and biasing the caps is certainly one option but adds greatly to the overall simplicity of the concept. A wattless dropper maybe ? to derive a bias voltage.
Overall though it seemed to work well, and despite initial reservations I would be happy to use it - on the results so far 🙂
They are are good points. My set up was just a litle experiment into how effective it all was. I tend to forget that we generally have pretty decent mains supplies in the U.K. and that many locations will be far worse, not just in voltage stability etc but the purity as well.
There are always ways and means and biasing the caps is certainly one option but adds greatly to the overall simplicity of the concept. A wattless dropper maybe ? to derive a bias voltage.
Overall though it seemed to work well, and despite initial reservations I would be happy to use it - on the results so far 🙂