advertisement


What's the best DIY DC blocker circuit?

irb

pfm Member
I have the dreaded buzzing toroid problem, in two different power amps. I reckon I need to do something about it. I've read a fair bit here on pfm and elsewhere, and here's what I've learnt:
1. The most likely cause is DC offset on the mains, and the prescribed cure is a mains DC blocker of some sort.
2. There are several commercial products available but some of these are expensive, and most get mixed reviews. Some include other filtering, which may or may not be a good idea.
3. A DIY solution is possible, as a DC blocker is not terribly complicated. Rod Elliott's pages describe a possible circuit, here: https://sound-au.com/articles/xfmr-dc.htm. Several sources, including ATL Audio and Sjöström Audio, will supply a suitable PCB or even a built module, using Elliott's circuit or a variation on it.
5. The heavy duty DC blocker by Sjöström Audio (their DCT03) is a particularly well-liked incarnation of this type of circuit - more than one fishie has good things to say of it.

So, why not just buy a Sjöström DCT03 board, along with the necessary diodes and caps, and build my own DC blocker? Well, because the web has offered up a different DC blocking circuit, which might be somewhat smarter. It was suggested by a regular poster on AoS some years ago - see here: https://theartofsound.net/forum/showthread.php?11219-DIY-DC-blocker

The circuit looks like this:

DC_Blocker01 by grilled snapper, on Flickr

If it will work, it offers two advantages: first, it is simpler (why build a blocker with sixteen diodes if only two are needed?); and second, it should block all DC votages, whereas the other circuit has a limit of about 0.7v for each diode in the chain (so something like 2.8v for the DCT03).

What do others reckon - will it work, or is there a good reason why so many people use the more complex option?
 
The more diodes and more caps the more dc it blocks, simples. The big sjostrom is my goto. I have smaller units, 2 caps six diodes, they're not as effective.

It's tried and tested, this is knocked up by some bloke on AOS
 
Thumbs up for the DCT03, it cured a friends DC buzz and hot transformer on his New Marantz amp
This is the one I built for him
DCT03-3 by Alan Towell, on Flickr
DCT03-4 by Alan Towell, on Flickr
I did change the caps in one of the filters to work at a slightly different frequency and the changed the mains connectors to the DCT03 board to Wago type as the give a much firmer grip on the cable
IMG_3958 by Alan Towell, on Flickr
Alan
 
Powercons, fofelix style rfi filters, I have all the parts for just the same built sat waiting for a rainy day.
 
It's tried and tested, this is knocked up by some bloke on AOS

I agree! It's very offputting that no-one else seemes to be using the simpler circuit. But I suppose I'd like to get my head round the options, even so. I suppose I was half-hoping that someone here might have tried the AoS circuit. Or be able to comment of whether it would work.

The more diodes and more caps the more dc it blocks, simples.

Well, the circuit on AoS is a bit different - the diodes don't shunt the caps, so the value of DC it could block isn't limited to the usual 0.7v ish per diode. (If it works!)
 
Last edited:
........Rod Elliott's pages describe a possible circuit, here: https://sound-au.com/articles/xfmr-dc.htm. Several sources, including ATL Audio and Sjöström Audio, will supply a suitable PCB or even a built module, using Elliott's circuit or a variation on it.
5. The heavy duty DC blocker by Sjöström Audio (their DCT03) is a particularly well-liked incarnation of this type of circuit - more than one fishie has good things to say of it.
ESP site is very good and well worth reading.
The circuit looks like this:

DC_Blocker01 by grilled snapper, on Flickr
THAT circuit won't work. The diodes will only conduct through the adjacent capacitor as they face each other and are in 'anti-series'.
The diodes are only to stop large voltages occurring across the caps during current surge at turn-on. The caps then need to be large enough to produce only a small voltage across the pair at full power current so that the diodes don't conduct. As Rod says: 'The diodes are used to protect the capacitor bank, but it is the capacitor that blocks the DC - not the diodes.'
 
ESP site is very good and well worth reading.

THAT circuit won't work. The diodes will only conduct through the adjacent capacitor as they face each other and are in 'anti-series'.
The diodes are only to stop large voltages occurring across the caps during current surge at turn-on. The caps then need to be large enough to produce only a small voltage across the pair at full power current so that the diodes don't conduct. As Rod says: 'The diodes are used to protect the capacitor bank, but it is the capacitor that blocks the DC - not the diodes.'

Thanks! That makes a lot of sense.

The function of the diodes has been puzzling me. In the first circuit on Rod Elliott's page, he had the electrolytic caps in anti-parallel, and he implied that the function of the diodes was to avoid damaging the caps by applying reverse voltage to them - which makes sense. But when, further down his page, he put the caps in reverse series, surely that did away with the need to protect the caps against reverse voltage?

So I began to wonder why you couldn't dispense with the diodes altogether. However there is, as you say, the issue of turn on currents. I'm wondering, though: can you deal with that by making the caps big enough? The Sjöström circuit that others have had good results with increases the capacitance to over 100,000 uF. Am I right in thinking that the voltage you want to keep low is the drop produced by the capacitive reactance? Because with 100,000 uF, the reactance at 50Hz is only around 0.03 ohms, I think. So you'd need a surge current of almost 100 amps to get a voltage drop across the caps of 2.8v - which is the point that the chains of 4 diodes would start to conduct, if I've got this right. So the diodes in the Sjöström board might never actually be doing anything - no? (But I may have gone completely wrong somewhere!)
 
In the link to Rod Elliott's page, figure 1 represents a hair dryer or similar at half-power (half-wave rectified) which just draws current from one half of the sinewave, and that will produce a DC offset because of the impedance of the incoming mains and your ring circuit, represented by the 800 milliohms. Fig 3 represents that same half-wave rectified load, but with a blocker before the transformer on the same circuit. He then goes on to say 'While electrolytic caps can withstand a small reverse voltage (around 1V is typical), in the interests of longevity it is probably better to use the caps in series.' so instead of having a pair of 1000uF caps in anti-parallel (each of which suffers reverse voltage on opposite swings of the sinewave and thus splits the reverse leakage current) he uses a pair of 4700uF in anti-series so that as one is charged per the polarity markings, it protects the other against reverse leakage current and vicky-verky on the opposite swing.
Yes, lots of good (LOW ESR) capacitance will reduce the voltage across these caps at high (surge) currents. The diodes ensure no more than about 1.5-2 volts across the caps for 2 diodes in series, and as you say, higher V for more diodes.
 
Thanks for the replies. I was thinking of ordering a DCT03 board, and the parts, but I've done a little research and it seems some of the parts are rather hard to get at the moment. (None of the Vishay diodes available until April at least.)
I know there was a group buy a year or so ago, and quite a few fishies took part. I don't suppose anyone has a Sjöström blocker they no longer need, and would be willing to sell? Or parts that they never got around to using?
 
You could substitute the diodes with any that fits, even TO220 footprint types would probably drop in. Just look for other makes with comparable PIV and I-forward ratings.
 
You could substitute the diodes with any that fits, even TO220 footprint types would probably drop in. Just look for other makes with comparable PIV and I-forward ratings.

Thanks, Martin. Yes, maybe I could. There are some TO220 types available, I think. But the DCT03 board packs the diodes in pretty tight, and even if I could get TO220 types to fit, it wouldn't be elegant. If I can't get diodes in the right package I'm not sure the DCT03 board makes sense.
 
Nothing special, 6A 600v, I have p600j in mine widely available even on ebay

So they are. I was looking at Digi-Key, Farnell etc. Is eBay OK as a source for this type of component, generally?

Edit: Also, I see one seller with the Vishays, but they're £6 each, which is £96 for 16 - too much really. Sixteen unbranded equivalents can be had for only £4 or so total. Are the cheap ones really likely to be OK, though?
 
Last edited:
I've been reading more on this issue, and this morning I came across a thread on audiosciencereview (started by Colin Wonfor, as it happens). One post (here: https://www.audiosciencereview.com/...-to-help-stop-transformer-hum.948/post-455575) mentions a DC blocker produced by Akitika, who I'd never heard of, but who apparently make a LM3886 based power amp kit in the US. The photos and documents show that the Akitika blocker uses the same circuit as was suggested on AoS. (Actually it has the polarity of the diodes and caps reversed, but that makes no difference, I think, as long as you reverse both.) So, to be clear, it has a cap and diode in series for each half of the mains waveform. The diodes block reverse current from the adjacent caps, and I guess that the forward voltage drop across the conducting diode limits the reverse voltage that the adjacent cap sees. No?

As far as I understand it, this circuit will work. It has a couple of advantages: it should stop all DC, not just up to the switch-on threshold of the diodes; also it is cheap - 2 diodes, 2 caps, rather than 16 diodes and 6 caps, and a kit can be bought for $25 including pcb, diodes and caps.

But it will have a few disadvantages. For one, I guess it won't have the same protection against switch-on surges that the shunt diodes provide in the Rod Elliott/ATL/Sjostrom circuit. The ripple current rating of the caps would need to be good, to cope with surge current. Also, the current will be flowing through one diode or the other all the time, so they'll be switching on and off constantly, which might be noisy. Third, compared to the Sjostrom DCT03, it has a lot less capacitance, so more reactance at 50 Hz. If you were making your own version of this circuit, you could use bigger caps and add more caps in parallel, which would help with the first and last points. But noise from the diodes would remain.

Does that make sense? I'm not actually thinking of going with the Akitika circuit - I'm really asking to check if I'm understanding this stuff correctly.
 
51703129715_b1cdc335d7.jpg

As I see it: when 'L' is positive, the left diode conducts, charging the right capacitor and protecting the left capacitor from excess reverse voltage. This doesn't stop the left capacitor passing leakage current, and there is no way that this will stop a high voltage across the right capacitor under high current conditions as the right diode is reverse biased. When 'L' goes negative, then the right diode charges the left capacitor etc. So maybe I was wrong in post #6 above to state it won't work. However, the diodes now form part of the 'conduction' cycle but provide no 'protection' against high current scenarios.
All in my humble opinion of course.
 


advertisement


Back
Top