I will be starting a preamp project in the next few weeks, with another amplifier in the next year. I would like to hear some opinions on filtering RF from the AC line. Not really addressing speaker lead pickup, signal lines, or rectifier switching noise for this thread. I see a few options, or combination thereof:
1. Installing Corcom type filters (custom or OTS) immediately after input fuse. Most Off-The-Shelf modules use both X and Y caps, though Morgan Jones seems to favor a design addressing differential RFI only (I believe he says the Y caps for common mode add to the ground noise). Others seem to disapprove of these devices entirely, with the claim they degrade sonics.
2. Using electrostatically shielded transformers for CM noise rejection. I like this idea, along with using a split bobbin design for filament transformers. Unfortunately, many of the power trannies I would like to use do not have the shield, and I don't want to order customs.
3. Installing ferrite beads on the power supply, and adding ceramic caps to the heater pins. I also think common mode chokes for the heater supply is fairly easy and cheap.
4. Here and there I see someone using CM chokes in their B+ supply.
So my question is what do you typically build into your designs? All or some of these, and are there any others? I would always shoot for the shielded transformer if available. My preamp won't have one, so what would be the best options?
1. Installing Corcom type filters (custom or OTS) immediately after input fuse. Most Off-The-Shelf modules use both X and Y caps, though Morgan Jones seems to favor a design addressing differential RFI only (I believe he says the Y caps for common mode add to the ground noise). Others seem to disapprove of these devices entirely, with the claim they degrade sonics.
2. Using electrostatically shielded transformers for CM noise rejection. I like this idea, along with using a split bobbin design for filament transformers. Unfortunately, many of the power trannies I would like to use do not have the shield, and I don't want to order customs.
3. Installing ferrite beads on the power supply, and adding ceramic caps to the heater pins. I also think common mode chokes for the heater supply is fairly easy and cheap.
4. Here and there I see someone using CM chokes in their B+ supply.
So my question is what do you typically build into your designs? All or some of these, and are there any others? I would always shoot for the shielded transformer if available. My preamp won't have one, so what would be the best options?
Many COTS EMI filters wont handle amplifier psu duties due to the very high peak currents. The common mode choke in them is made to a price.
Some people dont like large X caps, but find smaller ones to be ok (ei 4.7nF v 22nF.) Y caps put leakage currents on the ground.
Ive ripped the ferrite toroid out of disused computer switching power supply and wound a common mode choke on it.
A drastic measure Ive seen taken is using balanced power. A large (preferably shielded) transformer provides 60V-0-60V power instead of the usual 120V-0. This is not to code outside the amplifier in most residential settings. The transformer must be large to avoid degrading attached PSU performance.
Some people dont like large X caps, but find smaller ones to be ok (ei 4.7nF v 22nF.) Y caps put leakage currents on the ground.
Ive ripped the ferrite toroid out of disused computer switching power supply and wound a common mode choke on it.
A drastic measure Ive seen taken is using balanced power. A large (preferably shielded) transformer provides 60V-0-60V power instead of the usual 120V-0. This is not to code outside the amplifier in most residential settings. The transformer must be large to avoid degrading attached PSU performance.
Tweeker said:
Agreed, but I will be running all choke input filters (including on DC filament PS), and PSUD shows me a reasonable crest factor on current.
I would also suggest running about 10X rated current i.e. if amp draws 2A current, I would pick a 20A rated Corcom module.
So your preference is to run more choke than cap, and eliminate the Y caps?
Some people dont like large (inductance wise, more current rating is better, asides stray capacitance) CMCs either. I cant say I have a preference on choke H v cap F, though I do favor eliminating Y caps (ideally with balanced power). I was reporting prior discussion at this site.
The thing that really needs eliminating is all the triacs in the house.
The thing that really needs eliminating is all the triacs in the house.
I live in a fairly large city with a lot of rf sources nearby, I have never found it necessary to do anything other than pay attention to internal signal grounding paths, and to use power transformers with electro-static shielding.
Those corcom modules are more effective for keeping conducted emi out of the wall socket from high speed digital noise sources like pcs than for anything else imho.
Generally I would otherwise echo Tweeker's comments about x and y caps as well as cm chokes.
You are far more likely to have problems with line and low level (phono) inputs than power or even speaker lines (as long as global feedback is not employed)
Fully enclosed metal chassis will work wonders if you use a little aforethought to keep most of the rf out of the box.
RCA jacks should never be directly grounded to chassis, either star grounding or bussed grounds back to a single star ground is preferred, however in RFI rich environments small ceramic caps from the jack gnd shell right to chassis will usually help. Grid stoppers on input tubes, and in some cases resistors right at the jack will help a lot as well - I have never had to use a bypass cap on a signal lead, but if needed use the smallest value that does the job and insert some series resistance ahead of it so that the driving source does not become unstable. Silver mica, and some low inductance foil types may do the job and not degrade sound quality too much - ceramic types are to be avoided due to the distortion they can generate when audio is impressed across them.
IMO for whatever reason in tube gear I have never found that a corcom entrance module results in better sound when employed - in fact much the reverse. I won't use them except in test equipment and digital projects.
Those corcom modules are more effective for keeping conducted emi out of the wall socket from high speed digital noise sources like pcs than for anything else imho.
Generally I would otherwise echo Tweeker's comments about x and y caps as well as cm chokes.
You are far more likely to have problems with line and low level (phono) inputs than power or even speaker lines (as long as global feedback is not employed)
Fully enclosed metal chassis will work wonders if you use a little aforethought to keep most of the rf out of the box.
RCA jacks should never be directly grounded to chassis, either star grounding or bussed grounds back to a single star ground is preferred, however in RFI rich environments small ceramic caps from the jack gnd shell right to chassis will usually help. Grid stoppers on input tubes, and in some cases resistors right at the jack will help a lot as well - I have never had to use a bypass cap on a signal lead, but if needed use the smallest value that does the job and insert some series resistance ahead of it so that the driving source does not become unstable. Silver mica, and some low inductance foil types may do the job and not degrade sound quality too much - ceramic types are to be avoided due to the distortion they can generate when audio is impressed across them.
IMO for whatever reason in tube gear I have never found that a corcom entrance module results in better sound when employed - in fact much the reverse. I won't use them except in test equipment and digital projects.
That's good advice.
One note, though, is that I am unfortunate enough to not have a readily available transformer with a shield, and could not locate any off the shelf's that fit my req's.
If I could, I would be less concerned about RFI. But now I face the H-L capacitance of my power tranny, and that can be significant. I am also vain enough to insist on using exotic wood chassis with top nickel plate, so I lose some chassis shielding. Nickel does provide a little magnetic shielding, tho.
One note, though, is that I am unfortunate enough to not have a readily available transformer with a shield, and could not locate any off the shelf's that fit my req's.
If I could, I would be less concerned about RFI. But now I face the H-L capacitance of my power tranny, and that can be significant. I am also vain enough to insist on using exotic wood chassis with top nickel plate, so I lose some chassis shielding. Nickel does provide a little magnetic shielding, tho.
You could line said fancy chassis with aluminium if it really needs to be. No amount of PSU filtering will help if the RFI is coming into an antennea on the amp.
On line filter C and L, it really doesnt take much to achieve a <160khz corner.
On line filter C and L, it really doesnt take much to achieve a <160khz corner.
Copper tape works wonders on the inside of wood chassis, and you can get it pretty cheap at most garden centers where it is sold as a slug repellent barrier. All that is required in addition is a metal bottom plate that is effectively connected to the shield and top chassis plate.
I should mention the tape usually has adhesive on the backside and adheres well to wood. Score the area where the tape overlaps to assure a good electrical connection and/or solder.
I should mention the tape usually has adhesive on the backside and adheres well to wood. Score the area where the tape overlaps to assure a good electrical connection and/or solder.
Thanks for the copper tape purchasing tip. How does it repel slugs? Is it some sort of chemical reaction, or are high voltages involved? I imagine with their good contact, a few volts would be enough.
Tweeker said:
Ah, not so much a repellent then, more a death strip. I expect copper's toxicity is why my previous house used lead piping for the incoming drinking water. 😀
Getting back on topic, a foil E/S screen on the mains transformer or dual bobbin construction are the best ways of preventing mains-borne RF noise being capacitively coupled to your electronics. Compact internal wiring minimises the length of aerials to pick up noise, and a totally enclosed chassis metal chassis will screen against electrostatic pick up. Careful attention to earthing (short, fat wires to the right point) also helps greatly. Oh, and check for stability - marginal stability will enhance RF pick-up.
EC8010 said:
Good, that confirms what I have read elsewhere, and where I was leaning. I guess I will have to continue looking for a mains xfmr that fits the bill with an e-shield.
Sowter has one that would fit perfectly, but is $220 USD !!
Allied has a shielded model for $41, but the output is a little higher than what I want.
Most Lundahls don't seem to carry the shield, and they won't custom build. Any other sources off the shelf?
There used to be a very cheap way of obtaining the isolation. In the UK, Maplin Electronics used to sell a transformer kit with a dual bobbin, one half already wound with the primary, and the other empty for you to wind your secondary. This was ideal for obtaining weird combinations of heater voltage. Curiously, control transformers for industrial electrical systems are often made this way. A dual bobbin is almost as good as an E/S screen, but much, much cheaper.
I was planning on using 6A filters at the IEC input to my S/S. power amps.zigzagflux said:
Is this saying that is a bad idea?
AndrewT said:
Possibly. In practice, a typical capacitor input supply tends to draw about 4-6 times the static load current as ripple current. So, work out what your peak loudspeaker current is likely to be at full whack, multiply it by six, divide it by the turns ratio of the amplifier's mains transformer and see if 6A rating is enough. I'm guessing it probably isn't.
Thanks EC,
170W into 8ohms =6.5Apk into nominal impedance.
turns ratio is 240/40=6 and multiplier is also 6. So it appears 8ohms is right on the limit for a ClassAB SS amp.
Is this peak charging current limit to prevent the ferrite toroid becoming magnetically saturated? or does the winding prevent saturation and it is then a heat problem in the wire?
170W into 8ohms =6.5Apk into nominal impedance.
turns ratio is 240/40=6 and multiplier is also 6. So it appears 8ohms is right on the limit for a ClassAB SS amp.
Is this peak charging current limit to prevent the ferrite toroid becoming magnetically saturated? or does the winding prevent saturation and it is then a heat problem in the wire?
All 3XX series Hammond power transformers appear to have an electrostatic shield brought out on a separate lead, and this is true of the ones I have purchased. They are a little more expensive than the 2XX 120V primary series, but run much cooler, and are very quiet in operation as well.
AndrewT said:
Yes. At saturation, an inductor has zero inductance and therefore zero filtering effect. Secondly, saturation is nothing like as sudden as you might think. Inductance begins falling quite a way before complete saturation.
kevinkr said:
With great hope I double checked their website. As I suspected, I did not see any indication of an e-shield in their documentation. They only specify concentric wound for low stray fields (but high interwinding capacitance).
Just to be sure, I called the factory. He confirmed there is not a shield, but that the green wire you refer to is simply an internal chassis ground.
🙁
Oh well, the Allied isn't the worst option. Hope it runs quiet (at least low vibration).
My apparent mis-apprehension about electro-static shielding in the Hammond 3XX series was based on this document where there is a gray lead (NOT GREEN!) labeled SH. Mine all have this lead and IIRC it does not connect to the core:
- Status
- Not open for further replies.