Separate names with a comma.
Discussion in 'Amp Tech Center' started by elpico, Jun 5, 2018.
I tell you what...
This thread is really getting deep!
Yep this is inductive coupling we've been talking about. It's possible for signals to couple via the electric part of the EM field as well. In that case it's the rate of change of the voltage that's significant rather than the current.
Thanks for that Speedy454. I did learn this stuff once, but it's difficult for me to recall all the relationships just right. It's nice to have people to talk to about it because going through all this is really helping to refresh it in my memory.
There's an interview with Ken Fischer in which he explains his preference for turret boards/point-to-point because of the ability to run the connections in any direction. (As opposed to printed circuit boards with their single plane of connectivity.) This makes sense in the context of what we're discussing regarding the physical size of the current loops. Also it takes into account the direction in which the loops are flowing with regards to three dimensional space. (The same reason, I think, power and output transformers are preferably oriented in different directions. Keeping potentially interfering currents at right angles to one another is a widely used trick to minimize reactive inductance and, hence, noise.)
The downside of point-to-point that's been, uh, pointed out (sorry) is that all of those little wires inside your amplifier are minuscule antennae, no matter how neat the lead dress. So I guess it's like anything else; each side has its pros and cons, just take your pick and stick to it.
Maybe Mr. Trainwreck himself was on to something after all! No disrespect, mind you. Fischer, Dumble, Gjika, et al are interesting characters and very skilled craftsmen in their own right.
But I digress... So if the noise factor, if you will, of a current loop is related to its physical size, and said loop is "grounded," ie partially flowing through the chassis of the amplifier, would it be accurate to say the loop becomes larger, taking into account the current is now flowing through the entire physical space of the amp?
Interesting, I know loop antennas are directional and it does ring a bell that the orientation of the loop would be a factor in the possible coupling between them, I believe I did read that in some emc text or another. Sounds like Ken knows what he's talking about there.
Probably a good time for a reminder that I'm not a walking textbook for this stuff, or qualified to write such a thing. If you guys are hoping this will get down to the level of equations for determining the exact amount of crosstalk between two loops then you're going to be disappointed. You'll need to crack a real text book for that level of information. I usually remember the results at the end of the chapter (small loop good!) but the math used to arrive at that result leaks out of all the holes in my head shortly after. When I need to calculate some specific value I have to go looking for a book.
The question about using the chassis as part of the rectifier loop is a good example of this. I don't think I've ever seen that case considered in a textbook, it's not really something you'd see done in modern electronics. I do know that what I said about the current spreading out across the plate like a river is correct, but to what degree does that impact the loop's ability to radiate noise? I have only my intuition to go on there. Also I no longer work at the RF shop so I don't have access to equipment that would let me find out experimentally anymore. If anyone finds a textbook that holds that answer please do share it here.
interesting discussion about inductance.
so, every wire is both a transmitter of EMI, electromagnetic interference, and an antenna/receiver of EMI.
shorter wires and smaller loops are better.
does it have to be a loop?
I looked it up and technically it doesn't.
an open loop will pick up an electric field, but it takes a closed loop to pickup a magnetic field.
in amps it is electromagnetic fields, and open and closed loops pick those up.
how much electromagnetic flux?
one equation is:
flux = V * A * sin theta
where V is the voltage or voltage gradient, A is the area of the loop, and theta is the angle between the flux direction and the plane of the loop.
I do not believe magnetic or electro magnetic flux will cause current in a loop or a wire unless the flux is varying with time.
I found out more about twisted wires. This supports that loops are a big part of this.
not only does twisting wires together eliminate areas of loops, the smaller loops that are formed cancel each other out because twisting flips the area 180 degrees.
Does an antenna need to be a loop? above is the circuit symbol of an antenna and it is a loop.
below is a roof antenna that is not a loop.
then there is the Biscay cross.
the Biscay cross looks like a loop. It rotated and would give away the direction of radio signals.
No an antenna can be a loop, but doesn't need to be a loop. Loop antennas were a standard feature of AM radios so I'm sure you've all owned one before. This form factor was very common:
And since we're all tele nuts we all own multiple loop antennas in the form of single coil pickups. The loop area of a tele pickup may seem small but the wire loops around it thousands of times and net result of all those loops is something we're all too familiar with - they're extremely effective at picking up noise.
I'd like to talk about one more less than great loop in a typical guitar amp and then I promise I'll stop droning on about loop area.
The quality of electrolytic capacitors has apparently changed pretty radically in the last 70 years. I suspect they may have been quite prone to rupturing and leaking their electrolyte back in the 1950s because Fender layouts always seem to group every power supply capacitor in the amp into a little bunch, as far from everything else as they can get it. It's like they're being quarantined from the rest of the devices. Even if my guess about leakage is not the reason for this layout decision, I'm sure there must be a reason that forced them to do this because from an electronic perspective this is not at all what you want.
Let's trace out the current loop for V1 in a fender amp:
I don't even know what to draw for the part where they've used the chassis as a section of the loop, but no matter what you draw there it's clear that instead of making this loop the minimum size it could be, it's instead close to the absolute maximum size loop you could even fit on this board. And that's mainly a result of the decision to quarantine all of the power supply capacitors down at one end.
Now I can already hear people saying "yeah but the capacitor has very low impedance to AC signals so this loop is therefore low impedance and not likely to pick up noise". It's true the cap has low impedance so no interfering signal should appear at the capacitor, but the rising frequency of the interference sources in our environment these days means that this length of wire is no longer well filtered by a capacitor that far away. Look at the length of your phone, or the antenna mast on your wifi router. That's the length of wire that will resonate at the frequencies those things are putting out. So even if one end of that wire is "grounded" at the capacitor, the other end is far enough away that it's free to wave around at those frequencies and then we're back to the galloping cell phone interference sound from the first post.
Anyways, the fix is simply to move the capacitors to the things they're actually meant to filter. The first cap should be on a terminal strip next to the rectifier tube to reduce the size of that noisy loop and keep it far from the rest of the circuitry. The second cap is probably fine where it is. The third belongs down at the other end of the board between the two tubes it supplies. The connections between the capacitors should be made by a twisted pair.
One last note on the topic and then I promise I'll stop complaining One dimension of the loop we drew above was set (to some extent) by the large dimensions of a 1950s power supply capacitor. Choosing a very long capacitor with the leads at opposite ends means at least one part of your loop is going to be that wide. Modern capacitor designs aren't built this way anymore specifically because of this reason. Capacitors look like this now:
with closely spaced leads that don't force you into large loop areas. They're superior in every way to the big old dinosaurs of the 50s and they fit in very small spots so there's really no reason NOT to be putting them near the tubes they supply. Heck you can add a few more if you want, they only cost around $2. Sometimes we have a "bigger must be better" bias but I assure you that's not the case with electronic circuits. Smaller is better here.
Another loop and its proper grounding that should be mentioned, are the two loops of heater-grid capacitance hum, their influence, and the need and means of balancing them are covered quite well in this Radiotronics article.
Regarding placement of filter caps, Merlin (of course) talks about that in his book (possibly the same chapter on grounding). And some Hoffman kit layouts demonstrate this.
Vintage Fender amps (and others) use a single, shared common return path for everything, back to the wall. Often connections are randomly made direct to the chassis, theoretically with strong and weak currents flowing all over each other.
The next step in enlightenment seems to be the separate preamp bus, connecting to the chassis at the input jack. The PT, power supply connects to the chassis near the PT bolts. The only difference is that the connections to the chassis are more discrete and purposeful, and fewer, and placed farther apart. Theoretically the strong and weak currents are running through separate areas of the chassis, so the strong doesn't stomp on the weak.
And (back to filter cap placement) a further refinement would be to physically place each filter cap near to the circuit it powers. You still have a wire from the power supply to the preamp stages, it's just that the cap is further along this wire, instead of near the power. Is this because the unfiltered current is less noisy than the filtered, hence the desire to shorten the filtered run?
Somewhat separately from physical cap placement is the further refinement of separating the circuit into a subcircuit for each filter cap. Whether the cap is in proximity or not, cap+ feeds the preamp stage (triode and it's entourage), and the common path back from this section is isolated from every other part of the amp. So, instead of one common throughout the whole amp, and instead of splitting that one common into half placed onto the chassis 'far away' from the other half, we now have four or more discrete circuits and common return paths, one for each filter cap. All of these finally joining up back near the power supply, perhaps at the PT bolt.
The only way to isolate these circuits from the chassis is to use insulating bushings for each jack: input, reverb, footswitch, etc.
Now that everything is off the chassis, an even further refinement is to connect this one common 'return' point (not calling it ground, and saying 'common common' is confusing) through a cap/resistor network, such that the DC reference for the working part of the amp is no longer zero volts, but something slightly higher. I know this can be a good thing for noise control, but I'm not sure why. Anyone?
Some devices have more potential for noise than others. The dreaded ground loop - which we now know has little to do with 'ground', and everything to do with multiple circuits sharing a return path. Higher voltage circuits are more prone to noise.
An example of a low voltage 'loop', which is a total non-issue, is each of us holding a plugged in guitar. We touch the strings, we're directly connected to the amp's circuitry, who's common return path is through the wall socket, back to the panel. At the panel, all commons join together, and (as mentioned) also all safety grounds. Further, in a typical home installation, that common bundle connects directly to the metal rod driven into the earth just outside the building. But wait, we're standing on .... something, which we can assume has continuity back to the earth. So, our feet completes the ground loop. The amp now has two paths back, one through the wall, one through our feet. Because the voltage is extremely low, there is no noise to speak of, and we live to play on.
But consider an outboard reverb tank. It's a high voltage tube amp in it's own right. Both amp and tank plug into the wall. It doesn't matter if it's a different circuit, as all the commons join. But the devices are also connected through the instrument cable running between them. Now the current flowing in each device has a 'choice' which path to take back to the panel. Through it's own chassis / wall common, or through the instrument cable, into the other device, and out through it's chassis, etc.
This will likely be very, very noisy, unless some of the refinements I mentioned above are employed. Fender's 6G15 reissue goes to the trouble of isolating and elevating the chassis. That's saying something about the need, considering how much they could save (in volume) by taking a simpler approach.
One final thought: what about pedals? If we have a pedal that plugs directly into the wall, it's probably pretty noisy, for the same reason: Two high voltage devices, connected via instrument cable, and (implicitly) the wall common. But most modern pedals are low voltage. In addition, they don't actually connect to the wall. Either there's a battery, or an isolating transformer (wall wart).
Whew, I'm fried thinking about all this. I believe it to be pretty accurate, but I am the farthest thing from an expert. I did have to use all these refinements to get a noise-free Revibe build, so I've thought a lot about it. But nothing more. Corrections and clarifications are very welcome.
I believe this is correct. An EE explained to me years ago that it is the collapse of the magnetic field around a conductor that induces a current in another nearby conductor. He was talking specifically about AC power and signal cables running side by side. Hence things like 60 cycle hum.
Maybe I wasn't clear enough on that part, it is indeed the rate of change of the current in the loop that identifies it as a potential noise source. The rectifier current was selected as an example of a noise maker based on the plot that shows the current changing from zero mA to half and amp and back again very quickly. Similarly when we mentioned interference coupling via the electric field rather than the magnetic one, it would be the rate of change of the voltage that makes a potential noise candidate, not a steady high voltage. I've always been told that type of crosstalk is less of an issue than the inductive type in low frequency circuits (ie. audio) so I haven't given it any attention here. It's easily blocked by plain old shielded wire as well, magnetic coupling is not.
If I'm duplicating Merlin's work again I apologize for being repetitive. That link has been posted already and I had a read through it the other night. Since then I've been trying to focus on topics he left out, like radiated noise and loop area, rather than repeating what he's already so elegantly summed up.
I was lucky when I started out in this hobby in that many of the well known figures in the field were very active participants in the online discussion groups of the day. Merlin wasn't around back then but people like Ted Weber and Kevin O'Connor were likely to answer any question you had about speakers or amps directly. After some years Ted even created his own discussion forum on the weber site and he was incredibly generous with his knowledge. There's really never been anything before or since that can match the access to information about guitar speakers that he gave to all of us. Like the "aging" treatment touted by many companies now that make their new speakers sound like 50 year old ones. Diluted fabric softener misted on the cone and a tiny amount of acetone on a q-tip rubbed around the rings of the spider. When he got a hold of a new model of vintage speaker to study he'd post all the plots and parameters he created publicly for the benefit of all. Good luck getting that kind of information or interaction out of speaker companies today. It all seems to be hoarded as "trade secret" now. Thanks Ted. You are missed.
I bought a couple of Kevin's books back then and long before merlin his books always covered all the topics we're discussing here. He had modernized layouts in them for many popular amplifiers (champ, bassman, ac30, plexi, jcm800, hiwatt, SVT etc) and those included the improvements we're talking about like proper grounding, elevated heater voltages, and distributed power supply caps. So I was lucky that right from the first amp I ever built I had access to this information.
The thing you're talking about to avoid the ground loop between the amp and outboard reverb unit, that's called a ground lift switch and it's provided in each of Kevin's modernized layouts. Along with separate bias controls for each tube and other niceties. Most of us scattered when the forum we all used (ampage) went offline without warning one night. I didn't hear much of Kevin's voice online after that but I believe he still posts occasionally on diyaudio.
Oh, not a problem. I mentioned it for the person who hasn't yet found Merlin, or has just quickly browsed his excellent grounding article. There's more, there. Plus, it's always nice to have multiple explanations for (what is for me) a difficult topic.
OK. I understood a simple ground lift to mean just that, to bluntly remove the common return path for one of the devices (the tank in this case, because it's useless without the amp), so that it's return path is ALWAYS the instrument cable. In older amps, like Fender BF/SF, the ground lift poses a safety issue, because there was no separation from the chassis, and no dedicated chassis ground. So the chassis could go hot, and have no place to go.
Merlin (again, sorry) also uses the term [15.10], but mentions that completely breaking the connection between what he calls "audio ground" and "mains earth", presents safety problems, and that the small hum-block loop network is a compromise (implying more than just a ground lift). 10 ohm resistor *I think* essentially allows the power supply to use the near return path, but the other device, coming over the cable, won't prefer it. Forcing to each their own path. Also a cap so that the chassis still acts as an RFI shield (which it wouldn't do if ground were truly lifted, right?). Finally, a pair of diodes that will open when the voltage across them reaches 0.6V. So it appears blocked, but in case of a short, the unit doesn't become a hot nightmare with nowhere to go.
I'm right at the limit of my understanding here, perhaps a bit beyond. If this whole mess is normally called 'ground lift', well, I apologize for going the long way. Most of my experience with this stuff is a) recent, and b) on old amps, where ground lift meant simply that.
According to Rod Elliott, “The loop breaker works by adding a resistance in the earth return circuit. This reduces circulating loop currents to a very small value, and thus 'breaks' the loop”. I’m also a fan of multiple explanations, and though it’s aimed at builders of solid state hi-fi equipment, his article on grounding is well worth reading.
I'm taking a break halfway through. Good article.
Ah yeah I can see why you might be alarmed if you're thinking of the old fender amps with the ground switch. That was a very different situation. The problem there wasn't that switch, there was a problem with the design of north american wall sockets and plugs at the time. The plug looked like this:
Obviously that can be plugged into the wall two different ways. (pull it out, flip it over, plug it back in). Because of this the person wiring the amp had no way to guarantee the chassis would get grounded. It was a coin flip every time the user plugged the thing in whether the chassis would be connected to ground or 120Vac. This is why these old amps were unsafe, not the switch. It's kind of astounding that anyone thought this was an acceptable state of affairs, but I guess it happened.
That's not at all the situation now. We have three prong plugs that guarantee all devices have a proper connection to ground at all times. Although the term "ground lift" may sound like some kind of violation of that guarantee, if you do this properly every device remains fully safety grounded.
Rod Elliot's section on this seems to be addressing the topic of ground lifting devices that weren't designed with that function in mind. That's a bit more complicated, his jack is connected to the chassis so he's actually lifting the chassis from the earth. I don't think you want to get into that. If you build your devices with isolated jacks and a single point connection between circuit ground and the chassis then every chassis can remain fully connected to earth ground at all times.
In your situation with an amp and outboard reverb you have two earthed chassis and two circuits that each connect to their earthed chassis. The problem is when you connect the reverb to the amp their circuits are now connected to earth at TWO points: their own chassis and the other device's (via the signal cable). If you've built them with isolated jacks and a single point connection to the chassis then you can put a lift switch on that single point connection and toggle it on one device to remove the duplication. Everything is still grounded, both chassis were left connected to earth, and both circuits are still connected to one of the chassis.
Merlin's page seems to be a combination of the two ideas. He's talking about the same thing I am, leaving the chassis grounded and lifting only the circuit ground, but then he's copied the voltage limiting circuit from rod elliot's page, which is intended for a very different situation (lifting the circuit AND chassis from ground). I don't really see why he has it that way, but maybe he has a good reason. The series of events that would need to take place before those components could have any function is pretty out there. Let's go through it for the heck of it:
- you're standing on the wet concrete floor of your basement in bare feet
- you're playing an electric guitar that's plugged into an outboard reverb and then an amp
- there's an objectionable hum so you toggle the ground lift switch on the amp
- you're not being shocked
- you tire of playing guitar and start pulling cables out but you don't turn anything off first
- you disconnect the cable between the amp and reverb from the reverb but leave it plugged in to the amp
- with the amp's lift switch toggled the end of this cable is not connected to earth ground
- you're holding on to the end of this cable in your bare feet
- you're still not being shocked
- (you are now the ground reference for the circuit, which goes on operating normally)
- you're getting annoyed about how hard it is to get shocked so you ask a friend to come help
- the friend starts lifting the amp up and dropping it on the ground over and over
- you maintain your grip on the end of the cable
- after enough smashing something finally breaks loose inside the amp that connects the positive side of the power supply to the earthed chassis
- congratulations, you have been shocked
Personally I'm not going to loose sleep over a far fetched series of events like that, just driving to work is probably a thousand times more dangerous than that, but to each their own.