another neophyte question about AC voltage

[pone tony]

Can someone please explain to me:

Since AC currents can cause ground loops (hum), why bother with them at all?

Can't the heater filaments just be run on DC, since AC voltages in RMS are completely equivalent to DC voltages?

Am I wrong in asserting that the only reason heater filaments are run off of AC is because it's easier to just do it that way rather than add additional capacity to the B+ (DC) supply?

[Chuck H]

"Am I wrong in asserting that the only reason heater filaments are run off of AC is because it's easier to just do it that way rather than add additional capacity to the B+ (DC) supply?"

If your suggesting that the filaments be run from the B+ supply, then yes, you are wrong. The B+ rail will not supply the current needed for the filaments. As for running the filaments from a DC supply...I've never tried it, but many have and most conclude that the resulting regulator circuit and it's sensitivity to the heat created in tube amps and it's size due to current demands make it an impractical solution. There are many methods for quieting filament hum developed for guitar amps. Elevating the ground reference (positive bias) and using a twisted pair of leads for some inductance seems to offer enough of a solution for most aplications. If you need more noise rejection that that your very particular in your needs and may benefit from an enormous dc filament supply. There are examples used in high fi systems that may be useful if you really want to pursue this. But IMHO it's overkill for instrument amplification and is best resrved for sound reproduction amplifiers.

HTH

Chuck

[dai h.]

it's possible to get a quiet amp with AC heaters. One good example--the Soldano SLO100. There is some literature on having a quiet amp with AC heaters in the Radiotron book as well. If you do DC heaters and do it wrong, it can actually turn out worse with lots of hum. It can be more complicated too with more parts, huge caps, regulator, etc. The notion that running AC heaters is fundamentally wrong and therefore the answer to get a quiet amp is DC heaters is understandable but not really correct.

[Enzo]

First off, ground loop is a specific thing, and not a generic synonym for hum. The hum radiated by poorly places AC lines near signal lines is not a ground loop.

Good design practice with AC heaters will result in a quiet amp. Since adding all the extra stuff to run the heaters on DC adds to the cost and complexity of an amp, why do it when a quiet amp can be made with simple AC?

[mooreamps]

Since AC currents can cause ground loops

ahm, AC currents by nature do not cause ground loops. Ground loops are caused by having more than one path to ground from a singular circuit node.
-g

[dai h.]

but if you keep making multiple redundant connections a lot then you end up with a ground plane which is good which must be why an amp like an old Fender with redundant connections to ground must work with reasonably low noise plus things like grounds being grouped together correctly, correct layout.

[Enzo]

I am not sure, but I don't think that is what he meant.

A classic ground loop is two perfectly fine pieces of gear, both plugged into the same mains outlet. Then when their inputs are connected together, such as when using an A/B switch, it hums. There are two paths to ground from the input. If the currents inthem are not equal, the voltage dropped across their resistances won't be either, so current then has a net flow between them. SInce that current is usually AC, it gets impresssed upon the signal.

In a guitar amp, making up an example. Lets say you made the heaters grounded on one side, and also you grounded the input jack at the chassis. The heater on the tube nearest the input jack - and farthest from the power tranny - has its return current flowing through the chassis. The chassis is its return "wire." The inoput jack ground return also uses the chassis. The inoput and the heaters then share that "wire." Hum results.

If we then rwired so the heater return went via wire, not chassis, and we ran a ground return wire from the jack, all to a common ground point (star ground), then none of the circuit elements shares a current path with another.

It doesn't have to be heater AC either, ripple current from a filter can make hum too.

[dai h.]

wouldn't it depend on where you connected the heater ground? If you dropped it on the chassis far away it seems to work but if you say (worst case scenario) connected directly to the input jack ground lug, then I think you would get hum. The return path for the 1st tube DC (the static current since it's in class A usually around 1-3mA) would return through the chassis but again depends on how you connect and again you wouldn't want the heater current return (300mA x preamp tubes + .9A to 1.5A x power tubes) connected to the 1st stage tube cathode end ground but way far away grouped with the other higher current stuff it should be okay, or at least it seems that way. I mean, there must be some logic why Marshalls and Fenders with multiple grounds on the chassis can be reasonably quiet. I see it as sort of a plane and a wire the whole entire line from the gtr. end to the safety ground end or power source end. Or sort of picture a bus wire that grew and grew and got bigger and bigger into a plane.

[Enzo]

Because Fender, Marshall et al go for "good enough." What passes on a Fender would not on a high end stereo system.

I made up the heater example to demonstrate HOW it can happen. CLearly if we take steps to avoid that, it doesn't. If you removed those jumpers to ground along the Fender eyelet board and made them all into star ground branches, you might achieve a somewhat reduced hum level, not that it is excessive as it sits.

I'd agree that the class A B+ return would not likely add hum, but ripple from something like a reverb driver taken off an early B+ node with heavier ripple than the preamp node might. The return of the filter cap would be the hum source.

I have had to redo the grounds on a few poorly designed amps. I recall one cheap combo - is Core a brand? - like a keyboard amp or something, wherein there was a relay down near the input stage, and the ground for the relay coil was right there on teh same trace as the input stage ground. The not all that well filtered DC for the relay shared about 6 inches of trace from the first stage back to the power supply. Whenever the relay was energized, it added hum to the signal. cut the relay off the trace and ran its own wire back to the filter cap, no more hum.

There is something called "ground bounce" that digital engineers have to worry about. On logic boards, many chips share common power busses. Transitions to logic low - grounded state more or less - can cause currents to flow through the ground traces. Over the course of that ground trace, there is resistance. It is as if that resistance was added in series with the ground returns. Cirrent through that resistance causes a voltage drop across it. That means that at the far end of that ground trace, what is "ground" there can be a portion of a volt above ground. SInce logic levels are voltage determined, this can conduse a susyem if severe enough. SO they have to consider that when designing power distribution on the board. I know of it as a concept, but to an engineer it is a real consideration. My description is probably a little off the mark, but hopefully it gets the idea across.

[dai h.]

I've read a bit of grounding on PCBs that had to do with the sort of stuff you're talking about in technical articles in magazines. I don't remember (nor understood) all of it--things like segmenting grounds and having send on top of returns for lower noise, so I know grounding for things like digital circuits can be complicated, but as to the ground plane being inferior, how about like on a PCB where you have a ground plane and not traces in a star shape, or maybe a star point that expanded into a plane (or maybe a giant star?) as I mentioned before. Conceptually and in practice if that works and is okay to do on a PCB, why not with the chassis? I question that as being thought of as "wrong". I guess a typical chassis is not as conductive as copper, but there is a lot of mass, so should be low R. I stumbled on a pic of an old Fender Champ recently and noticed that the inside of the chassis looked shiny like it was maybe copper, and all the grounds were soldered to it. Maybe Leo had a radio background, and as I understand for HF circuits you use ground planes and maybe his background might explain the approach?

I'd agree that the class A B+ return would not likely add hum, but ripple from something like a reverb driver taken off an early B+ node with heavier ripple than the preamp node might. The return of the filter cap would be the hum source.

Well the whole idea (which I think I understand a little bit) is to try to be concious of what is going on in the grounds, and *apparently* even some engineers aren't quite aware (at least according to some of the articles). So if someone is aware they wouldn't hook the ground up in the wrong place.

Because Fender, Marshall et al go for "good enough." What passes on a Fender would not on a high end stereo system.

Fender seems better than marshall with the grounding. Not completely scientific, but I think Marshall seemed to make more errors off the top of my head (putting the bridge diodes ground in the wrong place and using 10 ohm R seemingly to correct incorrect ground placement comes to mind). So not quite understanding but just because they put something in the wrong place doesn't seem to equate to putting grounds down on the chassis and using it as part of the whole scheme necessarily being wrong. More that they did it wrong rather than using the chassis being wrong. Re: Fender noise vs. high end stereo noise, I'm not sure how fair it would be to compare them. Regular electric gtr. signal transfer is sort of fundamentally flawed anyway, being high Z unbalanced. You'd want to do what Les Paul does, then you could compare them. Besides that there is noise ALL OVER various recordings--Jimi, SRV, Eric Johnson, etc. and people still seem to be able to enjoy and put people like that and their music on a pedestal, so even if something is not quite clean and hi-fi like Steely Dan, it can still work obviously.

[Enzo]

We can quickly get over my head, if we have not already, but in high freq circuits such as RF or digital, a few inches of trace becomes a transmission line and all th attendant headaches of termination and reflections and such come into play. Ground planes are part of that equation.

But I know you have to be careful on pcbs, because a ground plane a millimeter away from all your traces becomes capacitance to ground all ovefr as well, potentially high end might suffer, not that in a guitar amp we worry about real high end.

A chassis is low resistance, sure, but if there is half an ohm from input jack to power supply ground, a mere 100ma flowing through it will drop 50mv. COmpare that to 100-150mv guitar signal at the input and it is a substantial amount of noise. What aer real world currents? I don't know, it is just a hypothetical.


I don't think Leo was into RF, it was no secret he lifted his designs from the tube maker manuals. That was why the RCA folks drew up the circuits in the first place - to make it easier for design companies to use RCA tubes in their products. I seriously doubt any RF wary design effort went into a Champ. SIngle ended amps like Champs are noisier anyway. I have a factory bulletin on a CHamp telling us service centers NOT to try to "cure" the hum, it was normal.

Marshalls 10 ohm from ground to chassis - and the cap and diodes in parallel - is there to help separate the circuit grounds from chassis. That resistor lets the circuit ground refer to the chassis without being wired right to it. Perhaps we should call the circuit grounds circuit commons instead. The common is then grounded through the 10 ohm resistor.

I am not really comparing Fenders with stereos, just pointing out that hum standards in guitar amps are a lot lower than in stereo systems. That is why you won't find a nice stereo amp with grounds to chassis all over the place. Well, less likely anyway. And that is why Fender et all don't bother with fancier grounding schemes. Leo "Good Enough" Fender. WHy waste money on 5% or 10% resistors when 20% resistors are good enough? WHy waste money on fancier grounding schemes when the level of hum in this Bassman is good enough?

Fender didn't care, it is the tweakos among guys like you and me that worry about the extra little performance issues. They are the ones using star grounds and such that Fender never bothered with, and they are the ones asking questions like the original post in this therad.

[Ptron]

FWIW regarding the PCB/ground plane thing: I'm a tech for a small company that makes high end processor, I/O, etc. cards for embedded computer systems. I'm currently being trained to do PCB layouts. We're talking about 14+ layer boards with hundreds if not thousands of nets. I mean these things get dense. I should qualify here and say that I'm pretty early on in my training, so I'm sure there's a lot I don't know yet, but...we have rules about clearance and not having sharp angles in traces. We have rules about plane edges being staggered. We have rules about laying out differential pairs and length matching of traces. We have rules about crosstalk between nets and tools that calculate it. We have rules about trace impedances and tools that calculate them. And so on and so on… Needless to say, there are lots of rules. But, no one has ever said squat to me about grounding. Just drop a via to the ground plane wherever it’s convenient. Connect it to other nearby grounds first if that’s convenient. Whatever.

I don't know why it doesn't seem to matter. It's funny you bring this up Dai because I had just been wondering about that. I'll have to see if I can get one of the engineers to explain it. Maybe even though a board can draw amps, no one current is big enough to matter.

[dai h.]

But I know you have to be careful on pcbs, because a ground plane a millimeter away from all your traces becomes capacitance to ground all ovefr as well, potentially high end might suffer, not that in a guitar amp we worry about real high end.

The aluminum "lid" in a Marshall cab (or adhesive backed alum. in Fenders?)--couldn't you think of that as a ground plane? It seems like a number of times I've seen this HELPING (i.e. affecting and mattering to performance despite the low frequencies being dealt with) with squealing apparently due to capacitive coupling, which should be helping to isolate individual wiring from each other even though they may be inches away from the wiring they are effecting.

A chassis is low resistance, sure, but if there is half an ohm from input jack to power supply ground, a mere 100ma flowing through it will drop 50mv. COmpare that to 100-150mv guitar signal at the input and it is a substantial amount of noise. What aer real world currents? I don't know, it is just a hypothetical.

yes, exactly why where you put the grounds matters. So you would (especially) try to keep the higher current returns in a smaller loop (keep higher current returns physically shorter or closer together and grouped more or less together to keep the voltage drops smaller). Conversely, since the lower (and super tiny) current returns means smaller voltage drops in the returns and smaller errors, you can stretch them out more with less probs. Plus I suppose less energy so less radiation for lower current loops, higher for higher. Also, so a trace has an R, a chassis could be thought of as many many traces with a specific R side by side. Maybe infinitesimal amount of R but if you have a trace with an R, then keep paralleling it with something the the same size and keep doing that, R should become lower and lower right? So maybe the sheer mass means superiority in less R and L.

I don't think Leo was into RF, it was no secret he lifted his designs from the tube maker manuals. That was why the RCA folks drew up the circuits in the first place - to make it easier for design companies to use RCA tubes in their products. I seriously doubt any RF wary design effort went into a Champ. SIngle ended amps like Champs are noisier anyway. I have a factory bulletin on a CHamp telling us service centers NOT to try to "cure" the hum, it was normal.

It says the Funk tube amp book for example, that "(i)n the 1920s, Leo Fender was a bookkeeper who got into ham radio as a hobby. This lead him to building and repairing radios for his friends, and renting out homemade PA systems. He eventually became a full time radio repairman... (Dave Funk's Tube Amp Workbook, page 1)". Another bit, "(t)he construction of this early Champ is interesting. A plain .060" steel chassis is formed by making one bend for the back panel. The chassis is PLATED with copper, not painted, as stated in another book. [my emphasis]Sounds like something built by a radio man.The copper provided an easy surface to solder to, good for grounds, and excellent for shielding." (Ibid, p.3) IIRC there was an old construction book on the Pete Millet site where they advised to use the chassis for the grounds.

So looking at things like that (RF circuits = use of ground planes as the norm. + ham radio guy + tradition?), maybe he just came from a time where it was normal to use the chassis and not that he was trying to reproduce radio frequencies through gtr. amps or anything.

re: the noisy (hummy) Champs, I do wonder if a builder took a more theoretical approach you couldn't get a better signal to noise performance (I guess I'm questioning the seeming or supposed inevitability of the hum).

Marshalls 10 ohm from ground to chassis - and the cap and diodes in parallel - is there to help separate the circuit grounds from chassis. That resistor lets the circuit ground refer to the chassis without being wired right to it. Perhaps we should call the circuit grounds circuit commons instead. The common is then grounded through the 10 ohm resistor.

the 10 ohm is something that was described by someone tweaking and guessing that the 10 ohm made some tonal mojo and my guess to that was that Marshall made an error (ended up with a ground loop) and fixed it by connecting to the same wrong point through a low value R (but I might be completely wrong since I didn't continue to take a closer look and I just said to try bypassing the 10 ohm with a wire and if he got more hum doing that, then my guess was that my guess was correct).

I am not really comparing Fenders with stereos, just pointing out that hum standards in guitar amps are a lot lower than in stereo systems. That is why you won't find a nice stereo amp with grounds to chassis all over the place. Well, less likely anyway. And that is why Fender et all don't bother with fancier grounding schemes. Leo "Good Enough" Fender. WHy waste money on 5% or 10% resistors when 20% resistors are good enough? WHy waste money on fancier grounding schemes when the level of hum in this Bassman is good enough?

Fender didn't care, it is the tweakos among guys like you and me that worry about the extra little performance issues. They are the ones using star grounds and such that Fender never bothered with, and they are the ones asking questions like the original post in this therad.

I don't know if I'd agree with that. For example in Champs, you have the AC line going near a VOLUME POT with it's high Z wiring for the volume/pwr. on-off sw., BUT later on blackface(or maybe earlier?), you have the power and stndby. in the back corner near the pwr. input which from a noise theory perspective I think this is more ideal (though maybe not from an ergonomic perspective where on the front where they are visible/noticable and easily accessed by the user might be better). (Therefore you saw an improvement--I think Bruce pointed this out a while back.) Besides that the guy is dead, so how does anybody know how the amps would have developed further (or if they wouldn't have) in that respect. Maybe they would have progressed, but of course no one lives forever so we can't fault him for that. On Marshalls, the grounding scheme changes from the copy of the decades old 5F6-A type scheme in the later JCM800s, so that could be taken as an indication of "caring" and wanting to make an improvement, although they probably could have done better (looking at the Soldano SLO100 scheme and layout which looks like there is more solid theory behind the way they do things to acheive low noise with high gain).

As far as stars vs. planes, from what last I was thinking about this stuff, since as I understand planes are better (and used) for high frequencies and stars are worse, then it might be a good idea to star the power section in an amp. Maybe that will give a tiny bit of help in preventing oscillation.

enzo, why do you have to be so difficult? just kidding of course, lol...

[dai h.]

(hmm... I guess it took me almost an hour to write my stupid post, lol...)

Ptron, thx I'd definitely be interested in what someone way more knowledgable has to say. (I'm just a hobbyist, so not much of it registered in my brain, but) I've read some articles on grounding in those sort of circuits, and yes I gather things can be quite complicated.

[R.G.]

But, no one has ever said squat to me about grounding. Just drop a via to the ground plane wherever it’s convenient. Connect it to other nearby grounds first if that’s convenient. Whatever.

I don't know why it doesn't seem to matter.

It's because it's digital. The routing of signal traces is CRITICAL and that's where you get that equal-length, constant impedance, differential, edge effects, stuff.

Something that's grounded, particularly an input that's always grounded will never move enough to cause a logic glitch, and with CMOS logic never pulls enough current to cause any change. So yes, nearest ground of any kind is fine. The noise immunity of CMOS inputs means that the local ground would have to move a huge amount to ever be a problem.

Analog voltages, by definition, have ZERO noise immunity. Logic is different.