How does transformer tap grounding work?

[sahazel]

I've been building tube amps, and I've developed a question or two about power transformer grounding:

A common arrangement is to connect one PT secondary tap to (chassis) ground. With that tap connected to the chassis, and the chassis connected to safety ground, it makes sense that the chassis remains at ground potential, while the voltage on the other tap(s) swing positive and negative with respect to that reference. So far so good.

But, what happens when the amp has a two-prong plug, or is connected to an outlet with no safety ground connection? Especially in the case of a two-tap secondary, I'd now imagine that the floating chassis is participating in the full voltage swing on the secondary. If I were grounded and touched that chassis (or my guitar strings), wouldn't I be in trouble?

And then, I've noticed that, according to my multimeter, the usual tube amp arrangement with a grounded secondary center tap puts the hot wire from the power outlet at 60VAC with respect to the chassis, and the chassis at 60VAC with respect to the outlet neutral. I understand there's no circuit and thus no potential for ongoing current flow between the chassis and the outlet neutral, but still, with a peak voltage of ~85V with respect to neutral / ground, shouldn't the chassis be able to deliver a brief shock to anything connected to ground?

 

[mandu]

You should avoid using 2 prong powered amps where you can touch metallic chassis. Ground the amp for your safety from electric shock. There is some capacitance between the primary and secondary windings, which causes a small leakage current flow to secondary and so to the chassis. Depending on how the transformer is built and how you plugged the 2 prong to the power, and power line frequency, the leakage varies. One type is the primary and secondary overlap over insulation, another uses two separate bobbins, and another uses different magnetic core for primary and secondary. If the primary and secondary are wound overlapping, the leakage is most when the outer layer of the primary is the live. For many of us using a cloth iron, table fan, with two prongs, the leakage may not be felt. But little current always flow through your body. I am sensitive to leakages and always ground all electrical equipment. Grounding an electrical equipment keeps you at the same electrical potential as the chassis so you do not receive a shock. Have a safe earthing. Regards

 

[JMFahey]

But, what happens when the amp has a two-prong plug, or is connected to an outlet with no safety ground connection? Especially in the case of a two-tap secondary, I'd now imagine that the floating chassis is participating in the full voltage swing on the secondary.

NO.
Transformer secondary voltages are measured tap to tap, or across winding ends.

IF you touch chassis and nothing else, you are NOT receiving a voltage.
The chassis isn´t "participating* in anything.

IF you touch chassis with one hand and any transformer winding end which has at least one tap grounded you will receive the full voltage available from that winding, even fatal if high enough ... buy why would you do that?

If I were grounded and touched that chassis (or my guitar strings), wouldn't I be in trouble?

NO, you are touching chassis and "nothing else" from that transformer.

And then, I've noticed that, according to my multimeter, the usual tube amp arrangement with a grounded secondary center tap puts the hot wire from the power outlet at 60VAC with respect to the chassis, and the chassis at 60VAC with respect to the outlet neutral.

Not at all.
IF you measure that there is something very wrong.
Please post a diagram showing transformer and mains wiring you have.

Not a cut and pasted drawing saying "I did the same" but what YOU actually have.

I understand there's no circuit and thus no potential for ongoing current flow between the chassis and the outlet neutral, but still, with a peak voltage of ~85V with respect to neutral / ground, shouldn't the chassis be able to deliver a brief shock to anything connected to ground?

Not sure since we don´t actually know what you built.

Please post that diagram.
A picture may add some detail but the main info is a hand drawn diagram with all connections.

 

[PRR]

...I've noticed that, according to my multimeter, the usual tube amp arrangement with a grounded secondary center tap puts the hot wire from the power outlet at 60VAC with respect to the chassis, and the chassis at 60VAC with respect to the outlet neutral....

This can happen if the chassis is NOT grounded.

It is un-safe to play an un-grounded amplifier!!!

However, the theory here is: The transformer has internal capacitances. Probably under 1,000pFd. If there is no other path to ground reference, and if the capacitance on each lead is equal, then yes the transformer body will "float" to halfway between the two live wires. 60V on 120V power.

However, draw the whole circuit, with the meter also. The meter loads the capacitances. But a 10Meg DMM won't load it much.

When you try to get a shock from this 60V floating transformer to a water pipe or a grounded microphone, current is limited by the 1,000pFd leakage capacitance. At 60Hz this is 2.8Meg, at 60V this passes 22uA, which is far below the usual 1mA threshold of feeling.

Many-many 2-pin hi-fi amplifiers work this way.

Stage amps are different. Hi-Fi users do not hold onto the equipment the way a guitarist clutches his strings (which are bonded through the cord to the amplifier). Hi-fi users may be less likely to be very sweaty, or taking heart-stressing drugs. Usually all the gear in a Hi-fi must be bonded together via RCA cords to work, while gear in a no-ground stage is usually not bonded guitar to bass to PA and they float to various potentials. Also stage gear has a tough life and insulation breakdown happens.

 

[sahazel]

Ah, transformer body floating due to internal capacitances starts to make some sense.

Alas, I live in an old building where the only outlets with safety grounds are in the bathroom and behind kitchen appliances. Best I can do is a GFCI outlet.

I'll see if I can make a good representative drawing later this weekend, but I have observed this phenomenon with three of my amps now, so it's not the behavior of one particular circuit.

Maybe the simplest example is a 5f1 champ clone that matches schematics online and which I can easily describe:

Primary side: Hot wire from the wall goes through the fuse, through the switch, to one PT primary tap. Other primary tap connects to the neutral from the wall.

Secondary side:
- 5v heater taps go to the 5y3 heater / cathode.
- 6.3v taps connect the lamp, 6v6, and 12ax7 in parallel. There's no 6.3v CT, so a made a virtual CT with two 100Ω resistors, one from each 6.3v tap, other ends soldered together in a chassis ground lug.
- The 325-0-325 taps each go to their own 5y3 plate. CT is soldered to a chassis ground lug.

Multimeter has common probe clipped to the chassis, red probe clipped to the point where the primary tap meets the switch. Measuring AC voltage I get 60VAC.

 

[PRR]

> I get 60VAC.

OK, with a 10Meg meter.

Now clip-lead a 10k resistor across the meter. This is more-like the rough impedance of your body (<1k to >500k). I hope the voltage comes WAY down. Like less than 1V.

Some Electrician's Meters have in-built or selectable "anti-ghost" mode with a similar body-like loading. If a wire has NO connection at either end, but runs in cable/conduit with other AC wires, a hi-Z meter will "show voltage". Often 1/2 or 1/3 of the local line voltage. But for a reasonable length of cable, a lo-Z meter will show that it is not a "hard voltage" but just leakage "ghost".

 

[JMFahey]

Did you follow the schematic in detail, including the typically .05uF x 600V cap from one mains leg to chassis?

All old amps included that; modern ones do NOT (those are called Death Caps for a reason) and instead use 3 wire mains with good ground, PERIOD

You must upgrade your rehearsal room to real grounded outlets, again PERIOD.

Just for kicks, enjoy this video, it happened in Brazil to a band I know, amps were made by a friend.

He works very well, his amps are properly grounded, etc. but venue wiring had some outlets wired wrong and a few were upside down; to boot lazy "Electricians" sometimes do NOT run the proper dedicated ground wire but connect Ground pin to Neutral at the outlet, thinking "it´s practically the same" ... an accident waiting to happen:

YouTube

Notice sparks flying between guitar strings and microphone holder stand; also that at one point he holds strings with one hand, mic stand with the other and can´t let go.

He survived only because he was a healthy athletic 27 y.o. at the time, plus a Ton of luck, both he fell to the floor, was kicked away from stand, his Father in Law was trained in CPR and by sheer chance an ambulance arrived to take care of somebody else.
Even so he has a massive scar tissue in his heart "as if he had suffered a massive coronary" and he has to live with that for the rest of his life.

 

[sahazel]

I am aware of death caps and did not include one in my 5f1 build. The video reminded me that I used to play with a guitarist who had a ritual of touching his strings to the mic at the start of every show. Not a bad habit to get into, I guess!

I tried clipping a 10k resistor across the meter leads. (And thus, between the hot wire and my chassis.) Meter now reads ~0.5VAC. And my 1/4W resistor is not on fire...

 

[sahazel]

Okay, finally got a chance to make a sketch of the project I was actually working on when I noticed this phenomenon. This is an old Knox Model 602 series-filament widowmaker that I'm trying to make safe & reasonably quiet. I've added an isolation transformer and removed the original floating ground and death cap setup.

I was trying to verify that I had the wiring of the isolation PT correct when I originally discovered that my multimeter reported significant AC voltage between the chassis and both the hot and neutral wires from the wall. Went and checked the 5f1 and another amp and immediately found similar behavior with those. Here I've shown my measurement tonight between the line and the chassis ground: 86VAC.

Does the ghost voltage idea still make sense in this setup? Does the schematic look like the right way to wire an isolation transformer?

 

[PRR]

This is the right way. If you ever move to a better place, or get tired of hum/buzz and pull a 3-wire extension cord from a properly grounded outlet. (Yes, extension cords are a risk, but a hot-chassis amp is not a large load.)

 

[sahazel]

Great, that's exactly what I intend to do. I was honestly worried about making this connection because I understand safety ground is connected to neutral somewhere in the building and I'm reading voltage between chassis and neutral.

I'll run an extension cord from one of the few grounded outlets I have and verify that it doesn't blow the fuse, and the meter reads ~0VAC.

 

[sahazel]

Ran an extension cord from a grounded outlet. With the safety ground connected, chassis to hot wire from the wall measures 120VAC, and chassis to neutral measures 0VAC, on both the 5f1 and the Knox. My lightbulb current limiter remains nice and dark, so I infer that there isn't any significant current flowing between the PT secondaries and ground to establish this reference. So that mystery seems to be resolved. Thanks for your help!

While I have a thread going with the Knox schematic, I have another question: the Knox had some very loud 60hz hum (after replacing filter caps, etc), which I've been working to track down. I initially suspected the filaments, but running them on DC didn't help at all! Eventually I discovered that the problem was at the volume control.

With the amp as pictured in my schematic, hum decreases to nothing as I turn the volume down. As I turn the volume up, I get a very loud 60hz hum, but then it suddenly falls off to a reasonable level as the volume reaches maximum.

Surprisingly, with amp volume at max, the guitar's volume control worked great. I tried replacing the volume pot, but got the same behavior with a new pot. Moving wires and components around inside the amp didn't make any difference, and replacing components along the signal path into the 12au6 didn't help either.

Finally I had the idea to just bypass the input volume pot, and move it to between the 12au6 and 50c5. That setup works great! Hum is at the previous, reasonable "max volume" level when volume is all the way up, but is reduced as I turn the volume down.

My question is, what could be going on here? Why would the volume controls on the guitar and between the 12au6 and 50c5 work fine, while one at the input to the amp produces enormous hum at medium settings?

 

[PRR]

> hum decreases to nothing as I turn the volume down. As I turn the volume up, I get a very loud 60hz hum, but then it suddenly falls off to a reasonable level as the volume reaches maximum.

The wire off the pot wiper goes past some nasty 60Hz. (Is the on/off switch on the back of the pot?) Full-down it shorts to ground. Full-up, it is damped by whatever drives the pot. At half-resistance you have hundreds of k resistance so any external 60Hz comes in easy.

Putting the volume pot in front of the whole amp is dumb. The 1st stage hiss does not go away when turned-down. In guitar amps we like to take some gain before the volume control.

Moving Vol to between 6AU6 and 50C5 moves from a 0.050V level to a 7V level so the same hum is relatively much-much smaller.

Problem here is that the 6AU6 has a *lot* of gain. If you put the pot between 1st and 2nd stages it is possible a hot guitar will overload the 6AU6 before it gets to a point you can control it (Of course you do have a knob on the guitar.) This type of guitar amp is always a compromise between short gain and early overload. Of course the world is full of "better designs" and the reason to use a cheap piece of kit like this may be beCAUSE of its rude tone.

 

[PRR]

> hum decreases to nothing as I turn the volume down. As I turn the volume up, I get a very loud 60hz hum, but then it suddenly falls off to a reasonable level as the volume reaches maximum.

Here's a simulation of this effect, which should be familiar to any guitar-amp builder.

Because of sim stupidity, I have modeled with DC. The "100Meg" impedance represents typical leakage from a hum-source to pot wiring. (Approximately 3pFd at 60Hz.) Turn the pot zero to max. The leakage increases from zero to half-resistance, then falls off at max.

 

[JMFahey]

No need to follow old design to a T if that creates problems and does NOT provide any sonic bliss.
I´d suggest you mod inputs to Fender classic 2 inputs, high and low, which allow you to plug guitar into the attenuated one if too hot.

Then you can remove that cheesy input pot and place it where it belongs, at the 50C5 grid.

Best of both worlds and IF you want to use it as the original , you can simply set amp volume pot to 10 and control everything from the guitar ... *exactly* as before.

 

[sahazel]

Ah, yes, I hadn't noticed when testing with my Tele, but my 335's humbuckers do indeed overdrive the 12au6 a bit unless I turn the guitar down. Not an entirely unpleasant quirk for a cheap and unusual amp.

Looks like Harmony made some amps based on the same tube set (like H-303A) that put the volume on the 50C5 grid and used something somewhat like the Fender input arrangement, as JMFahey suggests. Rather than struggle to reduce hum around the input pot, I think I'll take the Harmony design as inspiration and see if I can find a balance I like. Thanks for the explanations!

 

[jjman]

Alas, I live in an old building where the only outlets with safety grounds are in the bathroom and behind kitchen appliances. Best I can do is a GFCI outlet.switch, to one PT primary tap. Other primary tap connects to the neutral from the wall.

Although most of the outlets may not have a ground pin, the boxes may be metal with metal conduit to each. All that metal may be grounded and ready to be accessed by replacement outlets.

Or cheat:

 

[sahazel]

Imitating Fender inputs was a great idea (actually I went with a 3-input variant from a Harmony schematic, and some 47k resistors I had handy, but same design as 5f1 inputs when the 3rd input isn't in use). Gives me a clean input for humbuckers, grounds the input when I unplug, and the input resistance has stopped the amp picking up radio.

Interestingly, even though the low input puts the guitar through a 47k/47k voltage divider, it doesn't induce hum the way the original input volume pot did. Not sure if that's because it's less total resistance than the pot, or because the input is further from the rest of the circuit, or both.

Anyway, I'm happy to report that the remaining hum is now as quiet as my 5f1 at the same volume level. I'll post a video when I get it buttoned up.

 

[sahazel]

Came up with some further tweaks and questions on this one.

First, I realized that my 50C5 screen voltage was too high at 137V. Data sheets variously give a max of 110 or 120V. I guess the problem was that the output of my isolation transformer on modern power is around 127V, more than the amp was designed for.

To fix it, I first considered a resistor on the AC input to the rectifier, or increasing the 1k dropping resistor producing B++, but I already had most of the voltages in the amp about right (maybe 10% high). Finally I settled on a small 33Ω resistor on the AC input to get most voltages right where I wanted them, and a 10k resistor on the 50C5 screen to get to 110V.

This turned out to be a great chance to learn what happens when screen voltage is insufficiently stiff. It dropped easily with even a little signal, and the result was a slightly distorted, compressed sound. A 10uF cap from screen to ground made the screen voltage stiffer and cleaned up the sound. It now sounds better to my ears than it did originally with excessive screen voltage.

So, old design was rectifier output to a 60uF cap for B+, through a 1k resistor to a 40uF cap for B++. Now I have an additional 10k resistor and 10uF cap for the screen... B+++, I guess. Does this design seem reasonable? Would I be better off sticking with a single dropping resistor and less total capacitance?

Second question: I found someone's instructions online for adding this same isolation transformer to an amp with a very similar circuit. Unlike some other posts I've read, he recommends adding a silicon bridge rectifier before the 35W4 to use both halves of the AC waveform, because of the load put on the transformer by half-wave rectification. Is this really necessary? My math says it's not, but maybe I'm doing something wrong:

Transformer is rated 50VA, 435mA. AC heaters use 150mA and I assume I should take that off the top. The remaining 435-150=285mA, times 0.28 per Hammond recommendations for half-wave rec is 80mA. Adding up the highest quoted max-signal values for plate and screen currents in the 12au6 and 50c5, I get 74mA, which I infer should be about the max DC current draw (and actually I think the 12au6 is running pretty cool vs. datasheet specs). Am I missing something, or just this transformer just barely able to handle this amp?

In practice, the transformer doesn't get hot from use, though maybe lukewarm or maybe that's just the nearby tubes warming it up.

 

[PRR]

....the highest quoted max-signal values for plate and screen currents in the 12au6 .... actually I think the 12au6 is running pretty cool vs. datasheet specs).....

Obviously this AU6 is not running the 10mA-15mA on the datasheet. It has a 470k plate resistor, so even if the tube were flat-short it could only pull 0.3mA; to "work" it must be pulling ~~half of that.

When a voltage-amplifier is *choke* loaded, as in an RF/IF amplifier, best gain comes at HIGH current. The datasheet guys like those numbers.

When a voltage-amplifier is *resistor* loaded, as in most audio amplifiers, best gain comes at LOW current. Limited by the load of the next stage. The fall-off of Gm with current is not as much as the increase of gain with bigger plate resistor.

I would not panic running 50C5 G2 somewhat over the book values. It's a soft limit, and 50C5 are still very cheap. However if 10K gives a good drop, do that. There is some loss of gain and max output with such a large resistor; as you found a capacitor will smooth it out for speech/music and even e-guitar (with some "bounce" on overload transients). If you were trying to build a stadium-blaster, there might be some better optimum; but you didn't pick a 50C5 to blast stadiums or even tiny clubs. It's a big bedroom amp and I doubt a tenth-Watt more or less really matters.

Don't put a bridge rectifier in there!! It can be done but with the confusing "ground" connections in a cheap amp you are very likely to get it wrong and blow it up. Cleanliness suggests losing the bottle rectifier, running both plates and heaters off a FWB... but all the voltages will have to be dropped and the ~~15W of excess heater power means a resistor bigger than the 35C4 bottle. Leave it be. As you figure, the PT is not so close to its ratings that the half-wave underutilization will hurt much. Add filtering if needed for 60Hz hum, e-caps are cheap today. While you might stick to 50-60uFd first cap for rectifier stress, you could then go 150r 10W to a 470uFd second cap to feed OT, that will be very clean.