The top of the primary should go to XLR pin 2. The bottom of the primary to XLR pin 3. XLR Pin 1 is not ground as many believe, it is shield. You can connect XLR pin 1 to ground via a 51 ohm resistor in series with a 0.01uF and a SPST ground lift switch in parallel with the resistor and cap.
se
Just reading back through the thread this morning (it's easier to follow with a cup of coffee than a glass of wine).
Wouldn't the above not actually be balanced output? My understanding of balanced is that both phases must have an equal impedance return to ground. You already mentioned your feelings on balanced headphones though, so maybe you don't mean to say that this would be balanced.
Wouldn't the above not actually be balanced output? My understanding of balanced is that both phases must have an equal impedance return to ground. You already mentioned your feelings on balanced headphones though, so maybe you don't mean to say that this would be balanced.
It would be balanced as long as the secondary isn't tied to ground on the primary side. Most of the tube circuits I see however seem to have the secondary connected to ground on the primary side. I assume this is done for safety reasons because of the high voltages involved.
But as long as the secondary is galvanically isolated from the primary (which is sort of the whole point of a transformer otherwise you could just use an autoformer) you could tie the bottoms of the two secondaries together and each headphone driver would still be driven from a balanced source.
se
When you're saying secondary tied to ground on the primary side, do you mean the center tap tied to ground on the primary side? I'm not quite following. For differential tube output stages, B+ is often delivered through the center tap of the output transformer primary. I've seen a couple of parafeed arrangements that do it differently, but those are rare for push pull.
I think we're talking about two different things. An output transformer can be used to drive headphones unbalanced (common grounds) from a balanced/differential source. Like a regular push pull amp. I'm trying to come up with something that has balanced output without the common grounds between channels. Now I know the merits of this are dubious, but if I'm going to pretend to believe in it, I might as well go whole hog.
Look at your schematic. The 4 ohm tap on the secondary, which you're using as a center tap, is tied to ground on the primary side.
Yes, I know. But I'm talking about the secondary. You're using the four ohm tap as a center tap and you're tying it to ground on the primary side. Why? Even regular push-pull amps tie the bottom of the secondary to primary side ground. I can only assume this is for safety.
Let's say your transformer is being driven by a single-ended circuit. Now connect the top and bottom of the secondary to a headphone driver. NO CONNECTION TO GROUND ON THE PRIMARY SIDE. In this case, how is the headphone driver NOT being driven from a balanced source?
Further, take two channels of this. Connect the bottoms of the two secondaries together. How is each headphone driver still NOT being driven from a balanced source? The two channels don't see each other and have no effect on each other. You might just as well connect the bottoms of the secondaries to the tip of your nose.
This is how you can drive two headphone drivers from a balanced source using a TRS plug and jack.
But it seems to be common practice for high voltage tube circuits to always tie the secondary to primary side ground for safety reasons as I can think of no other reason for doing so.
se
Thanks for clarifying, Steve! I follow your explanation now.
My understanding is that the secondary should be tied to ground to prevent it from floating. My understanding is also that balanced requires equal impedance to ground for both ends of the signal. Grounding the center tap of the secondary addresses both of these. My understanding could be wrong
This is where I think my understanding may be failing me. The way I'm looking at it: in AC terms, the single ended circuit sees ground through the power supply or the tube even if the primary is not directly grounded. Likewise, the secondary will be seeing this impedance to ground through the OPT. If we connect the 'bottom' of each of two channels, doesn't this now create an imbalanced impedance to ground in relation to the 'top' of each channel? The 'bottoms' would see the impedance to ground in parallel.
My understanding is that the secondary should be tied to ground to prevent it from floating. My understanding is also that balanced requires equal impedance to ground for both ends of the signal. Grounding the center tap of the secondary addresses both of these. My understanding could be wrong
This is where I think my understanding may be failing me. The way I'm looking at it: in AC terms, the single ended circuit sees ground through the power supply or the tube even if the primary is not directly grounded. Likewise, the secondary will be seeing this impedance to ground through the OPT. If we connect the 'bottom' of each of two channels, doesn't this now create an imbalanced impedance to ground in relation to the 'top' of each channel? The 'bottoms' would see the impedance to ground in parallel.
Hello Sodacase,
try to see signal and ground as diffrent things. The output of an transformer is every time a balanced signal. In the moment you take one end of the secondary to ground the signal went to a unbalanced signal. A balanced signal is only to see between the two signal wires. Let the ground out of your mind. A speaker or a headphone only see the difference between the two wires.
try to see signal and ground as diffrent things. The output of an transformer is every time a balanced signal. In the moment you take one end of the secondary to ground the signal went to a unbalanced signal. A balanced signal is only to see between the two signal wires. Let the ground out of your mind. A speaker or a headphone only see the difference between the two wires.
Thank you, Dieter!
So as long as the secondary is ungrounded, it is inherently balanced. I follow that. Maybe I am just preoccupied with the safety precaution of grounding the secondary. SY's earlier suggestion of a high value resistor on each leg of the secondary to ground addresses this. A grounded center tap on the secondary would as well without upsetting the balanced nature, but those can be difficult to find in the right values and can't be relied on to be a perfect CT.
I have to admit that I don't understand exactly what charges the secondary when it is floating and presents the safety hazard in the first place though (maybe magnetic coupling with other parts of the circuit?).
So as long as the secondary is ungrounded, it is inherently balanced. I follow that. Maybe I am just preoccupied with the safety precaution of grounding the secondary. SY's earlier suggestion of a high value resistor on each leg of the secondary to ground addresses this. A grounded center tap on the secondary would as well without upsetting the balanced nature, but those can be difficult to find in the right values and can't be relied on to be a perfect CT.
I have to admit that I don't understand exactly what charges the secondary when it is floating and presents the safety hazard in the first place though (maybe magnetic coupling with other parts of the circuit?).
Thanks for clarifying, Steve! I follow your explanation now.
My understanding is that the secondary should be tied to ground to prevent it from floating. My understanding is also that balanced requires equal impedance to ground for both ends of the signal. Grounding the center tap of the secondary addresses both of these. My understanding could be wrong
"Ground" is commonly misunderstood.
Floating isn't necessarily something that needs to be prevented. It just means there's no physical connection to some other point. And yes, balanced means equal impedances to ground. But that ground can be a floating ground.
Ultimately, the secondary doesn't know ground from well, a hole in the ground.
Whatever is driving the primary, whether it's push-pull or single ended, pushes current through the primary creating a magnetic field which couples to the secondary inducing a voltage across it. Transformers have galvanic isolation between the primary and secondary (which is why they are used to prevent ground loops between equipment) so whatever is used for ground on the primary side is irrelevant to the secondary. It is only responding to a magnetic field.
The secondary will have an electrical center at the center of the winding. It doesn't matter whether you put a physical tap there or not. If you have 100 turns, the electrical center will have 50 turns on one side and 50 turns on the other side and the impedances on either side of this point will be the same, hence balanced.
The voice coil of the headphone driver also has an electrical center with an equal number of turns on each side.
The salient point here is that it doesn't matter if there's an actual physical connection between the electrical center of the secondary and the electrical center of the headphone driver's voice coil. If you were able to stick a probe at each of these points, there would be no potential difference between them when carrying a signal. So having an actual physical connection between them is moot.
And getting back to what I said previously, if you have two channels, and you connect the bottoms of the two secondaries together, neither channel sees the other and they behave exactly as they would if you did not tie them together, hence, you could use a TRS plug/jack and each driver would still be driven from a balanced source. The electrical centers of the secondaries and the voice coil do not change nor do the impedances between those electrical centers and each side of those centers.
se
I assume the safety issue concerns a possible insulation failure between the primary and secondary windings. Remember, your B+ is tied to the primary's center tap so if a primary winding shorted to a secondary winding, then you've got 200 volts on your headphone jack.
se
Ahh, I think I see the light now. Thank you Steve. I was thinking too much in DC terms maybe.
Regarding the TRS example, you mentioned single-ended earlier. It seems that the same would still be true of differential amplification, wouldn't it? The drivers will only see the difference between the two secondary outputs.
One of the cited benefits of balanced is lack of crosstalk. I wonder how that comes into play.
Yes, that's what I was getting at. Whether the primary is driven by a single-ended circuit, a push-pull circuit or a balanced differential circuit, if the headphones are connected to the secondary, they are being driven balanced.
That really hasn't anything to do with balanced as it does with the TRS jack which has a common ground contact resistance. Each channel's signal current flows through that resistance and there will be a tiny voltage drop across that resistance.
Personally I think it's insignificant. It would have to be pretty bad before you'd ever hear it while listening to music.
Besides, many people want their headphones to sound like loudspeakers instead of the middle of your head effect you get with headphones and what's the crosstalk of a pair of loudspeakers?
Also, if you're using open back headphones, you're going to get quite a lot of acoustic crosstalk not to mention the capacitive and inductive coupling from the cable.
You could simply use a four pin XLR with an unbalanced amp and get rid of it if it really kept you up at night. Really this is all just fodder for clinically neurotic audiophiles.
se
It's crazy.
The first "balanced" headphone amp (and nearly all that have followed) was just two amplifier channels bridged together. So the "balanced" input wasn't even differential and had zero common-mode noise rejection which is the whole raison d'etre of balanced interfaces. Any common-mode noise would simply be amplified and passed on to the output. As I said, the rest of the audio world would call this a bridged amplifier but it was called a "balanced" amplifier and the insanity ensued.
Among the "benefits" was "four times the power." Four times the power relative to what though? Of course it was four times the power compared to a single channel of the channels that were being bridged. But of course this was meaningless as you could get the same amount of power from a single-ended amp.
Another was "twice the slew rate!" Which again is meaningless unless you're slew limiting and I'm not aware of any headphone amplifiers that are slew limiting unless someone is making a headphone amp using 741 opamps or something.
The promotion of "balanced" was nothing but a load of meaningless hype, but people ate it up.
I suspect some of that had to do with an old audio salesman trick. We humans tend to perceive louder as not simply louder, but as higher quality. Unscrupulous audio salesmen took advantage of that and made sure that the system they wanted to sell was always played just a little louder than the system's being compared to.
If you have a bridged amplifier that also has a single-ended output, the bridged output will be a bit louder than the single-ended output so people would perceive the bridged output as sounding better.
se
Ok, updated schematic below.
I really appreciate the help and guidance so far! We found a more affordable output transformer option (which happens to have a CT on the secondary). Also added a u-pad style shunt attenuator.
The grid leak resistors and shunt attenuator are a confusing network. I think the resulting Zout value would be the grid leaks in parallel with the value of the shunting pot. It should be good enough to not roll off the treble too badly with the ~100pf input capacitance of the ECC99.
Input impedance would be the grid leaks in parallel with the shunt plus the 12.5ks, I think?
I don't much like how both will be affected by rotation of the volume pot, but I'm not sure how else to get around it. I suppose it's no different than other designs though. We want to do this with a single stereo pot to control volume on both channels.
I really appreciate the help and guidance so far! We found a more affordable output transformer option (which happens to have a CT on the secondary). Also added a u-pad style shunt attenuator.
The grid leak resistors and shunt attenuator are a confusing network. I think the resulting Zout value would be the grid leaks in parallel with the value of the shunting pot. It should be good enough to not roll off the treble too badly with the ~100pf input capacitance of the ECC99.
Input impedance would be the grid leaks in parallel with the shunt plus the 12.5ks, I think?
I don't much like how both will be affected by rotation of the volume pot, but I'm not sure how else to get around it. I suppose it's no different than other designs though. We want to do this with a single stereo pot to control volume on both channels.
indeed waste of time, unless your amp is BTL or something special.
( Different story; electrostats are symmetrical by design, with push-pull stators. )
BTW remove those 62k, just makes signal weaker.. input sec has ct..... put gnd there
input opt, primary gnd you may experiment, and if connecting to usb dac like me, you do not want to wire gnd. (hum)
( the xlr system does not recommend wire cable shield to pin1 (you should not wire ct to gnd, since signal is +- symmetric), and if accidentally used with cinch; you may short one half of primary.)
you may think about it like 300R tv line
Last edited:
That's a great idea. Thank you! Grid will still see DC ground through the input secondary that way and I don't have to deal with the elaborate resistor network. That would make Zin about 1/2 the shunt plus the 12.5k resistors. Zout would be the shunt value.
You're also recommending to remove the input primary center tap ground? It could be put on a switch to lift it if needed, I guess?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.