What's wrong with this balanced audio generator output?

[Greybeard]

Hi,
this audio generator has a balanced XLR output.



When I set 1 Vrms output voltage, I can measure:
+signal to signal GND : 506 mVrms
-signal to signal GND : 506 mVrms
+signal to -signal : 1012 mVrms

So far so good.

But when I switch the output GND switch (ground lift) from "GND" to "floating", I measure this:
+signal to signal GND : 428 mVrms
-signal to signal GND : 584 mVrms
+signal to -signal : 1012 mVrms
The + and -signals are not symmetrical to signal GND!

I measured the resistance from GND pin of the XLR output to chassis GND:
3.6 Ohms (independent of GND switch position).
There is no "floating GND" (AC coupled or open GND)!

So what may be wrong with it?


This is just a random find of "ground lift" google search, I don't think the generator uses a real output transformer:







 

[CaptDon]

Not only is it balanced above ground, but also floating. Any sort of reading is possible
and you can probably shift the reading significantly just by placing the resistance of
your finger from one of the leads to ground. I used to pull my hair out trying to explain
'open delta' three phase wiring to my industrial electrician co-workers. It was 480 phase
to phase but anybody's guess what you would see phase to ground since it wasn't
ground referenced. Theory said I could have placed a hand on ground and grabbed
any phase (expecting 277vac) but with no current path to ground on the open delta
I would continue living. I was never crazy enough to try to prove the theory although
we did use a low wattage 277 volt lamp for leakage testing. A neon tester would always
glow as expected!!!

[TimFox]

In general, a floating output (such as a transformer secondary without a grounded center tap) only becomes “balanced” when connected to a balanced input (equal impedance to ground at both ends).  It is common to have a ground connection at only one end of the cable.

[DimitriP]

Have you looked at the output with a 'scope ?
I'm curious if the sinewave is symmetrical in all cases.

[Greybeard]

Have you looked at the output with a 'scope ?
I'm curious if the sinewave is symmetrical in all cases.

Both signals are perfectly sinusoidal, just different amplitudes.

[Greybeard]

Rout is specified < 15\$\Omega\$.
V+signal - V-signal falls to 50% when I connect a load of 26\$\Omega\$, so it's 13\$\Omega\$ per "half" output transformer winding.

Yes, I can "pull" the voltage down relative to GND, using a resistor about some k\$\Omega\$.
If I use 1.3 k\$\Omega\$ it's half the voltage as unloaded, so Rout to GND is  1.3 k\$\Omega\$.

[Cerebus]

The initial assumption that pin 1 on an XLR has anything whatsoever to do with the signal is wrong. Pin 1 can be the cable screen, possibly chassis ground, and possibly the return for phantom power on microphone circuits. What it is not is the reference level for the hot and cold signals. It might be and frequently will be, by coincidence, at roughly the same potential as the reference for hot and cold, but it is not and should not be used as such. The signal is what appears differentially between hot and cold, not between hot and anywhere else, or cold and anywhere else.

The whole point of having floating, differential signals on professional audio equipment is to divorce them from the local grounds of the equipment at both ends of the link.

A note on grounding centre taps: It's a very bad idea. If you have phantom power applied to a transformer wired this way you will short the phantom power to ground via the transformer coils. On a well designed input this will result in the phantom power going into current limiting, on a bad design it will break. In either case, you will be pushing DC through the transformer, which won't improve its performance. Centre taps should only be used for phantom power injection or extraction.

[Greybeard]

What's the correct way to wire the balanced output, if I need an unbalanced (asymmetrical) signal for testing (guitar) amps?
In most cases, the amp's input GND is connected to mains safety GND

[Cerebus]

With a well designed output stage you can just "unbalance" it by connecting the cold from the output to the signal ground reference at the input.

[Greybeard]

With a well designed output stage you can just "unbalance" it by connecting the cold from the output to the signal ground reference at the input.

That means:
Pin 2, +signal ---------- amp input hot
Pin 3, - signal ---------- amp input cold (internally connected to mains safety GND)
Pin 1, signal GND: unused
GND switch: "floating"

Correct?

[Cerebus]

Yes, but:


+signal ---------- amp input hot
- signal ---------- amp input cold (internally connected to mains safety GND)
signal GND cable shield: unused shield/chassis

You want the shield to be connected to a nice low impedance (i.e. chassis) both for safety reasons and because you want a nice low impedance sink for any signals induced in the shield. If you do encounter a ground loop, then and only then might you consider floating the shield at one end only. If you can solve a ground loop problem by using a floating or pseudo-floating connection always do that in preference to 'lifting' any grounds. 'Lifting' a ground is a last resort, and should only be done when there is still a safety ground in place at every piece of equipment involved.

Anyone who 'lifts' grounds by removing the safety ground of equipment should be promptly taken outside, beaten flat with the blunt end of a drummer to prevent inadvertent re-use, and then discarded as "unsafe for use".

[Greybeard]

Lifting GND was meant as lifting signal GND, NOT lifting safety GND !!!

[Cerebus]

Grounding shield on both ends will short - signal (pin 3):

No, why would you think so?

Can I remind you that illustration is just a random one you found somewhere and is a pretty good illustration of "how not to do it".

[Greybeard]

Grounding shield on both ends will short - signal (pin 3):

No, why would you think so?

Can I remind you that illustration is just a random one you found somewhere and is a pretty good illustration of "how not to do it".

I referred to the 2nd circuit:



If GND switch is in not set to "floating" position, -signal (pin 3) would be shorted at amp's cold input pin.
The circuit is weird  (I never would do it this way), but it nearly reflects the behavoiur of the real generator.

[Greybeard]

Can I remind you that illustration is just a random one you found somewhere and is a pretty good illustration of "how not to do it".

Then you can give me an illustration "how to do GND lifting", please?

[Cerebus]

I referred to the 2nd circuit, because it reflects the behavior of the generator:

I hope that circuit does not reflect the behaviour of the generator because, as I have said, grounding the centre tap of an audio mic/line transformer is not the thing to do.

Taking a "random", to use your own description, diagram and then assuming it applies to your equipment is a very odd thing to do.

Do you have an actual circuit for the actual output of the generator? You'd be better served by looking at that rather than random illustrations that are (1) about something else, (2) actively unhelpful. What make and model is your generator?

[TimFox]

You didn’t post the model number of your generator.
I own a Wavetek 98 DDS-synthesized sine wave generator.  It’s unbalanced 50 ohm output (BNC) works from 1 uHz to 1.1 MHz.  The 600 ohm output is through a transformer, which limits it to 30 Hz to 200 kHz (the digital control throws an error if you request a frequency outside that range for the 600 ohm output).  The transformer secondary connects to two banana jacks (canonical 3/4 inch spacing) with the CT brought out to a binding post (ungrounded).  The voltage measurements on the 600 ohm output are consistent with what you measured.

[Greybeard]

It's an A2, I don't have schematics and I did not find any.
I just can measure how it behaves and think about which circuit could fit.

[Audiorepair]

This is getting way more complicated than it needs to be.

Make a very short female XLR to mono jack cable, wire the hot to the tip of the jack and the cold and the shield to the ground.
Then run a standard XLR cable from your generator and plug your adaptor into it and into the guitar amp.
Then you have a balanced XLR cable you can test anything else with, and an adaptor for amps etc.

If you ground lift your generator, the cable will still be shielded by the amp, so shouldn't be noisy, but the amp will see no ground from the generator.
This will then mimic a guitar, which also has no ground of its own.

[Cerebus]

Perhaps you should take the time and trouble to find the manual and read it.

Took me five minutes to find, download and extract these two diagrams:





There is a limit to how much people will work to help you if you don't even seem to be willing to do your own homework first.

Edit: Found just underneath the second diagram:

[Greybeard]

Thank you, Cerebus!

If this circuit is correct, then my device is defective or wired wrong.

1) Pin 1 is always connected over 3.6\$\Omega\$ to chassis and protective GND. Float/GND switch position doesn't matter. I have measured this, as I wrote. I also measured the switch is connected to chassis GND on one side and on the other side it's not connected to pin 1, it is connected to "center point".

2) Both output voltages are not symmetrical (428mV/584mV), if GND switch is set to "floating", so I thought both centering resistors are not exactly equal. I have checked these by measuring from pin 2/3 to GND switch, but both are equal (7.8\$\Omega\$).

3) Maybe there is a bypassing network over the GND/Float switch, because "center point" is not fully free floating.

[Greybeard]

Make a very short female XLR to mono jack cable, wire the hot to the tip of the jack and the cold and the shield to the ground.
Then run a standard XLR cable from your generator and plug your adaptor into it and into the guitar amp.
Then you have a balanced XLR cable you can test anything else with, and an adaptor for amps etc.

If you ground lift your generator, the cable will still be shielded by the amp, so shouldn't be noisy, but the amp will see no ground from the generator.
This will then mimic a guitar, which also has no ground of its own.

That is exactly what I suggested a few posts before:

Pin 2, +signal ---------- amp input hot  (= cable center)
Pin 3, - signal ---------- amp input cold (internally connected to mains safety GND)  (= cable shield)
Pin 1, signal GND: unused
GND switch: "floating"

[Bassman59]

Anyone who 'lifts' grounds by removing the safety ground of equipment should be promptly taken outside, beaten flat with the blunt end of a drummer to prevent inadvertent re-use, and then discarded as "unsafe for use".

In my experience, every end of a drummer is blunt. (And some of my best friends are drummers.)

Back when I was working house sound at the local venue, I would have guitarists come in with cheater/ground lift "death cubes" on their amps' power cords. I would take them off. "Why did you do that?" "Because I really don't want to clean bodies of dead guitarists off of my stage." "But I get shocks from my amp without it." "Fix your fucking amp."

In most cases the problem was that the amps were the Fender style with that silly "ground" switch which just selected which leg of the AC mains got connected to that big death cap to chassis. The only fix is to eliminate that switch and that cap and install a proper 3-wire mains cord with the ground going to the chassis. "... but my amp is vintage!" "doesn't matter if you're DEAD." My '62 Deluxe, '62 Bandmaster, '64 Bandmaster and '66 Bassman amps are all modified with proper grounding and they are quiet and I don't get shocks.

(also when touring clubs I always brought a DMM and tested outlets.)

[Audiorepair]

That is exactly what I suggested a few posts before:

Pin 2, +signal ---------- amp input hot  (= cable center)
Pin 3, - signal ---------- amp input cold (internally connected to mains safety GND)  (= cable shield)
Pin 1, signal GND: unused
GND switch: "floating"

Well, kind of, but you have pin 1 unconnected, so if you extend that cable with another XLR cable, that cable won't be screened by the amp and will hum like hell.

It is better to make a short adaptor as I suggest, and then you can use any XLR cables you happen to have around.

[Greybeard]

That is exactly what I suggested a few posts before:

Pin 2, +signal ---------- amp input hot  (= cable center)
Pin 3, - signal ---------- amp input cold (internally connected to mains safety GND)  (= cable shield)
Pin 1, signal GND: unused
GND switch: "floating"

Well, kind of, but you have pin 1 unconnected, so if you extend that cable with another XLR cable, that cable won't be screened by the amp and will hum like hell.

It is better to make a short adaptor as I suggest, and then you can use any XLR cables you happen to have around.

OK, I get that point!