Power Supply Noise - Is this always a good solution?

[Labaris]

Hi,

I read on ESP that this components help to prevent noise on power supplies:



GND is the negative of DC supply for pedals
EARTH is the metal chasis of the PS enclosure (connected in one of the lugs of the transformer)
D1 and D2 are meant to stop fault currents
C1 is supossed to bypass radio frequencies
R1 is supossed to isolate the ground connection to prevent any loops

If this works, why use different transformers to isolate power supply outputs if you can do it only with a resistor? I built a completely floating-ground PS and it makes a huge noise so I want to understand what are the risks of building floating PS compared to building them the way I post here.

Thanks a lot.

[R.G.]

The short answer is that there are a large number of ways that hum and noise can get from power supplies and grounding into the audio path.

The circuit you show is an improved version of a ground lift, which helps with only a limited number of the ways that you can get hum. It is not a guaranteed cure for all cases.

And that's why we do all that other stuff as well.

[R.G.]

If this works, why use different transformers to isolate power supply outputs if you can do it only with a resistor? I built a completely floating-ground PS and it makes a huge noise so I want to understand what are the risks of building floating PS compared to building them the way I post here.

Sorry - I hit send too fast. I only answered the first part of your question.

The second part - the risks - need to be addressed. A completely floating power supply has some electrocution dangers unless it is done in certain very specific ways, which are themselves complicated to explain.

Tying the signal/circuit ground to earth/safety ground directly is the approved way that the safety-standards want it done. Interposing a resistor, cap and diodes between them as in the "ground lift" circuit is questionable in the view of the safety inspectors. It may or may not pass an inspection to safety standards.

So the earth ground connection is required in many instances for electrical safety. But if each piece of equipment has its own earth/safety ground connection by its power cord, you can get hum as all "earth ground" connections are not at the same electrical potential. When you connect the signal ground between them, you get a hum loop.

There are ways to work with this. Disconnecting signal ground from earth ground on all but one piece of equipment is one way. This "ground lift" is what was done in early studios. But this creates electrical hazards. The resistor/cap/diodes circuit helps with the safety a little bit, and may be enough, but is not guaranteed to be safe.

The real answer is to not connect signal ground between two pieces of equipment. Make the signal balanced +/- around its ground, and only connect its ground to its local shield on the sending end. Then at the receiving end, use a balanced differential receiver to convert back to a single ended signal. And this is indeed how most modern studios are designed. It's the reason for 600 ohm balanced line signals.

We cannot do this for guitar equipment for historical reasons, so we will always be fighting hum.

[Labaris]

The short answer is that there are a large number of ways that hum and noise can get from power supplies and grounding into the audio path.

The circuit you show is an improved version of a ground lift, which helps with only a limited number of the ways that you can get hum. It is not a guaranteed cure for all cases.

And that's why we do all that other stuff as well.

What other stuff?

The second part - the risks - need to be addressed. A completely floating power supply has some electrocution dangers unless it is done in certain very specific ways, which are themselves complicated to explain.

So, the Spyder for example, has those electrocution risks under control?

Tying the signal/circuit ground to earth/safety ground directly is the approved way that the safety-standards want it done. Interposing a resistor, cap and diodes between them as in the "ground lift" circuit is questionable in the view of the safety inspectors. It may or may not pass an inspection to safety standards.

So the earth ground connection is required in many instances for electrical safety. But if each piece of equipment has its own earth/safety ground connection by its power cord, you can get hum as all "earth ground" connections are not at the same electrical potential. When you connect the signal ground between them, you get a hum loop.

There are ways to work with this. Disconnecting signal ground from earth ground on all but one piece of equipment is one way. This "ground lift" is what was done in early studios. But this creates electrical hazards. The resistor/cap/diodes circuit helps with the safety a little bit, and may be enough, but is not guaranteed to be safe.

Very clear.

The real answer is to not connect signal ground between two pieces of equipment. Make the signal balanced +/- around its ground, and only connect its ground to its local shield on the sending end. Then at the receiving end, use a balanced differential receiver to convert back to a single ended signal. And this is indeed how most modern studios are designed. It's the reason for 600 ohm balanced line signals.

I've always wonder where those 600 Ohm came from.

We cannot do this for guitar equipment for historical reasons, so we will always be fighting hum.

Ok, understood. So if use this technique (the improved ground-lift) to get rid of HUM in a multiple isolated outputs PS, will I be losing all the benefits of having build it isolated? Can an isolated-outputs PS have HUM in some cases but not always?

Thanks.

[tubegeek]

I've always wonder where those 600 Ohm came from.

600 ohms is historical: it's the impedance used by early telephone equipment.
Same answer for 19" rack spacing: phone company.

[Labaris]

I've always wonder where those 600 Ohm came from.

600 ohms is historical: it's the impedance used by early telephone equipment.
Same answer for 19" rack spacing: phone company.

Thanks for the info

[R.G.]

What other stuff?

Many things. A whole thread in itself.

So, the Spyder for example, has those electrocution risks under control?

The Spyder is by no means complete as shown, especially for AC mains risks.

In particular, for electrical safety, all metal surfaces which the user can touch must either have continuity to AC mains safety ground, or be isolated from the mains by double or reinforced insulation. All of these issues are safety-standards items, and require instruction and practice. The Spyder shows a way to isolate only certain secondary outputs, with the idea that the signal grounds they connect to will be properly grounded for safety reasons. Isolating the grounds on a 9V pedal supply (for instance) allows the grounding point to be set by the amplifier, thereby reducing ground hum from multiple sources of safety ground. This is one of the things I alluded to earlier, and a different way of breaking ground loops than using balanced signals.

I've always wonder where those 600 Ohm came from

As the earlier poster noted, it's from telephone company practice. 600 ohms is about the resistance of some distance (1mile? 2 miles?) of single twisted pair used for telephone communications. We got the "phone jack" from the telephone companies as well.

Ok, understood. So if use this technique (the improved ground-lift) to get rid of HUM in a multiple isolated outputs PS, will I be losing all the benefits of having build it isolated?

All the benefits? No. But you must understand that the whole idea of an isolate power supply is to let the signal ground of the rest of the chain of pedals and amps provide the grounding point instead of imposing its own idea of what "ground" is.

As an aside, if you are using a really, correctly isolated "9V" power supply to power a pedal and you still have bad hum, either the power supply is not correctly isolated the way you think it is, or there are hum issues in the rest of the signal path outside the power supply.

Can an isolated-outputs PS have HUM in some cases but not always?

I would state that idea differently. An isolated outputs power supply will not contribute hum that is not already there. It is not a magic amulet that prevents hum from other causes.

[PRR]

> We got the "phone jack" from the telephone companies as well.

Good point. Someone was asking about "patching". When Ma Bell had to patch, manually, thousands of calls an hour, they developed a 1/4" plug and plug-racks, the manual switchboard. The guitar plug is a very cheap almost-compatible knock-off of the Bell patch plug. If you are doing intense patching, this IS the way to go. (But our needs are seldom so intense. There is a Bantam plug a bit smaller yet adequate for studio patching. And RCA still serves in minimum-cost patching.)

> 600 ohms is about the resistance of some distance (1mile? 2 miles?) of single twisted pair used for telephone communications.
> 600 ohms is historical: it's the impedance used by early telephone equipment.

There's a vast array of cable types. None are 600 ohms per mile, not even as an average. And are you saying series resistance? Audio impedance?
FWIW, the fat-copper WIDE-pair (open pair on glass insulators) does run 400-1000 Ohms "characteristic impedance" (infinite length). Above some frequency which sadly is usually in the telephone band. Open-pair is so expensive that after very little thought AT&T went to twisted-pair cable, which (like all twist-pair) tends to 100 ohms characteristic impedance. The series resistance can be a few ohms or many hundred ohms (even a couple thousand ohms).
Early telephony was trial and guess. Carbon mikes tend to 100 ohms, receivers similar. This is a good "match", which is necessary when you have NO amplification (except the amplifying action of a carbon mike).
This is fine for a short line. But when you want to expand your service, you soon have hundreds of ohms in the line. You reduce this with fatter wires but that gets VERY expensive. To promote Universal Service, AT&T went to quite thin wires (in cable). For other reasons there was a transformer in the instrument. A mild step-up gives better volume, a large step-up gives really bad treble. Ratios like 1:2 and 1:3 were used. There's no real "standard impedance. Late in the analog cable days there was a 900 ohm reference, but that seems to be a measurement yardstick not a goal.
No, 600 ohms came from meters and specifically Radio Networks. In the 1930s and 1940s, AT&T made the bulk of their money not from residential, business, or long-line switched telephony, but from the radio networks. (This trickles-down to AT&T's non-invention of tape recording.) Into 1937, the most common passive meter was rated 500 ohms at a VERY high level and was much too sluggish to indicate speech/music peaks. AT&T's internal metering systems were very expensive and complicated. The radio networks were having problems setting levels across many-hop continental feeds. AT&T listens to its best customers. Simpson had built sensitive light-meters and was brought in.
The networks were of two schools of thought about impedance. Some low (like 200) some high (like 500). Most interconnects were through transformers, often hybrid-coils, which have two windings on each side. They can be connected two ways, 4:1 impedance. It turned out that Simpson could not approach all goals with anything under 600 ohms nominal impedance (rectifier drop is a big factor). The committee settled on nominal 600 ohms, with 150 ohms as an option. (I think it was ABC who stuck with 150 ohms well into the 1970s and perhaps to the end of nominally "matched" lines.) May 1939. ASA C16.5-1961
Here we are metering "short" lines. Specifically the _point_ where AT&T connected to the broadcaster. For convenience, this might be a run down the block then a run upstairs, but always far under a mile, and usually with "negligible" loss.
This is, of course, the VU meter. Sensitive enough to show speech/music peaks, without loading a live line or needing an amplifier, cheap enough to have one at all key points in the network (and ultimately everywhere).
600 Ohms was sealed when H & P calibrated the 200AB for 600 Ohm output standard.
So The Telephone Company was at the table, but 600 Ohms was not their idea, and it wasn't "early days".
Interestingly the Talking Movie racket didn't have much to do with 600 Ohms until it became common in radio networks. They did take up 600 Ohms in a big way later.
We all know "600 Ohms" because much recording gear was based on radio gear, which was pretty good and much less expensive than the costly confections the movie industry had grown into by the late 1930s.

[tubegeek]

No, 600 ohms came from meters and specifically Radio Networks.

Well I'll be damned. Thanks Paul!

[R.G.]

Hmmmm. I was told it was miles of wire.

Ah, well. Another fine legend bites the dust.

[jblack547]

This is my favorite topic of the audio transmission realm. 600ohm balanced vs 10k single ended. When I finally understood it, it was an enlightening moment.

When you guys say "hum" do you mean the 60hz from the power line? That note that is note quite Bb?

[Labaris]

Thanks PRR and everyone for the comments about the 600 Ohm myth

Now, going back to my PS, here's a (bad) picture of it. This is the noisy one:



I've done some tests and with an amp from a sound reinforcement company in a live situation. Result: HUM in your face, but just as the same as the original noise I found.
Now I connected the same pedalboard to a different amp at some friend's home. There's a lot less noise or not at all. At least it's different.

I got another PS I built years ago and it's very quiet. It has the resistor and cap between negative and chasis ground (no diodes) and all the jacks are connected with separated *twisted cables (not the right word but I think you get the idea). It's the typical common ground PS, but it's been always very quiet.

So, the differences between noisy/quiet PS are:

1. Improved GND LIFT/Improved GND LIFT with no diodes
2. One bare tinned cable (see photo)/Twisted pair cables

And my questions are:

1. If I want to make the GND LIFT option switchable with a SPDT ¿What would be a good GROUND option? I guess it could be direct connection between negative and chasis ground.
2. Is this bare tinned cable a huge mistake and possible cause for the HUM?

Thanks

[R.G.]

I've done some tests and with an amp from a sound reinforcement company in a live situation. Result: HUM in your face, but just as the same as the original noise I found.
Now I connected the same pedalboard to a different amp at some friend's home. There's a lot less noise or not at all. At least it's different.

I got another PS I built years ago and it's very quiet. It has the resistor and cap between negative and chasis ground (no diodes) and all the jacks are connected with separated *twisted cables (not the right word but I think you get the idea). It's the typical common ground PS, but it's been always very quiet.

So, the differences between noisy/quiet PS are:

1. Improved GND LIFT/Improved GND LIFT with no diodes
2. One bare tinned cable (see photo)/Twisted pair cables

Actually, there are a whole lot of differences, not the least of them being a different power transformer and different wiring - right?
There are many sources of hum. As you noted, one of them is which amp the power supply ground gets connected to.

2. Is this bare tinned cable a huge mistake and possible cause for the HUM?

No.

[Labaris]

I've done some tests and with an amp from a sound reinforcement company in a live situation. Result: HUM in your face, but just as the same as the original noise I found.
Now I connected the same pedalboard to a different amp at some friend's home. There's a lot less noise or not at all. At least it's different.

I got another PS I built years ago and it's very quiet. It has the resistor and cap between negative and chasis ground (no diodes) and all the jacks are connected with separated *twisted cables (not the right word but I think you get the idea). It's the typical common ground PS, but it's been always very quiet.

So, the differences between noisy/quiet PS are:

1. Improved GND LIFT/Improved GND LIFT with no diodes
2. One bare tinned cable (see photo)/Twisted pair cables

Actually, there are a whole lot of differences, not the least of them being a different power transformer and different wiring - right?
There are many sources of hum. As you noted, one of them is which amp the power supply ground gets connected to.

2. Is this bare tinned cable a huge mistake and possible cause for the HUM?

No.

So, you mean that every build has its own risks of HUM?
What's your advice for the ground selector switch?

[R.G.]

So, you mean that every build has its own risks of HUM?

Yes. Very much so. Hum can come from many places, as I've noted. It can even come from an unfortunate arrangement of wires near a transformer, poor choice of which wire goes from the transformer/rectifiers and to/from first filter cap in the circuit.

What's your advice for the ground selector switch?

If you want to try it both ways, put in a switch to "short" around the whole mess.

Be aware that you're playing with AC mains safety issues here. If you don't know what's right, the internet is not a good place to get info to bet your life on.

[PRR]

> I was told it was miles of wire.

There was the "TU", Transmission Unit, which was nominally the _loss_ of a specified length of a specified line with a specified load. Different actual cables would have different TU per mile. But the nice thing is that TU would just +add+ : if the cable in hand is 1.2TU per mile, then 5 miles is 6 TU loss. Repeaters (amplifiers) were also rated in TU. So a long string of lines and repeaters could be planned without multiply/divide and without much decimal-point confusion.

You of course recognize that this is the same concept as the Bel and deciBel. And in fact the TU worked out to like 1.07 dB (or the practical unit of Bels was selected to be similar to the familiar units of TU), so many of the old papers can be read without conversion.

[jblack547]

The supply in the photo looks like a linear supply. Most likely based on an LM317 on the heat sink. The transformer says 220v. Some of these transformers have parallel and series primaries and allow you to wire it for either 120 or 220. What is yours wired for? If you are wired for 220 and you are using 120 the output is cut in half and you won't get much regulation, if at all.

Check the voltage after the diode bridge. Right at the input to the regulator itself. It should be at least 11 to 12 volts for a 9v output. Check the output voltage and make sure it is adjusted to 9volts. Looks like a trimmer on the board. Try that. Does the hum get worse as the system stays on longer? Another problem could be that you are drawing over the rated current. Stay below the rated output.

[Labaris]

Yes. Very much so. Hum can come from many places, as I've noted. It can even come from an unfortunate arrangement of wires near a transformer, poor choice of which wire goes from the transformer/rectifiers and to/from first filter cap in the circuit.

There might be some tricky thing happening, I don't know.

If you want to try it both ways, put in a switch to "short" around the whole mess.

Ok, thanks.

Be aware that you're playing with AC mains safety issues here. If you don't know what's right, the internet is not a good place to get info to bet your life on.

Thanks for the advice.

The supply in the photo looks like a linear supply. Most likely based on an LM317 on the heat sink. The transformer says 220v. Some of these transformers have parallel and series primaries and allow you to wire it for either 120 or 220. What is yours wired for? If you are wired for 220 and you are using 120 the output is cut in half and you won't get much regulation, if at all.

We have 220V in Chile, where I live.

Check the voltage after the diode bridge. Right at the input to the regulator itself. It should be at least 11 to 12 volts for a 9v output. Check the output voltage and make sure it is adjusted to 9volts. Looks like a trimmer on the board. Try that. Does the hum get worse as the system stays on longer? Another problem could be that you are drawing over the rated current. Stay below the rated output.

The output voltage is OK, I've not measure the voltage after the bridge.


Current situation:
Since I have the other power supply, which has been always quiet, I've been modifying the noisy one to be just like the other. I guess the problem could be not having enough difference between output and input voltage at the regulator (it's actually less than 5V). Yesterday it got very much silent when the only digital pedal on the board was unplugged from the PS. Today it was noisy as usual but got silent again when the cabsim pedal was unplugged. I really think there some ripple issue also.

[jblack547]

If the hum is 50hz, it is coupled. If the hum is 100hz, it is due to poor regulation. 50hz is straight from the mains, 100hz is after the bridge rectifier.

[Labaris]

If the hum is 50hz, it is coupled. If the hum is 100hz, it is due to poor regulation. 50hz is straight from the mains, 100hz is after the bridge rectifier.

And poor regulation could be a consequence of having the LM317 working with a low input/output voltage difference?