Anyone know what this noise is in my class D power Amps?

[Electrobumps]

I have hypex class D amps. No issues in the UK. Over in the USA I get this noise. It is worse on power up after the amp warms up it seems to become less frequent. Sometimes it is in the left channel sometimes the right. I don't hear it across both speakers at the same time. I have 2 x stereo amps in the chassis and they both have the issue.

I've recorded the sound with an iPhone.

Thanks
View attachment Amp Noise.mp3

 

[user 130681]

That sounds like interference from a mobile phone to me.

 

[ccaudle]

Sounds like demodulated EMI.
Did you build the amps yourself from Hypex modules, or is it a commercial design?
If you built them yourself, describe the input connections (single ended, balanced, shield connector clamped to chassis, wired to chassis, pin 1 problem with shield wired to circuit ground, etc.)
Is your model one which has LM4562 buffers on the front end? That op-amp has been reported to be EMI sensitive and may benefit from additional filtering on the input.

 

[Electrobumps]

Hi, thanks for the response. These were self build.

Balanced i/p pin 1 to chassis. Also tried lifting i/p ground between pin 1 and chassis and cap between pin 1 and chassis.

They are grounded as Hypex documentation/papers.

They are UCD180HG V2 With HxR powered by SMPS400V13. They are stated to have the LM4562 buffers.

I couldn't find much info about problomatic EMI on these modules when searching. One post did suggest sweeping with a kids walkie talkie. I did do this and when both radios talk buttons are pushed at the same time I pick it up at about 2 meters, I guess this is all relative to the strength of the signal.

 

[ccaudle]

They are stated to have the LM4562 buffers.

How is the connection made from the XLR input connectors to the amp buffers? If you have wires, you may be able to add ferrite beads to the wires and reduce the EMI.

 

[Electrobumps]

I used balance Van Damme XKE very short run. I've not had any experience using ferrite beads should this go as close to the input as possible? I've also seen a build that had one on the AC power in. Would this be worth doing too?

Thanks

 

[Ike Zimbel]

A few thoughts:
1) There's another thread going about cell phone interference getting into audio here.
2) Neutrik EMC connectors on your input cables might help (ie: the XLR's that you are plugging into the amp).
3) I wonder if the RF is actually getting in via the speaker cables? Not really sure what you could do about that, except try ferrite clamps on the speaker cables, right at the output terminals.
4) The red lead on the 2nd XLR (the one on the right in your pic) looks almost like it's broken off. Could just be the camera angle.

 

[ccaudle]

The earlier mention of following the Hypex wiring guide I think likely refers to this:
Hypex amplifier signal wiring page

That page provides bad advice, and conflicts with the information written up by the previous head designer (Bruno Putzeys) in their "pin 1 problem" app note:
Hypex Pin 1 Problems app note

There are a couple of things going on in the way you connected the inputs:

• the pin 1 connection has a long looping connection back to chassis. That will be a high inductance connection, so very high impedance at radio/TV/cell phone frequencies. That basically means it cannot do the job you expect it to. Instead use a piece of copper braid and connect pin 1 directly to the connector shell tab right above it.
• The pin 1 connection still connects to the circuit ground, which means that those noise currents are going to flow across your audio circuit (see the pin 1 problems app note linked above). What you want instead is that the amplifier ground gets referenced to the chassis at some point away from the input connectors (possibly through the power supply), so that currents flowing on the shields can travel whatever path gets them back to the source, be that through another cable shield, or a power line safety earth, or parasitic capacitance to the power supply.

In case you look for some of the references from that Hypex app note, be aware that Rane has reorganized their business, so the links to the Rane app notes are no longer correct in the Hypex paper.
The updated links are here:
Rane app note 110, Sound System Interconnection
Rane app note Grounding and Shielding of Audio Devices
Whitlock Rane tech support story referenced

I've not had any experience using ferrite beads should this go as close to the input as possible

Yes, they just slide over the wires (i.e. the separate hot and cold wires from pins 2 and 3). Longer is usually better, but at some point they become unwieldy and difficult to deal with, as well as making the wires heavier. You can also put multiple in series on a single wire.

I used balance Van Damme XKE very short run

That looks very cumbersome. The wires are inside of a chassis, so do not need a heavy outer insulation to protect them, and the wires are inside the chassis, so do not need an overall shield. Just a twisted pair will work and be much easier to deal with.
I like to take a piece of Ethernet cable and cut off the outer jacket, and use the inner twisted pairs as hookup wire. The wire is low capacitance, easy to get, and inexpensive, and already comes nicely twisted for you.

 

[JohnRoberts]

IIRC Bruno even designed a line of interface cables, have you tried their customer service.

JR

 

[Lampie519]

The mains voltage and frequency is different in the US therefore i guess you use a transformer of some kind.
It could be the source of this issue.

 

[Electrobumps]

The earlier mention of following the Hypex wiring guide I think likely refers to this:
Hypex amplifier signal wiring page

That page provides bad advice, and conflicts with the information written up by the previous head designer (Bruno Putzeys) in their "pin 1 problem" app note:
Hypex Pin 1 Problems app note

There are a couple of things going on in the way you connected the inputs:

• the pin 1 connection has a long looping connection back to chassis. That will be a high inductance connection, so very high impedance at radio/TV/cell phone frequencies. That basically means it cannot do the job you expect it to. Instead use a piece of copper braid and connect pin 1 directly to the connector shell tab right above it.
• The pin 1 connection still connects to the circuit ground, which means that those noise currents are going to flow across your audio circuit (see the pin 1 problems app note linked above). What you want instead is that the amplifier ground gets referenced to the chassis at some point away from the input connectors (possibly through the power supply), so that currents flowing on the shields can travel whatever path gets them back to the source, be that through another cable shield, or a power line safety earth, or parasitic capacitance to the power supply.

In case you look for some of the references from that Hypex app note, be aware that Rane has reorganized their business, so the links to the Rane app notes are no longer correct in the Hypex paper.
The updated links are here:
Rane app note 110, Sound System Interconnection
Rane app note Grounding and Shielding of Audio Devices
Whitlock Rane tech support story referenced


That looks very cumbersome. The wires are inside of a chassis, so do not need a heavy outer insulation to protect them, and the wires are inside the chassis, so do not need an overall shield. Just a twisted pair will work and be much easier to deal with.

I appreciate the feedback on this. I only added shielded cable yesterday as hypex has shielded cable in their diagrams. It's Van Damme XKE so not that difficult to work with.

This is how the other amp in the chassis is wired.




I did have the pin 1 to chassis loop that is now removed. The chassis mount XLR's multi pin 1 to shell on inserting an XLR. So no braid is needed. This amp seems to have no EMI now.

The other amp is still picking up EMI. So I will upgrade to XLR's as the one used are plastic and the shell tab is redundant on the inputs.

Thanks for the advise and links.

 

[Electrobumps]

The mains voltage and frequency is different in the US therefore i guess you use a transformer of some kind.
It could be the source of this issue.

They have a high and low voltage Jumper on the PSU's. I used this in multiple locations in London and never had issues. Hopefully this is now solved!

 

[ccaudle]

chassis mount XLR's multi pin 1 to shell on inserting an XLR

What model connector is that? I do not think I have run across that feature of automatically connecting pin 1 to shell in the connectors I have used.

 

[Electrobumps]

What model connector is that? I do not think I have run across that feature of automatically connecting pin 1 to shell in the connectors I have used.

They don't exist! When the XLR to the balanced jack is plugged into my monitor controller I got continuity between the shell and pin 1.

The continuity was due to the safety ground pin one connection in the monitor controller! Not the XLR input.

Properly connecting pin one and shell reintroduced the EMI problem. However, I am connecting from balanced Jack to XLR and following the ream connection table lifting the ground on the XLR cable solved that issue!

One last thing to figure out is the pop I get when turning it off and on. This doesn't happen after several minutes of use. Weirdly also something that was never an issue in the UK!

What about the ground input that I have wired to pin 1 on the module? Should this be left floating or connected to the chassis?

 

[ccaudle]

Properly connecting pin one and shell reintroduced the EMI problem.

Seems likely it wasn't actually "proper" in that case. You connected pin 1 to the shell with a short connection, and got rid of that connection from pin 1 into your audio circuit ground? Your follow on question leads me to believe that you did not remove the connection to the audio circuit ground.

What about the ground input that I have wired to pin 1 on the module? Should this be left floating or connected to the chassis?

The best place to connect to chassis would be the GND terminal of the power supply. The GND pin on the module input would be left floating, so that the combination of power supply and amplifier module just has the single connection to chassis.

 

[Electrobumps]

The best place to connect to chassis would be the GND terminal of the power supply. The GND pin on the module input would be left floating, so that the combination of power supply and amplifier module just has the single connection to chassis.

The SMPS PSU's ground to chassis via PCB screw hole, amp modules ground via heat sink to chassis. I always thought best practice would be to star ground these. Legacy Pin 1 problems document argues against star ground implementation. Its confusing as on one hand it advocates AES48 which I thought stated the following - The safety ground, chassis ground, and circuit ground is tied at a common star grounding point.

Hpex Amp modules connect to chassis via heatsink and by design is not meant to be isolated.

This Hypex Doc B1) Use balanced (XLR) inputs. This allows the whole thing to be earthed unless the ancillary equip- ment has problems.

So potentially should I also be looking at my monitor controller as creating the issue!

 

[ccaudle]

One last thing to figure out is the pop I get when turning it off and on.

Are you using a delay on the on/off control, or just letting the amp come on right away when the power supply ramps up? The on/off control is supposed to be pop-less, so putting a power on delay on that control to delay amp turn on until the power supply is fully stable would likely solve the pop problem.

The Hypex SMPS1200 power supply has an amp enable line for that purpose. Are you using that supply, with the amp enable connected between supply and amp modules?

Are you just removing AC power with a switch, or using the power supply standby control to turn power on and off (assuming you are using a Hypex supply)?

The SMPS PSU's ground to chassis via PCB screw hole

OK, verify that there is actually a low resistance connection from the PS GND output pin to some other point on the chassis. Ideally you would be able to control the grounding point, but as long as your chassis actually is conductive should be OK. It looks like the PS would screw to the lower panel of the chassis, so pay special attention to checking from the PS GND to the rear panel, since it is easy to get high resistance connections when you have multiple panels connected together.

amp modules ground via heat sink to chassis

I'll take your word on that, it is not mentioned in the datasheet. In cases like that you have to just hope the layout designer did the job properly and doesn't let interference current cross the audio circuitry, or have a heat sink which is isolated from chassis. Those amps don't put out much heat, so it is easier to have an isolated heat sink compared to a traditional design which might need substantial back panel space for heat sink fins.

I always thought best practice would be to star ground these.

It may depend on how you define your star. The goal is that interference currents have a path to travel which does not include the audio circuitry. You need to have the audio circuitry referenced to the chassis at one point, not multiple points for that to work correctly. That is a logical star, but not necessarily a physical star (i.e. a star ground doesn't necessarily involve long wires all going to one point, it could just involve single conductive path to one point, e.g. short pin 1 to chassis connection which then use the chassis as an "arm" of the star).

This is how the referenced AES48 standard draws the recommendation:

So potentially should I also be looking at my monitor controller as creating the issue!

Output connectors can have pin 1 problems as well. Start by getting your amp wiring correct, then evaluate whether any changes are needed to your monitor controller. Same rules apply everywhere, so you can pop open the case and tell in a couple of minutes if it has a potential problem or not.

 

[Electrobumps]

Thanks for the detailed reply.

Are you using a delay on the on/off control, or just letting the amp come on right away when the power supply ramps up? The on/off control is supposed to be pop-less, so putting a power on delay on that control to delay amp turn on until the power supply is fully stable would likely solve the pop problem

The Hypex SMPS1200 power supply has an amp enable line for that purpose. Are you using that supply, with the amp enable connected between supply and amp modules?

Are you just removing AC power with a switch, or using the power supply standby control to turn power on and off (assuming you are using a Hypex supply)?

Switching mains voltage onto SMPS supply.

The amps were not popping on startup in the UK. The amp module's delay start should create a pop-free startup. SMPS sends an on signal to the amp module 1.5s delay.

This pop only happens on one channel. The other channel does not pop. So some further investigation is needed.

OK, verify that there is actually a low resistance connection from the PS GND output pin to some other point on the chassis. Ideally you would be able to control the grounding point, but as long as your chassis actually is conductive should be OK. It looks like the PS would screw to the lower panel of the chassis, so pay special attention to checking from the PS GND to the rear panel, since it is easy to get high resistance connections when you have multiple panels connected together.

This has good continuity.

I'll take your word on that, it is not mentioned in the datasheet. In cases like that you have to just hope the layout designer did the job properly and doesn't let interference current cross the audio circuitry, or have a heat sink which is isolated from chassis. Those amps don't put out much heat, so it is easier to have an isolated heat sink compared to a traditional design which might need substantial back panel space for heat sink fins.

Don't take my word for it, I'm wrong. Just removed a module to verify. I have little understanding of amps and with no schematics.....it makes it hard to know what is going on. This below is mentioned in the Hypex Pin 1 Problems app note

“Your amp module connects the audio ground to the chassis. This will cause hum when connecting. Can you please insulate the heatsink”

I took that to mean that this is what is happening.

Output connectors can have pin 1 problems as well. Start by getting your amp wiring correct, then evaluate whether any changes are needed to your monitor controller. Same rules apply everywhere, so you can pop open the case and tell in a couple of minutes if it has a potential problem or not.

Absolutely agree. It's a PreSonus Central station + I opened it up it's a potential problem.

 

[Electrobumps]

I have now got these fixed. Hypex support were excellent in getting me sorted on this. They asked me to measure for DC on the output. This needs to be as close to 0mv as possible and below 10mv. You can tweak the trimmer on the daughter board to adjust this. Do not touch the other trimmer as this can damage the Amps.

I had 1.5v on startup and this slowly dropped to below 10mv. The suggestion was to change the two 22uf 63v buffer caps. This got rid of the pop on startup and the EMI/noise.

The PCBs are not the best quality and the trace holes lift very easily so any repairs on these need to be done carefully.