LM3886 hum

morky

Member
2016-02-16 1:19 pm
I know there are many similar threads but please bear with me.
I built my first amp using two LM3886t and after putting it all together I can hear a hum from the speakers. Unfortunately its quite noticeable otherwise I wouldn't mind. You can hear the hum even without inputs and the volume all the way down. I shorted the inputs and the hum is still there. My guess is it comes from the transformer, since it's not a toroidal and I got it really cheap cause I was trying to keep the initial budget low. It's a 2x24V 3A transformer.
What can I do about the hum that won't break the bank?

I provided pics of the schematic and of my setup.

LM3886 amp - Album on Imgur
 

wintermute

Administrator
2003-08-03 11:43 am
Sydney
Looks like you are using 1000uF per rail on the Power supply. I originally had this on my LM3886 and did notice some hum on the output. I added an additional 4700uF per channel and this reduced the hum level significantly (but didn't completely eliminate it). I checked my thread but I never posted any objective measurements of the difference in hum level (pretty sure I did check it on the scope at the time, but I do not remember the magnitude of the drop).

Tony.
 
I will do better than guess.
Your error is in the way you have "grounded" all the various wires/references/returns that all do different jobs.

Every signal, whether it's from a walkman, or to a speaker, is a TWO WIRE connection.

Forget about grounds.
You MUST use TWO wires for every signal connection. Signal Flow and Signal Return MUST be kept together for the whole route from source to receiver.
SORT that first.

eg. look at J1.
It shows a signal flow wire as a single wire from signal socket to amplifier.
It shows the signal return (RCA barrel) as a single wire connection to some "ground".
That is WRONG !!!!!!!!

The problem you have is that >99% of schematic diagrams are drawn that way. It takes some experience to recognise the ERROR and think about the CORRECT way that it needs to be wired.
 
Last edited:

morky

Member
2016-02-16 1:19 pm
I will do better than guess.
Your error is in the way you have "grounded" all the various wires/references/returns that all do different jobs.

Every signal, whether it's from a walkman, or to a speaker, is a TWO WIRE connection.

Forget about grounds.
You MUST use TWO wires for every signal connection. Signal Flow and Signal Return MUST be kept together for the whole route from source to receiver.
SORT that first.

eg. look at J1.
It shows a signal flow wire as a single wire from signal socket to amplifier.
It shows the signal return (RCA barrel) as a single wire connection to some "ground".
That is WRONG !!!!!!!!

The problem you have is that >99% of schematic diagrams are drawn that way. It takes some experience to recognise the ERROR and think about the CORRECT way that it needs to be wired.

I'm using two wires. those black wires have two wires in them and are shielded. I just don't have a better picture to show you right now since I'm at work. I'll get a better one when I get home. Also I already tried shorting the inputs and the hum is still there.
 
Last edited:
You need to improve safety.

Uninsulated mains connections close to where you are working with tools and fingers is not good for continued posting at this Forum.

I suggest you dismantle the whole thing and place the transformer and ALL the mains connections at the far end of the bench where you can't accidentally touch them/it.
Then run a long 3wire isolated Low Voltage AC to the amplifier PCB. This places you and your hands in an isolated environment where it is relative safe to experiment.
Without a chassis, you can't have any inadvertant connections that could be creating hum.
Get this mono version working first.
 

morky

Member
2016-02-16 1:19 pm
You need to improve safety.

Uninsulated mains connections close to where you are working with tools and fingers is not good for continued posting at this Forum.

I suggest you dismantle the whole thing and place the transformer and ALL the mains connections at the far end of the bench where you can't accidentally touch them/it.
Then run a long 3wire isolated Low Voltage AC to the amplifier PCB. This places you and your hands in an isolated environment where it is relative safe to experiment.
Without a chassis, you can't have any inadvertant connections that could be creating hum.
Get this mono version working first.

That's what I was thinking of doing. getting a second separate box and placing the transformer with the capacitors and bridge in, separate from the PCB and its chassis.


As for the safety comments, I totally agree with you and I'm being really careful.
 
I'm not suggesting a new permanent location for the transformer.

It's getting the amplifier out of the chassis, that I am more concerned about.
The chassis can complicate fault finding.
No chassis equals simpler layout and connections. Equals simpler fault finding.

Since the PSU is built into the amp PCB, I think you don't have the option to move the transformer to a remote permanent location. The worst non audio current pulses are between the transformer and the smoothing capacitors via the bridge rectifier. It is my view that this group of three MUST be kept together and made very compact with LOW LOOP AREAS.
 

morky

Member
2016-02-16 1:19 pm
I'm not suggesting a new permanent location for the transformer.

It's getting the amplifier out of the chassis, that I am more concerned about.
The chassis can complicate fault finding.
No chassis equals simpler layout and connections. Equals simpler fault finding.

Since the PSU is built into the amp PCB, I think you don't have the option to move the transformer to a remote permanent location. The worst non audio current pulses are between the transformer and the smoothing capacitors via the bridge rectifier. It is my view that this group of three MUST be kept together and made very compact with LOW LOOP AREAS.

Ok then, I'll try removing the transformer from the chassis at first and see if the hum is still there. If I were to remove the transformer permanently I would remove the bridge and the capacitors along with it to a separate chassis. I wouldn't keep the capacitors and bridge on the PCB.


I'll post back here again when I remove the transformer after work today.
 

morky

Member
2016-02-16 1:19 pm
Ok, so I placed the transformer out of the chassis about a meter and a half or so and the hum is still there... I don't know how to proceed next.

I'm thinking of buying another chassis and placing the transformer, the bridge and two extra capacitors in there and do something like a Faraday cage. That way I'll have only DC going inside the box with the PCB.
 

wintermute

Administrator
2003-08-03 11:43 am
Sydney
If the hum is the same with the transformer 1.5M away then putting it in a farady cage is not going to help. The problem is not radiation of noise from the transformer itself.

One thing very odd I noticed about the schematic is the 10 ohm resistors in series with the output... Not sure if they would cause any problems with hum but certainly they would have a huge effect on power delivery and the Q of the speakers the amp is connected to. I've never seen that before and it does not look right. also your circuit board appears to have output inductors, whereas I don't see any on the schematic.....

Tony.
 

whizgeek

Member
2010-02-16 11:47 pm
Check the wiring on transformer secondary. it looks like a half bridge wiring into full bridge rectifier.
Pull out each fuse on main pcb and measure the current and voltage,for both neg and pos supply. They should be similar, and in line with the datasheet. How much voltage drop when the fuses are inserted ?
Check the chips are not electrically connected to the heatsinks.
Otherwise:
The bridge rectifier is too small for good use, and you don't have enough space in the box to fit sufficient capacitors.
 

morky

Member
2016-02-16 1:19 pm
If the hum is the same with the transformer 1.5M away then putting it in a farady cage is not going to help. The problem is not radiation of noise from the transformer itself.

One thing very odd I noticed about the schematic is the 10 ohm resistors in series with the output... Not sure if they would cause any problems with hum but certainly they would have a huge effect on power delivery and the Q of the speakers the amp is connected to. I've never seen that before and it does not look right. also your circuit board appears to have output inductors, whereas I don't see any on the schematic.....

Tony.

I don't know about that resistor and the inductors tbh but I found this online (http://www.diyaudio.com/forums/solid-state/125679-function-output-inductor.html
And also if I'm not mistaken the inductors you mentioned are on wrapped around those two resistors.



Check the wiring on transformer secondary. it looks like a half bridge wiring into full bridge rectifier.
Pull out each fuse on main pcb and measure the current and voltage,for both neg and pos supply. They should be similar, and in line with the datasheet. How much voltage drop when the fuses are inserted ?
Check the chips are not electrically connected to the heatsinks.
Otherwise:
The bridge rectifier is too small for good use, and you don't have enough space in the box to fit sufficient capacitors.


What do you mean when you are saying "a half bridge wiring"? If i remember correctly the Voltage was about 33 Volts on both fuses but I had measured this with the fuses on. Also the chips are not electrically connected to the heatsinks (I already made that mistake and ended up burning one of the two LM3886).

The way the transformer is connected is this:

Red wire from plug to transformer, Black wire from plug to the fuse.
Red wire from the fuse to the switch, Black wire from the transformer to the switch.
And then the three wires (2 red, 1 black) from output of the transformer to the PCB.

I also have soldered the led to one of the capacitors (those next to the fuses) under the PCB. I did it because I didn't know at the time that the PCB already had an output for the LED. That's why you see two wires going under the PCB connected to the LED. But I don't think this has anything to do with the humming.

I also took a few more pictures last night which you can find in the link bellow.

https://goo.gl/photos/z4YdSWA8DydJ7Yi59
 
Ok, so I placed the transformer out of the chassis about a meter and a half or so and the hum is still there... I don't know how to proceed next.

I'm thinking of buying another chassis and placing the transformer, the bridge and two extra capacitors in there and do something like a Faraday cage. That way I'll have only DC going inside the box with the PCB.
That is what I expected. I want you to remove the amplifer from the chassis. The chassis can complicate the wiring and interferes with access to nodes that need to be measured and/or altered.
Get rid of the chassis while you investigate the wiring error/s.

Please attach your pics.
 

morky

Member
2016-02-16 1:19 pm
That is what I expected. I want you to remove the amplifer from the chassis. The chassis can complicate the wiring and interferes with access to nodes that need to be measured and/or altered.
Get rid of the chassis while you investigate the wiring error/s.

Please attach your pics.

Ok, so I can remove the PCB from the chassis after I get back from work no problem there, but can you explain how the chassis complicates the wiring? What should I look for after removing the PCB? What nodes are you talking about?

Also while removing the PCB from the chassis should I also remove the RCA plugs etc.? Or just put longer wires and pass the input/output signal from the plus on the chassis?

more pictures here: https://goo.gl/photos/z4YdSWA8DydJ7Yi59
 
Get rid of the chassis.
The metal chassis introduces the chance, whether deliberate or accidental, of making a connection that is not required.
AND, assemble a mono first. Let's get the mono working with low measured hum i.e. output Noise+Hum with input shorted, should read 0.0mVac, or at most 0.1mVac. You will need a DMM set to 199.9mVac to check this.

Please attach your pics.
 
Last edited:

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
Please attach your pics.

+1 for attachments rather than links to external sites.

I looked at the schematic you posted and am not seeing anything that would prevent the circuit from working. Yeah, the supply reservoir caps (2200 + 1000 uF) may be on the low side, but should be fine. You should certainly not have hum at idle. The 10 Ω resistor that's drawn in series with the output appears to be the Thiele network and have an inductor in parallel in reality. Good!

You're building the circuit on a PCB. Did you buy this from somewhere or did you lay it out yourself?

My best guess for why you're having trouble is that you have a ground loop somewhere. Probably on the input. From your images, it does not appear that your RCA connectors are isolated from the chassis. They need to be. You can buy RCA connectors with a plastic collar for this exact purpose.
You can find this out by getting the board out of the chassis as Andrew suggests. Or you can short the input at the input connector on the board rather than at the RCA connector.

You need to use the proper TO-220 mounting hardware for attaching the ICs to the heat sink. In addition to the thermal pad between the IC and the heat sink, you need to use a shoulder washer on the mounting screw. Here's an example: Digikey P/N: 36-3049-ND. I'm surprised you haven't blown a fuse yet with the attachment you have.

Tom
 

morky

Member
2016-02-16 1:19 pm
+1 for attachments rather than links to external sites.

I looked at the schematic you posted and am not seeing anything that would prevent the circuit from working. Yeah, the supply reservoir caps (2200 + 1000 uF) may be on the low side, but should be fine. You should certainly not have hum at idle. The 10 Ω resistor that's drawn in series with the output appears to be the Thiele network and have an inductor in parallel in reality. Good!

You're building the circuit on a PCB. Did you buy this from somewhere or did you lay it out yourself?

My best guess for why you're having trouble is that you have a ground loop somewhere. Probably on the input. From your images, it does not appear that your RCA connectors are isolated from the chassis. They need to be. You can buy RCA connectors with a plastic collar for this exact purpose.
You can find this out by getting the board out of the chassis as Andrew suggests. Or you can short the input at the input connector on the board rather than at the RCA connector.

You need to use the proper TO-220 mounting hardware for attaching the ICs to the heat sink. In addition to the thermal pad between the IC and the heat sink, you need to use a shoulder washer on the mounting screw. Here's an example: Digikey P/N: 36-3049-ND. I'm surprised you haven't blown a fuse yet with the attachment you have.

Tom

Ok so I don't know how to attach pictures that's why I link other websites. Sorry about that.


The PCB was given to me at a lab in my college about 5 years ago from my teacher. We were building an amplifier in that lab but I never finished it, I just soldered the components to the PCB and just now I started working on it again and placed everything in a chassis.

My RCA connectors are isolated with a plastic collar as you mentioned. Also I have already shorted the inputs on the PCB and not on the RCA connectors and the hum is still there. That's why I was thinking it was the transformer at the first place.


As for the chips they are isolated from the heatsinks. I did that mistake at first cause I hadn't noticed that the chips had a V- on the back and I ended up burning one. After I replaced it I isolated the chips from the heatsinks. I didn't have two TO-220 mountings (just one that wasn't even enough for one chip) but I had some other mica laying around and isolated both, that's why it doesn't look that good.

I'm still not entirely convinced the hum is not from the transformer since the bridge is still on the PCB and I would like to remove it and place it closer to the transformer. But I don't want to do that until I have eliminated all other possible reasons. I didn't have time to remove the PCB today but I hope I'll have time to do it sometime tomorrow. Other than that does anyone else have an idea of what could be wrong? Things to check?

Also thanks to everybody that has tried helping me so far!


PS: Something weird that I noticed on the PCB is under the R22 resistor there is a path with an empty connector that nothing is soldered to it. and also it doesn't seem to need to be actually connected to anything. And the same goes for two other spots under each chip. You can see that on the photos.

Also could there be something wrong with the relay that I removed and shorted?
 
Last edited: