So I've build my very first power amplifier using an LM3886 IC and I'm about as proud as a teenager can be, considering I've spend only £30 and am now capable of pi**ing off my neighbors haha
I'm pretty inept when it comes to circuitry, as I've never had any experience with something like this. The most I've ever done is repair a few PCB's with broken solder or blown caps. So please excuse my ignorance with this kinda stuff.
The issue I'm having is that I get an overwhelming hum when the amp is on. It's so severe that I can barely hear the music over it.
I've attached the schematic I followed so that you can see exactly what I've built. (To clarify, I have not done anything with the circuitry on the left of the amp diagram. I have no idea what that's for...)
At first I had the SGND and GND separate, so I tried to connect the two thinking ground is ground. Unfortunately, I no longer get any output with the two connected.
I know I'm doing something wrong here so any suggestions on how to resolve this would be appreciated!
I'm pretty inept when it comes to circuitry, as I've never had any experience with something like this. The most I've ever done is repair a few PCB's with broken solder or blown caps. So please excuse my ignorance with this kinda stuff.
The issue I'm having is that I get an overwhelming hum when the amp is on. It's so severe that I can barely hear the music over it.
I've attached the schematic I followed so that you can see exactly what I've built. (To clarify, I have not done anything with the circuitry on the left of the amp diagram. I have no idea what that's for...)
At first I had the SGND and GND separate, so I tried to connect the two thinking ground is ground. Unfortunately, I no longer get any output with the two connected.
I know I'm doing something wrong here so any suggestions on how to resolve this would be appreciated!
Attachments
Here are a few images of the build. A mess, I know 
The only cable I had available at the time has a brown sheathing. So don't be alarmed that the brown is seemingly connected to everything haha I will rectify this in the future
The only cable I had available at the time has a brown sheathing. So don't be alarmed that the brown is seemingly connected to everything haha I will rectify this in the future
Attachments
Last edited:
You can start by trimming the leads in all those components and resoldering. They're unnecessarily long. Pairs or triplets of relevant wires (AC, DC + and return, signal + and return, etc) used all throughout the build should be tightly twisted, too.
BTW I don't see thermal paste out the sides of the 3886s (no matter how much you use there's always some extra that comes out due to mounting pressure), you should add some to improve heat transfer to the heatsink.
EDIT: I don't see the mounting screw being used with a nylon washer, to me it looks like the 3886s are shorted to the heatsink with the mounting screw. Another huge problem there...
BTW I don't see thermal paste out the sides of the 3886s (no matter how much you use there's always some extra that comes out due to mounting pressure), you should add some to improve heat transfer to the heatsink.
EDIT: I don't see the mounting screw being used with a nylon washer, to me it looks like the 3886s are shorted to the heatsink with the mounting screw. Another huge problem there...
Last edited:
I've only just built the amp and I am trying to resolve the hum issue before I go about applying thermal paste and cleaning up the wires etc.
At the moment the amp is never on for longer than a minute for testing, so there is no heat issue.
I may decide to reconfigure the position of the components when I start to think about the enclosure. So nothing is permanent just yet
At the moment the amp is never on for longer than a minute for testing, so there is no heat issue.
I may decide to reconfigure the position of the components when I start to think about the enclosure. So nothing is permanent just yet
I mentioned the long component leads and the untidy wiring because these two can be (especially the wiring) a potential source of hum. Fix one at a time and eventually you'll end up with a hum free amplifier, given the correct wiring. You're using metal tab 3886s, those have the tab connected to V-, and you are shorting them through the heatsink. That could do things much worse to the chips than just make them output hum.
SGND is the 3886's gnd pin, it controls the mute function. If this plus pin 8 aren't set up correctly you'll end up with a muted amplifier, it explains why you didn't get any sound doing this connection. Check the datasheet to understand how it works. Give this a look too, lots of practical information here. We all should buy tomchr a few beers for that.
SGND is the 3886's gnd pin, it controls the mute function. If this plus pin 8 aren't set up correctly you'll end up with a muted amplifier, it explains why you didn't get any sound doing this connection. Check the datasheet to understand how it works. Give this a look too, lots of practical information here. We all should buy tomchr a few beers for that.
both the chipamp backplates are connected to -ve supply pin.
The heatsink just becomes a tapping on the -ve supply rail.
No harm done till you drop a screwdriver.
Dismantle one channel and lay it aside till you have a working mono channel.
Rebuild the mono channel sticking strictly to using twisted wire pairs for EVERY connection you make. The input is two wires. Make them a twisted pair. The output is two wires, make them a twisted pair.
The power supply is an exception. It is three wires. Make that a twisted triplet.
The transformer, and smoothing caps are three connections.
Yes, the Input Signal Return Wire needs to see the same reference voltage as the Speaker Signal Return wire. Take both to the Main audio Ground.
The heatsink just becomes a tapping on the -ve supply rail.
No harm done till you drop a screwdriver.
Dismantle one channel and lay it aside till you have a working mono channel.
Rebuild the mono channel sticking strictly to using twisted wire pairs for EVERY connection you make. The input is two wires. Make them a twisted pair. The output is two wires, make them a twisted pair.
The power supply is an exception. It is three wires. Make that a twisted triplet.
The transformer, and smoothing caps are three connections.
Yes, the Input Signal Return Wire needs to see the same reference voltage as the Speaker Signal Return wire. Take both to the Main audio Ground.
Last edited:
There are some pretty nice point to point builds to refer to and draw
inspiration and ideas from.
Here are a couple from a quick google search -
Mick Feuerbacher Audio Projects
http://www.diyaudio.com/forums/chip-amps/187527-point-point-lm3886.html
inspiration and ideas from.
Here are a couple from a quick google search -
Mick Feuerbacher Audio Projects
http://www.diyaudio.com/forums/chip-amps/187527-point-point-lm3886.html
So, I've removed the screw from one of the IC's and instead held it down with a plastic clamp for testing purposes. However, the hum is still there...
There is also a poping sound from the speakers when I turn off the amp (which didn't occur before)
Could there be an issue with the capacitors? Or have a perhaps damaged the IC's after shorting them out?
There is also a poping sound from the speakers when I turn off the amp (which didn't occur before)
Could there be an issue with the capacitors? Or have a perhaps damaged the IC's after shorting them out?
The hum is caused by the bad layout. If it looks bad it will sound bad.
Don't bend or twist the leads where they enter the IC or they will break.
Another example of P2P
http://www.diyaudio.com/forums/blogs/wintermute/134-construction-p2p-lm3886-gainclone-part-1.html
Don't bend or twist the leads where they enter the IC or they will break.
Another example of P2P
http://www.diyaudio.com/forums/blogs/wintermute/134-construction-p2p-lm3886-gainclone-part-1.html
By "layout" he means how the components are placed together. A good layout has short connections and few loops. Any loops are arranged to have a small loop area.
Some of the reasons for this is that short connections and small loop areas have lower inductance. Hence, those connections will experience lower inductive coupling to things like power transformers and stray fields.
That said, I don't think the layout is the problem in your case. You either have a hook-up error somewhere or the issues is caused by the VEE supply shorting to GND via the heat sink due to the lack of shoulder washers on the mounting screws.
You need one of these puppies on each LM3886: 7721-7PPSG Aavid Thermalloy | HS418-ND | DigiKey
You'll also need a thermal pad or washer between the LM3886 and the heat sink. It looks like you have one there already, but I can't see if it's a mica washer or just a piece of tape.
You can avoid all these issues entirely by using the isolated version of the LM3886. That's the LM3886TF
You'll need to add the decoupling caps as close to the IC pins as possible. I recommend a 1 uF, 50 V X7R ceramic cap soldered directly to the IC pins. Add a 22 uF OSCON (or electrolytic) in parallel with as short leads as possible. Then route off to 1000 uF of bulk capacitance where the power enters the circuit. You can read more here: Taming the LM3886 - Supply Decoupling.
Tom
Some of the reasons for this is that short connections and small loop areas have lower inductance. Hence, those connections will experience lower inductive coupling to things like power transformers and stray fields.
That said, I don't think the layout is the problem in your case. You either have a hook-up error somewhere or the issues is caused by the VEE supply shorting to GND via the heat sink due to the lack of shoulder washers on the mounting screws.
You need one of these puppies on each LM3886: 7721-7PPSG Aavid Thermalloy | HS418-ND | DigiKey
You'll also need a thermal pad or washer between the LM3886 and the heat sink. It looks like you have one there already, but I can't see if it's a mica washer or just a piece of tape.
You can avoid all these issues entirely by using the isolated version of the LM3886. That's the LM3886TF
You'll need to add the decoupling caps as close to the IC pins as possible. I recommend a 1 uF, 50 V X7R ceramic cap soldered directly to the IC pins. Add a 22 uF OSCON (or electrolytic) in parallel with as short leads as possible. Then route off to 1000 uF of bulk capacitance where the power enters the circuit. You can read more here: Taming the LM3886 - Supply Decoupling.
Tom
Last edited:
At the moment, I only have a single LM3886 screwed into the heatsink, the other is held by a plastic clamp. And there are washers on both (taken from the same amp as the heatsink). However the problem still persists.
I'm certain that I have followed the schematic correctly and have no made any errors. Despite this, I will attempt a tear down and start over since this seems like ym, only option atm.
Could someone tell me if the schematic is any good, or if there is a better one out there to follow? If I'm starting over, I might as well copy the 'best' (if there even is one) gainclone circuit
My edits to the schematic posted in Post #1:
R2 = 20 kΩ
R3 = 0 Ω
R6 = 0 Ω
C3 = 47 pF
C4 = DNP (Do Not Populate)
C10 = DNP
Add Cc = 180 pF from pin 9 to pin 10 (the two input pins) on the LM3886. This ensures stability near the supply rails.
Add Cz = 100 nF in series with Rz = 2.7 Ω, 2 W from pin 3 (OUT) of the LM3886 to GND. In series with the output connection (so from SPK to the speaker (+) terminal) add Lt = 2.0 uH (15 turns of AWG 18 / 1.0 mm diameter on an AA battery) in parallel with Rt = 1.5 Ω, 2 W. Rz, Cz form the Zobel network and Lt, Rt form the Thiele network. These networks are needed for stability, in particular with capacitive loads.
Keep SGND and GND as separate nodes and join them at one point only using R6 = 0 Ω (aka a piece of wire). This point should be at the speaker (-) output terminal for the best performance.
Keep the connections as short as you can.
Tom
R2 = 20 kΩ
R3 = 0 Ω
R6 = 0 Ω
C3 = 47 pF
C4 = DNP (Do Not Populate)
C10 = DNP
Add Cc = 180 pF from pin 9 to pin 10 (the two input pins) on the LM3886. This ensures stability near the supply rails.
Add Cz = 100 nF in series with Rz = 2.7 Ω, 2 W from pin 3 (OUT) of the LM3886 to GND. In series with the output connection (so from SPK to the speaker (+) terminal) add Lt = 2.0 uH (15 turns of AWG 18 / 1.0 mm diameter on an AA battery) in parallel with Rt = 1.5 Ω, 2 W. Rz, Cz form the Zobel network and Lt, Rt form the Thiele network. These networks are needed for stability, in particular with capacitive loads.
Keep SGND and GND as separate nodes and join them at one point only using R6 = 0 Ω (aka a piece of wire). This point should be at the speaker (-) output terminal for the best performance.
Keep the connections as short as you can.
Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.