Basically having a look around here makes me want to build a fully tested design for an amplifier so I don't have to go fiddling with it to get it to work. This would be my first time actually building an amplifier from scratch of this magnetude. I would also like to be able to upgrade the chassis to 5.1 in the future (so basically another stereo amplifier and a single me thinks). I was thinking of something quite simple although I want a healthy 100Wrms output per channel. I can't remember the name of them but I saw some ampliiers on here that were quite small that uses the LM3886 I think. Would this be suitible?
Thanks
Boscoe
Thanks
Boscoe
Not a simple question to answer without asking a few questions first.
1. What do you expect out of your amplifier ?
There are literally hundreds of Chip amplifiers and discrete component amplifiers that will provide 100W of reasonable audio quality sound that can be made with only a modicom of electronic knowledge.
2. How much do you want to spend ?
A chip amplifier can be built fairly economically whereas a true Hi-Fi amplifier will cost appreciably more.
3. What speakers and other ancilliaries are you using ?
A pair of B&W DM805d's will require substantially better amplification than a cheap pair of "run-of-the-mill" speakers.
Also consider the final design as part of your initial design.
A six channel 100W/Channel amplifier will require a power supply capable of delivering approximately 750W of power. The output devices will also need to be suitably heatsinked to operate at this power level. Better amplifiers use individual power supplies for each channel, this is costly and takes up a considerable amount of real estate in the design.
The LM3886 is only really good for about 38Watts, but it does this admirably well.
Amplifier power isn't everything. 100W is only twice as loud as 10W. Some of the really high end Class A amplifiers only operate around the 20-30W mark and are perfectly loud enough to annoy the deafest of neighbours.
Don't be too obsessed with published distortion figures. 0.0000000000001% of distortion is all very well. Even some of the £50K + speakers produce 1% of their own.
Personally, I would experiment with a GainClone LM3886 design in your case. It's a good beginning for a novice and they're pretty bomb proof. Take care with the size of the heatsinks otherwise you will cook the amplifier and don't skimp on your power supply.
1. What do you expect out of your amplifier ?
There are literally hundreds of Chip amplifiers and discrete component amplifiers that will provide 100W of reasonable audio quality sound that can be made with only a modicom of electronic knowledge.
2. How much do you want to spend ?
A chip amplifier can be built fairly economically whereas a true Hi-Fi amplifier will cost appreciably more.
3. What speakers and other ancilliaries are you using ?
A pair of B&W DM805d's will require substantially better amplification than a cheap pair of "run-of-the-mill" speakers.
Also consider the final design as part of your initial design.
A six channel 100W/Channel amplifier will require a power supply capable of delivering approximately 750W of power. The output devices will also need to be suitably heatsinked to operate at this power level. Better amplifiers use individual power supplies for each channel, this is costly and takes up a considerable amount of real estate in the design.
The LM3886 is only really good for about 38Watts, but it does this admirably well.
Amplifier power isn't everything. 100W is only twice as loud as 10W. Some of the really high end Class A amplifiers only operate around the 20-30W mark and are perfectly loud enough to annoy the deafest of neighbours.
Don't be too obsessed with published distortion figures. 0.0000000000001% of distortion is all very well. Even some of the £50K + speakers produce 1% of their own.
Personally, I would experiment with a GainClone LM3886 design in your case. It's a good beginning for a novice and they're pretty bomb proof. Take care with the size of the heatsinks otherwise you will cook the amplifier and don't skimp on your power supply.
I wan tto spend around £100 to start off with then add in over channels at a later date, I want 5 channels then an unpowered but amplified sub output.
the speakers I will be using are some homemade ones that I'm in the process of finishing.
Are your figures in rms or music power?
Thanks I did see that it does seem very simple but I would like a bit more power.
Thanks
Boscoe
the speakers I will be using are some homemade ones that I'm in the process of finishing.
Are your figures in rms or music power?
Thanks I did see that it does seem very simple but I would like a bit more power.
Thanks
Boscoe
You could build the SymAsym. It's a very good sounding amplifier and the AAK version gives you 100W at 8 Ohms.
He is talking about rms. Anybody talking in music power need to be hanged or at the very least a umbrella opened up in his @$$.
Technically for a true 100W x 5 amplifier you will need more around the 1kW off powersupply. But in real life situations you will never have a situation where all 5 channels will draw full power ,so i agree that 750W powersupply should be enough.
I do disagree about lm3886 though.
When I feed a lm3886 amp into a 8ohm dummy load and give it a 1khz input wave I can set input so that i get 50W output ,without sign of clipping (on oscilloscope) I left it running like that for at least 5 minutes constantly without any issues. using a 1.2C/W heatsink.
if you require a more severe test of the heatsinks ability to dissipate then reduce the sinewave output to ~ 1/3 maximum power i.e. ~20W to 30W for a 68W amplifier.
If you look at the National graphs you will see why.
Hi,
for a first build and to gain experience in all aspects of solid state design and assembly, I would recommend a 3886.
Do read the National datasheet thoroughly and do come back and ask if there are any (many) bits you don't understand. Make informed decisions, don't guess.
National describe some components as optional. Leave that choice to the experts. Use all the optional components to help ensure a successful build.
Do not use a DC coupled design.
Use an AC coupled design that has a DC blocking cap in the input and in the Negative FeedBack loop (NFB).
Start with a non-inverting design, not inverting.
Choose a PCB that has all these recommended features.
I'd suggest a monoblock without a chassis for first test and listening. You must be careful with all those mains components exposed to touching - no pets, no kids and an informed partner/visitor.
Build up a Mains Bulb Tester (dim bulb tester). It and much more of these details are in the various guides in Decibel Dungeon's site. He does show an inverted amp, don't choose that as your first build.
Safe Building.
for a first build and to gain experience in all aspects of solid state design and assembly, I would recommend a 3886.
Do read the National datasheet thoroughly and do come back and ask if there are any (many) bits you don't understand. Make informed decisions, don't guess.
National describe some components as optional. Leave that choice to the experts. Use all the optional components to help ensure a successful build.
Do not use a DC coupled design.
Use an AC coupled design that has a DC blocking cap in the input and in the Negative FeedBack loop (NFB).
Start with a non-inverting design, not inverting.
Choose a PCB that has all these recommended features.
I'd suggest a monoblock without a chassis for first test and listening. You must be careful with all those mains components exposed to touching - no pets, no kids and an informed partner/visitor.
Build up a Mains Bulb Tester (dim bulb tester). It and much more of these details are in the various guides in Decibel Dungeon's site. He does show an inverted amp, don't choose that as your first build.
Safe Building.
Last edited:
Yes at about 1/3w output power produces more heat. I was not worried about heat.
I was more worried about my power maximum power testing at that stage.
I was more worried about my power maximum power testing at that stage.
To a complete novice I have fond one supplier of an LM3886 kit.
DIY LM3886 Chip Amplifier (Gainclone) Project
However, they are also relatively easy to get on E-Bay.
The Datasheet is probably a bit daunting for a newcomer to DIY design, the link above shows a complete design.
As Andrew T saya, "breadboarding" as we call it, is fun, but all the dangerous voltages can be exposed and can kill if not cause substantial damage. All mains voltages to the transformer or transformers MUST be INSULATED and FUSED. Toroidal transformers, the transformers of most peoples choice, do have quite a high inrush current, so. A nominal 160VA transformer would be reasonably expected to have a steady state mains current of 1.4A at 115V or 0.7A at 240V. You will need anti-surge fuses nominally rated at 2 to 3 times this figure (ie 3A at 115V or 2A at 240V). Truly massive 1-2000VA transformers can have even higher inrush currents and some other form of limiting the current is often necessary. We'll stick with one amplifier (MONO) for the time being.
As with all GainClone amplifiers. Keep all leads to the IC as short as possible and keep the input leads away from the output leads - if you don't you may end up with a high power oscillator.
One word on Power Supplies.
Let us assume that the LM3886 datasheet is accurate and not misleading.
68W cont. avg. output power into 4Ω at VCC = ±28V
38W cont. avg. output power into 8Ω at VCC = ±28V
50W cont. avg. output power into 8Ω at VCC = ±35V
Consider these figures as ABSOLUTE MAXIMUMs.
For safety let us assume that your are using a good pair of B&W DM602's obtained second hand for about £100 - Excellent staring point for speakers. They are a nominal 8 Ohms but dip to about 5 Ohms at certain points in the audio spectrum.
For the safety of your LM3886 we will assume that the minimum load they will show to your amplifier will be 4 Ohms. The MAXIMUM power that the LM3886 can produce into 4 Ohms is 68W with a power supply of +/- 28VDC.
If you keep within this power supply you will certainly be OK.
To get 28VDC you will need a transformer with TWO secondaries each of 28 / 1.4142 = 19.7V. (21V would be close enough but is not a standard transformer)
You have two choices. Either use a standard 18V and accept that the power output will be slightly reduced (you'd be hard pressed to hear the difference) or use a higher voltage supply with a voltage regulator.
For the time being we will keep it simple and use an 18-0-18V transformer.
Let us look at the current requirement next. The LM3886 is a Class AB/B amplifier. It's pretty efficient and manages to put about 80% of the power into the load. We'll be generous here as well and call it 70%.
In order to put 68W into the load it must have a supply of AT LEAST 97W - lets call it 100W. The LM3886 has a maximum transient output of 135W so the PSU must have a transient capability of 193W (lets call it 200W)
Your PSU should be deisgned to deliver +/- 28VDC at 200W or 3.6A.
Now, for cost saving, this power does not have to be developed all the time. The transient current can be provided by the reservoir capacitors of the power supply.
Normal running is only +/- 28VDC at 100W or 1.8A.
Standard toroidal transformers are 160VA, 300VA, 500VA.
Each amplifier can be easilly supplied by a single 16oVA transformer or a stereo pair by a 300VA transformer.
As we are skimping a bit on the voltage of the transformer it would be beneficial to use Shottky Diodes, these have a lower Vf and work faster so more energy is recovered from the transformer.
You will see many designs for power supplies, in PSUs with "normal" diodes you may see capacitors across the rectifier diodes - if you use SHOTTKY DIODES - DO NOT USE capacitors across the rectifier diodes.
The reservoir capacitors are amongst the key components of your design. These need to be Good Quality High Ripple Current components - generally Computer Grade Capacitors. They are NOT CHEAP. You need a minimum of 4700uF at 35V but the more the merrier. If space permits use lots of smaller capacitors. 10x 4700uF will often be better than 1 x 330000uF cap. The LM3886 has a good PSRR (Power Supply Rejection Ratio) but it certainly benefits from a good solid PSU.
Each amplifier should have its own rectifiers and capacitors. Preferably each should have its own transformer. However, in a six channel design there will naturally be a bit of crosstalk between the rear and front channels so all the LEFT channels could use one transformer as could the RIGHT.
That leaves us with the problem of heat.
This is an inexact science and is better approached with an experimental attitude.
As we have already said, each LM3886 is about 70% efficient (its actually better than that but bear with me). If we are to develop 68W into our speaker, the amplifier will be using about 98W of power (we'll use 100W). Somewhere along the line we've got to help the LM3886 dissipate 30W of heat, the difference between the two figures.
How warm is your living room ? Let us say it's a hot summer day and it's 25 degrees.
How hot can the LM3886 get before it melts - From the data sheet = about 150 degrees.
For any sort of reliabilty it is recommended that the device is kept below 65 degrees.
That allows the temperature to rise 40 degrees above our hot summer day temperature.
We now have a simple calculation 40 degrees / 30 Watts = 1.33 Degrees / Watt
Heatsink manufacturers are ever so generous with the claims of heat dissipation from their heatsinks - you would be foolhardy to expect a 1.3 K/W heatsink to actually work at that level. Designers normally give the manufacturers a good 20% leeway. Let us take 1.33 K/W and multiply it by 1.2, that gives us 1.5 K/Watt.
Not a massive heatsink but still quite substantial.
The LM3886 must be mounted to the heatsign with a small amount of good heatsink compound and the heatsink MUST have a good FREE airflow around it with the fins being VERTICAL.
We have taken the device to its limits but have applied sensible maths to the design. This arrangement will operate well within the SOAR (Safe Operating Area) of the device and should give good reliabilty. It will also not be so hot that you can't touch the amplifier.
As a good rule of thumb, you should be able to hold your hands against a heatsink for about 5 seconds. If it's hot but doesn't burn then it is probably OK.
DIY LM3886 Chip Amplifier (Gainclone) Project
However, they are also relatively easy to get on E-Bay.
The Datasheet is probably a bit daunting for a newcomer to DIY design, the link above shows a complete design.
As Andrew T saya, "breadboarding" as we call it, is fun, but all the dangerous voltages can be exposed and can kill if not cause substantial damage. All mains voltages to the transformer or transformers MUST be INSULATED and FUSED. Toroidal transformers, the transformers of most peoples choice, do have quite a high inrush current, so. A nominal 160VA transformer would be reasonably expected to have a steady state mains current of 1.4A at 115V or 0.7A at 240V. You will need anti-surge fuses nominally rated at 2 to 3 times this figure (ie 3A at 115V or 2A at 240V). Truly massive 1-2000VA transformers can have even higher inrush currents and some other form of limiting the current is often necessary. We'll stick with one amplifier (MONO) for the time being.
As with all GainClone amplifiers. Keep all leads to the IC as short as possible and keep the input leads away from the output leads - if you don't you may end up with a high power oscillator.
One word on Power Supplies.
Let us assume that the LM3886 datasheet is accurate and not misleading.
68W cont. avg. output power into 4Ω at VCC = ±28V
38W cont. avg. output power into 8Ω at VCC = ±28V
50W cont. avg. output power into 8Ω at VCC = ±35V
Consider these figures as ABSOLUTE MAXIMUMs.
For safety let us assume that your are using a good pair of B&W DM602's obtained second hand for about £100 - Excellent staring point for speakers. They are a nominal 8 Ohms but dip to about 5 Ohms at certain points in the audio spectrum.
For the safety of your LM3886 we will assume that the minimum load they will show to your amplifier will be 4 Ohms. The MAXIMUM power that the LM3886 can produce into 4 Ohms is 68W with a power supply of +/- 28VDC.
If you keep within this power supply you will certainly be OK.
To get 28VDC you will need a transformer with TWO secondaries each of 28 / 1.4142 = 19.7V. (21V would be close enough but is not a standard transformer)
You have two choices. Either use a standard 18V and accept that the power output will be slightly reduced (you'd be hard pressed to hear the difference) or use a higher voltage supply with a voltage regulator.
For the time being we will keep it simple and use an 18-0-18V transformer.
Let us look at the current requirement next. The LM3886 is a Class AB/B amplifier. It's pretty efficient and manages to put about 80% of the power into the load. We'll be generous here as well and call it 70%.
In order to put 68W into the load it must have a supply of AT LEAST 97W - lets call it 100W. The LM3886 has a maximum transient output of 135W so the PSU must have a transient capability of 193W (lets call it 200W)
Your PSU should be deisgned to deliver +/- 28VDC at 200W or 3.6A.
Now, for cost saving, this power does not have to be developed all the time. The transient current can be provided by the reservoir capacitors of the power supply.
Normal running is only +/- 28VDC at 100W or 1.8A.
Standard toroidal transformers are 160VA, 300VA, 500VA.
Each amplifier can be easilly supplied by a single 16oVA transformer or a stereo pair by a 300VA transformer.
As we are skimping a bit on the voltage of the transformer it would be beneficial to use Shottky Diodes, these have a lower Vf and work faster so more energy is recovered from the transformer.
You will see many designs for power supplies, in PSUs with "normal" diodes you may see capacitors across the rectifier diodes - if you use SHOTTKY DIODES - DO NOT USE capacitors across the rectifier diodes.
The reservoir capacitors are amongst the key components of your design. These need to be Good Quality High Ripple Current components - generally Computer Grade Capacitors. They are NOT CHEAP. You need a minimum of 4700uF at 35V but the more the merrier. If space permits use lots of smaller capacitors. 10x 4700uF will often be better than 1 x 330000uF cap. The LM3886 has a good PSRR (Power Supply Rejection Ratio) but it certainly benefits from a good solid PSU.
Each amplifier should have its own rectifiers and capacitors. Preferably each should have its own transformer. However, in a six channel design there will naturally be a bit of crosstalk between the rear and front channels so all the LEFT channels could use one transformer as could the RIGHT.
That leaves us with the problem of heat.
This is an inexact science and is better approached with an experimental attitude.
As we have already said, each LM3886 is about 70% efficient (its actually better than that but bear with me). If we are to develop 68W into our speaker, the amplifier will be using about 98W of power (we'll use 100W). Somewhere along the line we've got to help the LM3886 dissipate 30W of heat, the difference between the two figures.
How warm is your living room ? Let us say it's a hot summer day and it's 25 degrees.
How hot can the LM3886 get before it melts - From the data sheet = about 150 degrees.
For any sort of reliabilty it is recommended that the device is kept below 65 degrees.
That allows the temperature to rise 40 degrees above our hot summer day temperature.
We now have a simple calculation 40 degrees / 30 Watts = 1.33 Degrees / Watt
Heatsink manufacturers are ever so generous with the claims of heat dissipation from their heatsinks - you would be foolhardy to expect a 1.3 K/W heatsink to actually work at that level. Designers normally give the manufacturers a good 20% leeway. Let us take 1.33 K/W and multiply it by 1.2, that gives us 1.5 K/Watt.
Not a massive heatsink but still quite substantial.
The LM3886 must be mounted to the heatsign with a small amount of good heatsink compound and the heatsink MUST have a good FREE airflow around it with the fins being VERTICAL.
We have taken the device to its limits but have applied sensible maths to the design. This arrangement will operate well within the SOAR (Safe Operating Area) of the device and should give good reliabilty. It will also not be so hot that you can't touch the amplifier.
As a good rule of thumb, you should be able to hold your hands against a heatsink for about 5 seconds. If it's hot but doesn't burn then it is probably OK.
The heatsink calc. should have been 1.3 / 1.2 = 1.0 K/Watt
But anything between 1 and 1.5 K/W will suffice under normal conditions.
But anything between 1 and 1.5 K/W will suffice under normal conditions.
- Status
- Not open for further replies.