| || || |
This is for how to make an other version of a GainClone amplifier. I named this one ‘Gain-X2’
(Gain-Chris-2). This is a High-End GainClone! If you want it to be Hiend, this is it!
OK, I admit, I am stretching the consept of a "Gain Clone" quite a lot here now...
The amplifier uses one or more mono amplifier chips, the LME49811 from National Semiconductor. It is more complex to use than the LM1875, LM3875 or -86, and not the easiest to build the amplifier from.
This amplifier uses Toshiba power MOSFETs at the output stage, but you could use equivalents as well.
You will need some test equipment to put it together - for example an oscilloscope or a PC-Scope.
However, amplifiers made with this chip is understood to be one of the best in sound quality compared to anything.
You will be able to make an extremely high quality class-A or -AB amplifier from this!
In the best conditions with 10W output power the THD will be way under 0.0005% for an 8ohm load at 1kHz.
The maximum output power is around 150W depending on the selected output mosfets, but you need a pretty high voltage PSU for that.
Minimum supply voltage is +-20V. You can use the PSU from Gain-X to start with, but a more powerful PSU is highly recommended. I used a +-24V,144W switch mode power supply, but we will come to that later...
Well guess what? I got inspired to build this one after first making the ‘Gain-X’, of course!
Idea behind that was to put together a small and still pretty decent, but low-cost amplifier.
Because of the good experience with that little thingy, I thought that what could actually been put together if I used some really big and expensive components and instead of making it small, what if I made it actually quite BIG? Well "big" is relative. It became still "relatively" small.
The idea of the Gain-X2 was born!
No alternatives and only one decision:
As I wrote in the Gain-X article, I did find out that people are building very high quality chipamps from the LM3875 and LM3886. But is that it? Or is there any better? Then I came across National Semiconductors LME-family of chips. And I was in love immediately! Those are exactly what I want! NS has some application notes about the LME-family. I read every single one of them extremely carefully. Plus the application notes about LM4702! And after reading them all, I read them again!
The LME-family includes LME49810, LME49811 and LME49830.
I decided to use the LME49811 and Toshiba's MOSFETs as the output stage.
No compromises. I wanted the best and I think I got what I wanted.
This time I did not bother about the size factor either. I could make this one a bit bigger.
Main reason for the bigger size is that a whole lot of heat must be dissipated. Big heatsinks will be required.
I still did not feel that I needed more output power than just a few Watts. But from those few, I am going to take all possible sound quality out for sure.
For that reason I biased it to Class-A!
The decision is, for it to operate in the class-A it shall be again only an 2*15W amplifier, which for sure is more power than I will ever need.
If you do not know or understand what the electricity coming out of the wall, from the power grid, can do:
DO NOT BUILD A PSU! BUY A READY MADE ONE!
+-24V, 220W for 4ohm speaker use. Minimum is about 120W, but it is good to have some head room. Note, for 4ohm speaker use I recommend the output stage to be separately powered with lower PSU voltages. +/-15V only.
For 8ohm speaker use; +-24V, 140W. Again, 50W/amplifier is the minimum, if you use the same supply to power the input stage and the output stage.
The national laws and regulations do not permit you to make any device that is directly coupled to the electricity net! This is one more good reason to use the switch mode power supplies available from almost anywhere.
Yes, I know! "Avoid them as far as possible!" - You are told by everyone... But not me! I decided to try them for this amplifier.
Because: They are crazy cheap! And they are crazy small! And they are crazy efficient!
They will of course be producing a lot of high frequency noise, but who says you must not filter it clean?
Of course you must! And let me show you how:
You could also use higher output voltage supplies, i.e. 36V or 48V per rail.
By doing so, you will get more output power from the amplifiers, but again you need to adjust the current limiting resistor values (if the supplies are not current limited themselves), and the 15kohm shutdown control resistors in the supply circuit to 22kohm for 36V supply and to 33kohm for 48V supply. And definitely adjust the bias resistor values in the amplifier circuit and the amplifier gain!
The values for your current limiting resistors depends on the current your supplies are capable of feeding.
These are their max values I'm talking about here now:
A 10ohm 50W resistor is fine for the 48V 5A supply.
For a 3A 48V supply you need a 20ohm 50W resistor.
36V 5A supply needs a 8,2ohm 50W resistor
and for a 3A 36V supply you need a 15ohm 50W resistor.
The resistors are needed for protecting the supply and the amplifiers output stage from short-circuit and from the crazy amount of current needed to charge the reservoir capacitors! They are also good for filtering out all sorts of noise and ripple.
These resistors must be fitted on a heatsink!
If the supplies are current limited, and therefore you wouldn't necessarily need the protection resistors, still a smaller value resistor could come in to question because of the noise filtering capability of it. Lets say a value of 1 to 2 ohm in that case, 25watt OK for most purposes...
If you decide to use the switch mode supplies, the output from them is already rectified and regulated. The supplies may also have thermal protection and short circuit protection. They are fused, so you may not need a fuse in front of them. Actually the power switch is also optional. You can keep the supplies totally external and connect their outputs to your amplifier with two lines and DC connectors. For sale is a master/slave extension cord/socket thingy, which could be your power switch. You connect the amplifier power supplies to the slave outputs and your signal source (a CD player?) to the master output. When you turn power on to your CD, also the amplifiers will be turned on! Pretty handy! And most important of all, your insurance company totally loves the solution! You haven't been messing up with any equipment in direct contact with the power grid - how nice!
Anyways, depending on the quality of the power supplies you got, the output of them may or may not be pretty noisy. We must eliminate that noise.
The answer is in the schematic above!
100mH (coupled) inductors in the circuit with the capacitors shown will eliminate all noise what so ever from the supplies.
I have got some questions about these inductors. Why so big? It is an over kill I admit, but the bigger the better. These are the keys to the gates of the heaven, believe me! At least use 10mH if 100mH is out of the question for you...
After the inductors and filter capacitors we need reservoir capacitors. Big ones!
As big as you can get and afford! I have two 47 000uF capacitors, one for each power rail.
Even better would be a whole capacitor bank, consisting of many smaller capacitors in parallel.
The high current diodes protect the power supplies from current getting back to them after switching them off. Because of these diodes, after switch off, the circuit also puts the amplifiers very soon into shutdown mode and you will not hear any loud snaps or anything from the speakers. Totally silent during power on and off! Nice!
The +++24V lines are for the amplifier shutdown control.
The ++24V lines are for amplifier input stage power.
The +24V lines are for output stage power.
A lot of GND contacts. This is a star ground for all system. Here you connect every ground line from everywhere. The speaker grounds, the amplifier grounds the chassis ground - everything! I did myself connect the input signal ground to the amplifier board itself with extremely good results, so that is an option for you too, I can recommend.
The -24V lines are for output stage again.
And finally the --24V are for the input stage power.
That's it! Power supply ready!
If you use this power supply for the LM1875 Gain-X, you only need to connect the +24V, GND and -24V lines. And of course the GND. You can use a 100kohm resistor connected to one of the ++24V and --24V lines to drain the reservoir capacitors.
Recommended also for the Gain-X2. Otherwise there will be almost no current to drain the capacitors when the amplifier is in the shutdown mode after power off.
Finally we are in the fact of the matter!
LME49811 is a High Fidelity power amplifier input stage capable of driving an output stage of bipolar or even some mosfet transistors.
National offers a very good data sheet about the chip. Read it! Also read all other data sheets about LME49810, 49830 and LM4702! Very useful information in them all.
I pretty much used the circuit from the LM4702 data sheet about driving mosfet output stage with some minor changes.
The schematic I made for the -811 is this:
This schematic is to be used with the +/- 24V power supply shown in the previous circuit only! If you plan to use more powerful supply, or a power supply with higher output voltage, you need to change some component values.
Firstly, the gain. I use here the gain of 20V/V. If you have more supply voltage, increase the gain accordingly. You can increase the gain by changing resistor RF to more resistive. Resistor RS should have same value. Change that too. I used RS as the volume control potentiometer with good results. Also CN needs to be changed to less capacitive.
Secondly, in case of having a different power supply you also need to readjust the bias to match it.
Component values shown here makes it Class A.
If you want to use different bias setting or want class AB operation, you need to change some resistor values.
Start by changing the bias resistor RB1 to say 560ohm. Decrease value step by step until satisfied. Do not go much lower than 330ohm. For an adjustable bias you can have a 500ohm trimpot in place of the RB1 with a 220ohm resistor in serial.
You can measure the bias current from the +24V supply line. Turn your current meter to the 10A setting and use the meter as the power lead between the PSU and the amplifier PCB.
For Class-A at 16W/8R you need 1A of bias current. Notice that you will then be dumping 48W power per amplifier all the time if you use 24V rails at the PSU. Remember this when choosing the heatsinks.
Also, note that 16W/8R needs only +/-16V supply rails! You could therefore make a quad rail PSU which gives +/-24V for the input stage (i.e. the LME49811) and lets say +/-18V for the output stage (i.e. the transistors). Make it +/-18V so that you have a little headroom there...
With 18V rails for the output stage you will be dissipating only 36W/amplifier. This helps a lot when choosing the heatsinks.
You can also experiment how changing the value of bias capacitor CB affects the sound quality. You might for example have an other capacitor paralleled with it, even a several microfarads capacitor could be tested here!
The VBE multiplier transistor 2SD438 can be replaced with TIP31.
If you use my PCB (sold out unfortunately, at the moment), the holes are drilled for 2SD438, but if you cut the TIP31 pins needle sharp, you can easily solder that to the place.
Compensation capacitor CC can be changed to lower value capacitor, as low as 10pF can be used. This capacitor is critical. Use Silver Mica capacitor here. If amplifier starts oscillating, increase value.
I used 1kohm trimmers in place of the gate resistors (RGN and RGP). Value can be adjusted for best possible results. Values shown in the schematic should be close starting values for the trimmers. Read National's AN-1645, you have all the necessary information about this adjustment in there.
After finding the right value, replace with 1% metal film resistors. 0,5-0,6W OK...
At the output stage I used the Thoshiba MOSFETs 2SK1529 / 2SJ200 pair, but you can use equivalents as well.
For example the Toshiba 2SK1530 / 2SJ201 pair.
Or from Renesas the 2SK1057 / 2SJ161 or 2SK1058 / 2SJ162 pairs.
Or from Magnatec the BUZ900 / BUZ905 or BUZ901 / BUZ906 pairs.
Magnatec and Renesas needs different biasing and gate resistors. Read the AN-1645!
But if enough interest rises, I may reorder some.
PCB size is 5cm*5cm (i.e. 2"*2").
I can sell them for 20e each, (i.e. 25$/pc) post and packing included.
You will also get the component layout and wiring diagram for them.
Payment by PayPal. Contact me by email, if you are interested...
There will also soon be a single sided PCB design downloadable, which you can then etch for yourself, so stay tuned...
For the power supply you need to design your own PCB. I have no PCBs for it.
The final verdict:|
Here is GX2 doing 100kHz test signal to a 8ohm resistive load at 60W.
Red is input. Blue is output.
And just for fun 1MHz 5W 8R...
Especially notice the crossover distortion - or actually the lack of it...
The THD measured with a Creative SB X-Fi HD USB soundcard and Visual Analyzer 2011 software at 48kHz sample rate, for a 1W 1kHz 8R signal and load is this:
Green is input signal from the soundcard to the amplifier. Red is the 1W output to the 8R resistive load.
Even though we could assume from this that the THD is only 0.0002%, and even though you could clearly see it from the graphs above... Let us not make this assumption just yet. As you can see, the main source to the distortions and noise is the soundcard itself and we therefore need way better equipment than this to measure more accurately what the exact THD performance of this amplifier will be. But I can still say and assure, that the THD contribution to the signal from the amplifier is totally negligible.
How does this all sound then?
It sounds not just good. It is better than anything! It is no longer just a stereo amplifier - it's three dimensional!
You can hear every peace of instrument just as it was recorded in that exact place it was when it was recorded, and not just from the speakers - it comes right inside your head! Like you were having headphones, only better! You can feel the sound.
If the Gain-X was already the Nirvana - this is beyond it, what ever there is...
Thanks for reading. And happy listening!