Hi,
(Based upon the datasheet design, somewhat)
After a LM4870 and a dual LM3886 amplifier I wanted a little bit bigger and better.
I’m using a 300VA toroid with 2 32VAC sec. That would gave me around 45V, in paper of course. In reality sometimes more than 50, so 63 V capacitors is needed for safety. I calculating around 100W in to 8 ohm, and 120 6 ohm, no 4 ohm loads (if this wound be a perfect world. It’s not). The two pair is adequate to it per chan. (2SC500/2SA1943). Pre drivers hmm maybe MJE30/50 or Toshiba I don’t know right now. Open to suggestions.
Input independence is low, but I wanted to keep the DC balance from the other side.
You can see the input LP and a HP, 3 hertz (same as in the NFB loop) and 136 kHz.
Also lowered the gain from 29 to 22, because this version doesn’t have a VI limiter and make it a little harder to clip it. And not least make the NFB resistance smaller.
Also the single pole is bigger for stability and I don’t need more. The slew rate around 25V/µs. Later I’m going to try two pole too, but now let see it is stable enough for it.
The interesting part is the clip pin. I want to know when it clips, so then I can ignore the PGA2310s values das no overload. I don’t know that I can just connect an arduino pin to it or I need a transistor for it, like in the scem.
Every part of the schem is free for argument and advice.
Thanks.
(Based upon the datasheet design, somewhat)
After a LM4870 and a dual LM3886 amplifier I wanted a little bit bigger and better.
I’m using a 300VA toroid with 2 32VAC sec. That would gave me around 45V, in paper of course. In reality sometimes more than 50, so 63 V capacitors is needed for safety. I calculating around 100W in to 8 ohm, and 120 6 ohm, no 4 ohm loads (if this wound be a perfect world. It’s not). The two pair is adequate to it per chan. (2SC500/2SA1943). Pre drivers hmm maybe MJE30/50 or Toshiba I don’t know right now. Open to suggestions.
Input independence is low, but I wanted to keep the DC balance from the other side.
You can see the input LP and a HP, 3 hertz (same as in the NFB loop) and 136 kHz.
Also lowered the gain from 29 to 22, because this version doesn’t have a VI limiter and make it a little harder to clip it. And not least make the NFB resistance smaller.
Also the single pole is bigger for stability and I don’t need more. The slew rate around 25V/µs. Later I’m going to try two pole too, but now let see it is stable enough for it.
The interesting part is the clip pin. I want to know when it clips, so then I can ignore the PGA2310s values das no overload. I don’t know that I can just connect an arduino pin to it or I need a transistor for it, like in the scem.
Every part of the schem is free for argument and advice.
Thanks.
Attachments
Hi,
Well technically BJTs are more linear. MOSFETs are more rugged and need more voltage (voltage controlled ampl.), and the most cases people chose it in big amps, because MOSFETs are somewhat immune to second breakdown (well Zeners are not a bad things on the Gate leg) . But lme49830 does not have a clip pin. If wanted to build a bigger then 200+W amp, I would build a FET one too. You can found a good ref design in High-Power Audio Amplifier Construction Manual by G. Randy Slone. If i remember correctly. (nope it was Designing Audio Power Amplifiers by Bob Cordell)
Also TI lm49830 page under "Application Notes". Awesome material.
Z.
Well technically BJTs are more linear. MOSFETs are more rugged and need more voltage (voltage controlled ampl.), and the most cases people chose it in big amps, because MOSFETs are somewhat immune to second breakdown (well Zeners are not a bad things on the Gate leg) . But lme49830 does not have a clip pin. If wanted to build a bigger then 200+W amp, I would build a FET one too. You can found a good ref design in High-Power Audio Amplifier Construction Manual by G. Randy Slone. If i remember correctly. (nope it was Designing Audio Power Amplifiers by Bob Cordell)
Also TI lm49830 page under "Application Notes". Awesome material.
Z.
Last edited:
Looking foward to this project, I have something similar in my mind, I might help with PCB design, my idea was to use single output pair because I have 8 Ohm speakers...
I pretty much have everything I need to build it right now but I need some time...
Maybe in the weekend I´d like to try your idea...
I pretty much have everything I need to build it right now but I need some time...
Maybe in the weekend I´d like to try your idea...
Well I planning to build it in the next 2 weeks when I go back to collage. Also thing about adding a cap in the NFB path to shorter the bandwidth. Maybe 22 pF, I don't, that is gonna be my back up plan, if there is any HF instability. But only way to tell it is to build it and watch it through an oscilloscope. Technically any HF stability would disappear because of C3/2 capacitors. Technically...
Well I’m back with the psu and with soft start circuit.
The power supply is the simplest I ever made. A regular dual unregulated DC and a regulated voltage doubler for the preamps (DAC).
The toroid can provide around 9.4A at max (300/(2*32)). The question here is how much should I uprate the DC fuses, because 10 A fuses would be perfect, but the fuses very effected by temperature and it would blow, when the music on its ultimate peak? (4.63A the highest level / channel) So I made it 12A. AC fuses are 3.5A ((300W/230V)*2). (ALL T type)
The soft start circuit is digitally controlled and houses the AC line fuse, the relay control and the ground breaker for the analog parts (if I ever find a metal chassis). The power comes from a small all in one ~12V source PSU. This is always on, even when the amp is in sleep mode (or something went terribly wrong). It gives me 5V too for the arduino and other digital parts (pga, leds).
I changed the original amp design too. I uprated the fuse ratings from 3.5A to 5A, just to be sure. Also I dropped the transistor silliness in clipping, I added a led with series with R18 as in the datasheet design, but I also added a 1k1 resistor, that goes to the arduino’s analog leg.
Everything what you see is pretty much the final version, before I build it.
BUT if somebody wants to correct me about it, such as … fusing or anything else, go for it.
The power supply is the simplest I ever made. A regular dual unregulated DC and a regulated voltage doubler for the preamps (DAC).
The toroid can provide around 9.4A at max (300/(2*32)). The question here is how much should I uprate the DC fuses, because 10 A fuses would be perfect, but the fuses very effected by temperature and it would blow, when the music on its ultimate peak? (4.63A the highest level / channel) So I made it 12A. AC fuses are 3.5A ((300W/230V)*2). (ALL T type)
The soft start circuit is digitally controlled and houses the AC line fuse, the relay control and the ground breaker for the analog parts (if I ever find a metal chassis). The power comes from a small all in one ~12V source PSU. This is always on, even when the amp is in sleep mode (or something went terribly wrong). It gives me 5V too for the arduino and other digital parts (pga, leds).
I changed the original amp design too. I uprated the fuse ratings from 3.5A to 5A, just to be sure. Also I dropped the transistor silliness in clipping, I added a led with series with R18 as in the datasheet design, but I also added a 1k1 resistor, that goes to the arduino’s analog leg.
Everything what you see is pretty much the final version, before I build it.
BUT if somebody wants to correct me about it, such as … fusing or anything else, go for it.
Attachments
the trigger transistors on the relay coils must be driven into saturation.
Assume a saturated hFE of 10, or slightly less.
The current from the arduino must be > coil current/10
2seconds is far too long for a soft start. try 200ms
How do you get +-10Vdc from a pair of 6V regulators?
Have you measured what the amplifier does if one supply rail fuse blows?
Assume a saturated hFE of 10, or slightly less.
The current from the arduino must be > coil current/10
2seconds is far too long for a soft start. try 200ms
How do you get +-10Vdc from a pair of 6V regulators?
Have you measured what the amplifier does if one supply rail fuse blows?
Last edited:
6V regulator is just a typo, just a place holder. In reality I would use a 10V or 9V reg.
Because I did not build it yet, so I can only speculate two things. DC or nothing. If DC happens then the UPC breaks the relays and sends a signal to the arduino so, then I can shot down the toroid itself in no time. The second guess nothing the LME loses its power, so no driving capability, just awkward pure silence.
Because I did not build it yet, so I can only speculate two things. DC or nothing. If DC happens then the UPC breaks the relays and sends a signal to the arduino so, then I can shot down the toroid itself in no time. The second guess nothing the LME loses its power, so no driving capability, just awkward pure silence.
Well, finally had time to build it. But there is a problem, as always. Limiter resistors are in hell temperature under no load.
In the fuse place there is a 27R 5W resistor, and its gets hot in 10 sec. Both sides. No DC output (~+24mv). First I thought that I made a short cut in the driver stage but no. There was a heat sink in the drivers and VBe transistors. The PSU 8A T fuses are holding. The output transistors are warm, but i think only the BIAS. There is no short cut far I can see, so maybe its a design problem.
Bias should be around 1.35, based on the resistors and the trimm pot (500R) is in middle position. I dint measured it yet. No wires reserved (good old + in - place and vice versa problem).I also measured the heat sink around 4mV, so only some static distract thing. Every transistor micad. No other thing gets hot.
Soo I have no idea what is the problem. I will measure the rails and bias voltages, but i think the problem lays in other place. Any suggestion?
In the fuse place there is a 27R 5W resistor, and its gets hot in 10 sec. Both sides. No DC output (~+24mv). First I thought that I made a short cut in the driver stage but no. There was a heat sink in the drivers and VBe transistors. The PSU 8A T fuses are holding. The output transistors are warm, but i think only the BIAS. There is no short cut far I can see, so maybe its a design problem.
Bias should be around 1.35, based on the resistors and the trimm pot (500R) is in middle position. I dint measured it yet. No wires reserved (good old + in - place and vice versa problem).I also measured the heat sink around 4mV, so only some static distract thing. Every transistor micad. No other thing gets hot.
Soo I have no idea what is the problem. I will measure the rails and bias voltages, but i think the problem lays in other place. Any suggestion?
Attachments
Last edited:
What are "Limiter resistors" ?
Vbias set to 1.35V is probably too high.
You are biasing a single pair of Vbe. That can be around 1.13V to 1.24V
Add on 10mV for each of R5 & R8 and you get Vbias from ~1.15V to ~1.26V
Do as Zol suggests.
Turn the VR to maximum resistance for first power on. Check all your voltages to ensure no errors.
THEN start setting Vbias by measuring distortion at the output. This will probably result in Vre ~5mV to ~10mV.
Vbias set to 1.35V is probably too high.
You are biasing a single pair of Vbe. That can be around 1.13V to 1.24V
Add on 10mV for each of R5 & R8 and you get Vbias from ~1.15V to ~1.26V
Do as Zol suggests.
Turn the VR to maximum resistance for first power on. Check all your voltages to ensure no errors.
THEN start setting Vbias by measuring distortion at the output. This will probably result in Vre ~5mV to ~10mV.
palstanturhin :
Yes, I have scope.
AndrewT:
Technically the limiter resistors are limiting the current so, if I made a bad wiring should heat up and show that "Jeh something is not right". But I also have fuses. but I don't want to waste them in the test runs. But in normal no load ( I haven't tried it with load) conditions in class-B should not dissipate to much heat, so jeh bias it is. I calculated it with avg diode drop== bias=0.65V*2. It seems that I was wrong. I will check it at Monday.
Yes, I have scope.
AndrewT:
Technically the limiter resistors are limiting the current so, if I made a bad wiring should heat up and show that "Jeh something is not right". But I also have fuses. but I don't want to waste them in the test runs. But in normal no load ( I haven't tried it with load) conditions in class-B should not dissipate to much heat, so jeh bias it is. I calculated it with avg diode drop== bias=0.65V*2. It seems that I was wrong. I will check it at Monday.
Hi,
I was build this combination about 2 year ago..its very stable.I m using since that time.First; you MUST connect to Q3 (mje340) same heatsink with power trans. second; replace R13:1.2k R14:330R c5:1u bias trimpot 200-500R.it should be work.
(Sory for my bad English)
I was build this combination about 2 year ago..its very stable.I m using since that time.First; you MUST connect to Q3 (mje340) same heatsink with power trans. second; replace R13:1.2k R14:330R c5:1u bias trimpot 200-500R.it should be work.
(Sory for my bad English)
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
I switched to 2 bias diodes. Bias around 1.21v somewhat over biased but good for know. The only problem now is only distorted sound comes out. Rails are 37.0 and 37.4v under 8r Load. I Checked the mute but 30k should be enough for 5-6 volts to get that 1-8 mA what is required to lme to be in play mode. Well jeh.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.