Headphone Amplifier
Hello, I recently build Rod Elliot's project 113 (headphone amp, low power speaker amp)
I've tested it on my speakers in series ( I only made one channel to test the schematic) 12ohm, around 2 watts, the sound is gorgeous..
Would putting more BD's in parralel increase power? Or Can I make a darlington design from BD's and some power transistors?
What about swapping the BD's with TIP41/42 for example, or I need some drivers even for the tip's.
At +-18v I can get around 25w which is enough.
Hello, I recently build Rod Elliot's project 113 (headphone amp, low power speaker amp)
I've tested it on my speakers in series ( I only made one channel to test the schematic) 12ohm, around 2 watts, the sound is gorgeous..
Would putting more BD's in parralel increase power? Or Can I make a darlington design from BD's and some power transistors?
What about swapping the BD's with TIP41/42 for example, or I need some drivers even for the tip's.
At +-18v I can get around 25w which is enough.
Last edited by a moderator:
Hi Bruno,
For that purpose, you need the higher rails and more powerful output transistors with higher bias, thermally stabilized with a bias spreader - practically, a full output stage re-design.
Look at the project 3A at Elliott's web site - it's got the OPS designed for lower load impedance / higher output power.
Cheers,
Valery
For that purpose, you need the higher rails and more powerful output transistors with higher bias, thermally stabilized with a bias spreader - practically, a full output stage re-design.
Look at the project 3A at Elliott's web site - it's got the OPS designed for lower load impedance / higher output power.
Cheers,
Valery
I think you could scale this one to maybe 5-10 W into 8 ohms using some more powerful output transistors with low beta droop, a more powerful opamp, a number of other component changes and some better mounting of bias diodes. Maybe 15 W with a bit of a following wind and stars properly aligned.
The plan to simply boost the current of an opamp is fine as long as you don't exceed the permissible supply rail maximum voltages. As power amplifiers are usually operated unregulated, the supply voltages would then have to be conservative to avoid damage. +/-15V should be maximum for most opamps used that way. This design was intended for regulated supplies though and the regulators used will restrict current to their internally set limit. That's useful but the effects of clipping can be unpleasant this way and with small amplifiers, clipping is more likely than with say, 40-100W amplifiers.
With a given speaker load, the output voltage determines the maximum power possible before clipping. Adding more or higher current output transistors only means you have more current available to meet the demands of lower impedance loads or a higher output voltage into the same load. However, they do nothing to generate the necessary higher output voltage. To increase power into the same 8 ohm load, you need to increase output voltage and current capabilty which usually requires a more complex design incorporating voltage gain too.
That's quite possible and has been done many times too but you need to use a more complex design and because with such complexity, stability is no longer a given thing, so just tinkering with a breadboard and a handful of parts, but no training or suitable test instruments, will often end badly
You could start with a look at ESP's P76 project: Opamp Based Power Amp
Elektor magazine presented variations on a few boosted opamp designs over 30+ years and there are already threads here discussing them.
Some more examples here too: How to boost the output voltage swing of an operational amplifier | How To Wiki | Fandom powered by Wikia
Here is a serious paper by Jim Williams (Linear Technology) that treats the topic in some depth: http://www.tenmilecreek.net/images/Op_Amp_Booster_Stages_pt1.pdf
With a given speaker load, the output voltage determines the maximum power possible before clipping. Adding more or higher current output transistors only means you have more current available to meet the demands of lower impedance loads or a higher output voltage into the same load. However, they do nothing to generate the necessary higher output voltage. To increase power into the same 8 ohm load, you need to increase output voltage and current capabilty which usually requires a more complex design incorporating voltage gain too.
That's quite possible and has been done many times too but you need to use a more complex design and because with such complexity, stability is no longer a given thing, so just tinkering with a breadboard and a handful of parts, but no training or suitable test instruments, will often end badly
You could start with a look at ESP's P76 project: Opamp Based Power AmpElektor magazine presented variations on a few boosted opamp designs over 30+ years and there are already threads here discussing them.
Some more examples here too: How to boost the output voltage swing of an operational amplifier | How To Wiki | Fandom powered by Wikia
Here is a serious paper by Jim Williams (Linear Technology) that treats the topic in some depth: http://www.tenmilecreek.net/images/Op_Amp_Booster_Stages_pt1.pdf
Something like this is more appropriate as a power amp:
PSU was made out of 200VA transformer (2x15V AC => +/-20V DC, single 10A diode bridge and 10,000uF/25V per rail is OK).
AD823 is supplied through 16V/1W Zener diodes and since it's a rail-to-rail OpAmp it will produce enough voltage swing to fully use the output stage made out of lateral mosfets source follower complementary pair. You can expect about 13W at 8 Ohms per channel or about 20W at 4 Ohms. Output stage has higher voltage power supply due to Vgs losses.
AD823 has JFET input and very low and stable DC offset so you can put volume pot at the input. Nothing to adjust - D1/D2 will reliably bias the output stage to about 150mA and the feedback loop will take care about DC offset.
Thermal stability is ensured through lateral mosfets' characteristics.
My build didn't need any external freq. compensation although it was used with 4-Ohm loudspeakers (medium to high reactive).
Similar results can be achieved with LM1875 chip but this design sounds noticeably better.
PSU was made out of 200VA transformer (2x15V AC => +/-20V DC, single 10A diode bridge and 10,000uF/25V per rail is OK).
AD823 is supplied through 16V/1W Zener diodes and since it's a rail-to-rail OpAmp it will produce enough voltage swing to fully use the output stage made out of lateral mosfets source follower complementary pair. You can expect about 13W at 8 Ohms per channel or about 20W at 4 Ohms. Output stage has higher voltage power supply due to Vgs losses.
AD823 has JFET input and very low and stable DC offset so you can put volume pot at the input. Nothing to adjust - D1/D2 will reliably bias the output stage to about 150mA and the feedback loop will take care about DC offset.
Thermal stability is ensured through lateral mosfets' characteristics.
My build didn't need any external freq. compensation although it was used with 4-Ohm loudspeakers (medium to high reactive).
Similar results can be achieved with LM1875 chip but this design sounds noticeably better.
Attachments
I have Recovered from my Pioneer CD player / and a deck Mitsubishi 5238A J-fet input ( 15V +- , 18 max ) Rail-to-rail , slew rate 20v / uS even more than NE5532 9V/uS.
I dont have money for expensive stuff, but compared to TL072 It sound lot better, lower distortion and " wider sound " Also I have Mitsubishi 5220 wich is 25v +- ( 22.5 typical , non rail-to-rail tho ) So I can use 20V +- supply.
If you said the op-amp can deliver enough for a pair of mosfets, cant it do the same for a pair of BJT's like 2sc/2sa for example
I dont have money for expensive stuff, but compared to TL072 It sound lot better, lower distortion and " wider sound " Also I have Mitsubishi 5220 wich is 25v +- ( 22.5 typical , non rail-to-rail tho ) So I can use 20V +- supply.
If you said the op-amp can deliver enough for a pair of mosfets, cant it do the same for a pair of BJT's like 2sc/2sa for example
I made a darlington out of BD139/2sc4468 and BD140/2sa1695
Works nice , BUT at only 100mA the transistors "see" 0.6v BE , because a darlingtone needs double the voltage on BE right?
And Maybe I need to stick the diodes or the BD's to the heatsink, because from 100mA quis current it got to 200 because heat , and now it is 400mA becuase even more heat, SO Im experiencing thermal runaway ..
I'v read somewhere that compound ( Sziklai pairs ) need less quisance current and theyr Vbe is 0.6v not 1.2v like the darlington, That's what Rod used in project 76 Opamp Based Power Amp
But I dont understand why on the ( Q3 Q4 ) from base of Q3 to emitter of Q4 doesnt need a resistor, the bias is made only by the upper transistors Q1 Q2 reistors R7 and the pot ?.
Works nice , BUT at only 100mA the transistors "see" 0.6v BE , because a darlingtone needs double the voltage on BE right?
And Maybe I need to stick the diodes or the BD's to the heatsink, because from 100mA quis current it got to 200 because heat , and now it is 400mA becuase even more heat, SO Im experiencing thermal runaway ..
I'v read somewhere that compound ( Sziklai pairs ) need less quisance current and theyr Vbe is 0.6v not 1.2v like the darlington, That's what Rod used in project 76 Opamp Based Power Amp
But I dont understand why on the ( Q3 Q4 ) from base of Q3 to emitter of Q4 doesnt need a resistor, the bias is made only by the upper transistors Q1 Q2 reistors R7 and the pot ?.
Also Juma thank you for the schematic , but I dont have mosfet's right now , and I would stick too BJT's for this one .
Yeah, everybody loves those mythical 2SJ162 2SK1058 parts. Just like they love unicorns.
There were some British lateral FET copies available for a while under other numbers but those are now joining the list of extinct beasts also I hear.
I can't even buy 2SA 2SC bipolars with heat tabs. I just had flown in from Leeds some plastic case TO225's for $9.68 freight. Same **** package as I got from digikey and mcmelectronics warehouses in US. What package you get is all in the suffix codes of datasheets that are NOT on the internet, and everybody available to US purchasers stocks high watt drivers that can only be cooled by psi power. Of course we can always buy fairy tales off E-bay.
There were some British lateral FET copies available for a while under other numbers but those are now joining the list of extinct beasts also I hear.
I can't even buy 2SA 2SC bipolars with heat tabs. I just had flown in from Leeds some plastic case TO225's for $9.68 freight. Same **** package as I got from digikey and mcmelectronics warehouses in US. What package you get is all in the suffix codes of datasheets that are NOT on the internet, and everybody available to US purchasers stocks high watt drivers that can only be cooled by psi power. Of course we can always buy fairy tales off E-bay.
Why mythical ?
I bought 10 pairs last month at reichelt.de at 4 EUROs a piece...
Nedis and Conrad and many other big reliable houses store them too, 2sk1058 and 2sj162 are still in production.
I bought 10 pairs last month at reichelt.de at 4 EUROs a piece...
Nedis and Conrad and many other big reliable houses store them too, 2sk1058 and 2sj162 are still in production.
Well in the western hemisphere, newark (farnell) mouser & alliedelec don't have any. Digikey says they are obsolete.
Avnet has 6 renesas 2sk1058 but the 2sj162 might be available in 16 weeks.
Avnet has a $50/line minimum order I believe.
The OP is in Europe I believe, so your find at a distributor in Germany may help him. That is the second schematic he has been shown with op amp input section that the special lateral FET's were recommended.
Avnet has 6 renesas 2sk1058 but the 2sj162 might be available in 16 weeks.
Avnet has a $50/line minimum order I believe.
The OP is in Europe I believe, so your find at a distributor in Germany may help him. That is the second schematic he has been shown with op amp input section that the special lateral FET's were recommended.
Last edited:
Ampslab in the US have been selling the Renesas parts for many years. Since there's only one package style - TO3P, there's no confusion there. Their US$7.50 price is post-free in the attached offer but of course, that comes out of their premium: 2SK1058 2SJ162 Audio Power Mosfets - Buy Online @ AmpsLab
I buy stuff from USA on a regular basis, with no problems. The other way round should work too...
Newark stores some nice laterals, Semelab's ALF16 parts at very nice price:
ALF16P16W - SEMELAB - MOSFET Transistor, P Channel, -16 A, -160 V | Newark element14
ALF16N16W - SEMELAB - MOSFET Transistor, N Channel, 16 A, 160 V | Newark element14
Each of them has actually two Renesas parts inside, matched and paralleled.
ALF16P16W - SEMELAB - MOSFET Transistor, P Channel, -16 A, -160 V | Newark element14
ALF16N16W - SEMELAB - MOSFET Transistor, N Channel, 16 A, 160 V | Newark element14
Each of them has actually two Renesas parts inside, matched and paralleled.
I have found this simple circuit, to learn more and try. I don't understand why is there a connection to ground from middle of R10-R11? That is the bootstrap?. Also R14-R17 what are they for?
It claims 50w rms...
It claims 50w rms...
Attachments
Last edited:
If the drivers Q1 & Q2 are not well matched, the ground connection helps keep the output in the middle of the power supplies and gives you more headroom against clipping when playing loud.
You should be able to put a pot somewhere to center the output up when idle. If the centering drifts off this could put DC current into your speaker while sitting around silent. Something like 100k pot between two 270 k resistors to plus & minus rail, wiper to the -input of the op amp. Audio op amps aren't trimmed to sit around stable with no offset voltage.
I don't have a clue about R14 & 17. It's 3AM in the UK, the experts wake up around 7.
The +-15 power supplies for the op amp are established here by 1.8k resistors. That would only work for that particular op amp from that factory and batch, likely. for your 20 v rail op amp I would use something like 1.3 W 18v zeners after the resistor, on the op amp side. Or low noise 18v regulators for the purists around here.
For the two ohm speaker double up on the output transistors and emitter resistors. I'd cut the base stoppers on the outputs to 22 ohms from 47 if using two.
Got any 1 ohm or .51 ohm 3 watt or 5 watt resistors yet? Back when I was a poor shop tech, making just above the rent + food, riding the bus everywhere, I bought my first ones of those in 1974 in a retail store. they are that useful.
You should be able to put a pot somewhere to center the output up when idle. If the centering drifts off this could put DC current into your speaker while sitting around silent. Something like 100k pot between two 270 k resistors to plus & minus rail, wiper to the -input of the op amp. Audio op amps aren't trimmed to sit around stable with no offset voltage.
I don't have a clue about R14 & 17. It's 3AM in the UK, the experts wake up around 7.
The +-15 power supplies for the op amp are established here by 1.8k resistors. That would only work for that particular op amp from that factory and batch, likely. for your 20 v rail op amp I would use something like 1.3 W 18v zeners after the resistor, on the op amp side. Or low noise 18v regulators for the purists around here.
For the two ohm speaker double up on the output transistors and emitter resistors. I'd cut the base stoppers on the outputs to 22 ohms from 47 if using two.
Got any 1 ohm or .51 ohm 3 watt or 5 watt resistors yet? Back when I was a poor shop tech, making just above the rent + food, riding the bus everywhere, I bought my first ones of those in 1974 in a retail store. they are that useful.
Last edited:
How much can I get out of this, because I already tried a darlington on the project 113 and it sound pretty good, around 10-15w on 4ohm
Someone suggested to change the output stage with the one of project 3a
Someone suggested to change the output stage with the one of project 3a
You picked a design with an old (and potentially unstable) CFP type output stage, where there is both voltage and current gain, the voltage gain being determined by the extra resistors. The principle is good and there were successful commercial and DIY products like this back in the 1970's, when the technique was popular. However, unless you have some experience and suitable instruments for troubleshooting power amplifiers, I wouldn't rely on the outcome. For comparison and a little understanding, here's Douglas Self's generic CFP output stage to show what the design might look like without voltage gain.
