I have finished yet, but not testing.
http://rapidshare.com/files/58923878/Low_Cost_20W.zip.html
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
http://rapidshare.com/files/58923878/Low_Cost_20W.zip.html
Kapibara said:
Hello
You will try the original version, try also the linuxguru modified version that I've done a schematic with few additions, I'm sure you will eard a good difference.
Gaetan
addendum
Alternative NPN low-noise, high-gain substitutions for Q1 (instead of 2n5961) include the jelly-bean bc550(c), mpsa18 and Toshiba 2sc3112(b). The last one is superb, with a hfe(min) of 1200 (1500 typ), NF = 0.5 db (typ), Vceo = 50V and fT = 250 MHz (typ) - however, it's probably hard to find.
The open-loop gain is important, because that's the only voltage-gain stage in this minimalist 3-transistor topology, and it helps to max it out to increase the loop-gain. The other two transistors only provide current gain.
It also makes a great choice for the input LTP and/or VAS in a wide variety of topologies, including that used in Symasym. I got hold of a few 2sc3112s, and I'll try them out shortly in a VAS. The only problem is the relatively low Vceo, which limits the rail voltages to +/- 25v if it's used in a VAS (this can be worked around using a cascoded VAS).
Alternative NPN low-noise, high-gain substitutions for Q1 (instead of 2n5961) include the jelly-bean bc550(c), mpsa18 and Toshiba 2sc3112(b). The last one is superb, with a hfe(min) of 1200 (1500 typ), NF = 0.5 db (typ), Vceo = 50V and fT = 250 MHz (typ) - however, it's probably hard to find.
The open-loop gain is important, because that's the only voltage-gain stage in this minimalist 3-transistor topology, and it helps to max it out to increase the loop-gain. The other two transistors only provide current gain.
It also makes a great choice for the input LTP and/or VAS in a wide variety of topologies, including that used in Symasym. I got hold of a few 2sc3112s, and I'll try them out shortly in a VAS. The only problem is the relatively low Vceo, which limits the rail voltages to +/- 25v if it's used in a VAS (this can be worked around using a cascoded VAS).
Update: What I got were 2sc3112a (probably genuine Toshiba), and the DC betas are clustered like peas in a pod tightly around 800, with a few outliers at ~750 and ~1100. It's still better than most bc550c and mpsa18 NPNs - I'd highly recommend it.
Hello
Which version it is, the original version or the one including Linuxguru modifications ?
How does it sound ?
Bye
Gaetan
Which version it is, the original version or the one including Linuxguru modifications ?
How does it sound ?
Bye
Gaetan
"D1-D3 reduce crossover distortion"..... sounds like class B biased to the edge of class AB. a lot of Pro-Audio amps as well as older Pioneer, Sansui, kenwood, etc... use fixed diode biasing. the diodes should be thermally bonded to the heat sink.
An externally hosted image should be here but it was not working when we last tested it.
Sound is better my ecpectation.
Better than TDA2030 or TDA2050 certainly for me.
Excellent choices - I like the TIP142/147 also. The Philips TIP142/147 seem to be sonically better than ST. YMMV, Philips hasn't made them since 1992 or so, and only NOS and counterfeits are available now. Other alternatives include ST BDX-33c/34c in a TO220-package.
You can try the 2sc3112 for Q1 ( if you can get it), instead of the bc550(c). It has a higher beta. It is also worth trying a one-transistor Vbe servo instead of the 3-diode + resistor chain. With bc550c and suitable resistors, the dynamic impedance of the transistor servo is lower - not sure of the impact on the sonics.
You can also experiment with even lower Miller caps on Q1 - say 33 to 47 pF, as long as it remains stable. Polystyrene or Silver Mica are sonically better than ceramic at this location. If you have to use ceramic, use a 1KV or 2KV rated device.
For the DC-blocking cap in series with the speaker load, try Panasonic FC - 2200 or 2700/25 should be adequate - 63v is way overkill. 2200/63 can be used for the supply filtering instead.
Try to use metal-film resistors for everything in the signal path, especially anything connected to the base of Q1.
Try a small MKP or MKT cap (say 0.47 to 1 uF) in parallel with input DC-blocking cap C1. The same can also be used in parallel with Q1 emiiter-resistor bypass cap C4. Both these will make the highs more detailed.
Your layout looks good - short signal paths, etc. I think I'll build one shortly if sounds as good as you say. It's a better alternative to a gainclone in some ways, because it can be sonically tweaked in numerous ways.
You can try the 2sc3112 for Q1 ( if you can get it), instead of the bc550(c). It has a higher beta. It is also worth trying a one-transistor Vbe servo instead of the 3-diode + resistor chain. With bc550c and suitable resistors, the dynamic impedance of the transistor servo is lower - not sure of the impact on the sonics.
You can also experiment with even lower Miller caps on Q1 - say 33 to 47 pF, as long as it remains stable. Polystyrene or Silver Mica are sonically better than ceramic at this location. If you have to use ceramic, use a 1KV or 2KV rated device.
For the DC-blocking cap in series with the speaker load, try Panasonic FC - 2200 or 2700/25 should be adequate - 63v is way overkill. 2200/63 can be used for the supply filtering instead.
Try to use metal-film resistors for everything in the signal path, especially anything connected to the base of Q1.
Try a small MKP or MKT cap (say 0.47 to 1 uF) in parallel with input DC-blocking cap C1. The same can also be used in parallel with Q1 emiiter-resistor bypass cap C4. Both these will make the highs more detailed.
Your layout looks good - short signal paths, etc. I think I'll build one shortly if sounds as good as you say. It's a better alternative to a gainclone in some ways, because it can be sonically tweaked in numerous ways.
Hello Linuxguru
Many years ago I did try to have more power using MPSA06 for input and TIP102 and TIP107 for output using 60 volt for supply but it never work. I never try find out why and I dumped it to go back to the 20 watt version.
Btw, if you tweak it show us the new schematic, maby a higher power version.
Thank
Gaetan
Many years ago I did try to have more power using MPSA06 for input and TIP102 and TIP107 for output using 60 volt for supply but it never work. I never try find out why and I dumped it to go back to the 20 watt version.
Btw, if you tweak it show us the new schematic, maby a higher power version.
Thank
Gaetan
Hi linuxguru,
This cap should be rated for 45 VDC at the minimum. 63 VDC is perfectly reasonable and a choice I might make. I would probably go with 50 VDC if there was a large price difference between the two.
Now, LF THD from the capacitor can become substantial in this application. You may find that 4,700 to 10,000 uF might be better. This has more to do with voltage across the cap than the -3 dB point.
The schematic posted by djk will sound much, much better. If you are fixed on using one bjt, then the highest hfe unit would be better. A darlington might be too slow. I wonder what a cascoded J Fet would sound like there, or a 40 VDC breakdown one (risky again).
-Chris
I'm sorry, but that statement is not true at all. HF performance of an electrolytic cap gets better with higher voltage for one. The main issue is that at lower or very high frequencies you may exceed the 25V breakdown you specified. That's nothing to say of transient turn on conditions and fault conditions.
This cap should be rated for 45 VDC at the minimum. 63 VDC is perfectly reasonable and a choice I might make. I would probably go with 50 VDC if there was a large price difference between the two.
Now, LF THD from the capacitor can become substantial in this application. You may find that 4,700 to 10,000 uF might be better. This has more to do with voltage across the cap than the -3 dB point.
The schematic posted by djk will sound much, much better. If you are fixed on using one bjt, then the highest hfe unit would be better. A darlington might be too slow. I wonder what a cascoded J Fet would sound like there, or a 40 VDC breakdown one (risky again).
-Chris
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Required Parts :
R1 10 K
R2 15 K
R3 120 K
R4 150 K Trimpot
R5 150 Ohm
R6 47 Ohm
R7 4,7 K 0,5W
R8 1 K 0,5W
R9 2,2 Ohm
C1 10uF 25V
C2 22uF 50V
C3 100uF 50V
C4 220uF 10V
C5 2200uF 63V
C6 100pF 63V
C7 470nF 63V
C8 4700uF 63V
C9 4700uF 63V
C10 470nF 63V
C11 470nF 63V
C12 2X680nF 63V
D1 1N4148
D2 1N4148
D3 1N4148
D4 FR302
D5 FR302
D6 FR302
D7 FR302
Q1 BC550C
Q2 TIP142
Q3 TIP147
Kapibara.... the nice animal you have used into your avatar
is called KapiVara here ...with a "V" in the place of "B".
Interesting.... do you have those animals in your place?
They leave in rivers, and only in South America if i am not wrong... teeth is alike Rabittz, the biggest Rabbit family member on earth..weigth goes to 100 kilograms, meat is appreciated and hairs are used to produce brushes...leather is special too....nice animals.
Anaconda and alligator are the natural enemies.
Very interesting, that you, so distant from here, use this name.
regards,
Carlos
is called KapiVara here ...with a "V" in the place of "B".
Interesting.... do you have those animals in your place?
They leave in rivers, and only in South America if i am not wrong... teeth is alike Rabittz, the biggest Rabbit family member on earth..weigth goes to 100 kilograms, meat is appreciated and hairs are used to produce brushes...leather is special too....nice animals.
Anaconda and alligator are the natural enemies.
Very interesting, that you, so distant from here, use this name.
regards,
Carlos
Attachments
Oh!...only you know the meaning or Kukubara
Your people may understand.
You see... i tried to explain....your idea is different...may be funny
regards,
Carlos
Your people may understand.
You see... i tried to explain....your idea is different...may be funny
regards,
Carlos
Re: You amplifier is fine Gaetan...sorry by the off topics...this one will be my last one
Nice small amplifier....
I have somewhere a lot of BDW93C - BDW94C, maybe I'll try to use them.
In your opinion, this amp can be used in bridged mode? I want power
and i hate big output caps 
burra is incorrect, butter in Italian is burro (male name...)😀 😀 😀
Nice small amplifier....
I have somewhere a lot of BDW93C - BDW94C, maybe I'll try to use them.
In your opinion, this amp can be used in bridged mode? I want power
and i hate big output caps destroyer X said:
burra is incorrect, butter in Italian is burro (male name...)😀 😀 😀
>In your opinion, this amp can be used in bridged mode? I want power and i hate big output caps
You'll need an external inverter (maybe a unity-gain inverting op-amp) to drive the other side of the bridge. There still will be problems with the output offset voltage, which can now be in either polarity, so you'll still need a non-polar cap or back-to-back electrolytics.
There's probably no easy way to get rid of the output electrolytic with a single-rail supply.
You'll need an external inverter (maybe a unity-gain inverting op-amp) to drive the other side of the bridge. There still will be problems with the output offset voltage, which can now be in either polarity, so you'll still need a non-polar cap or back-to-back electrolytics.
There's probably no easy way to get rid of the output electrolytic with a single-rail supply.
- Status
- Not open for further replies.