LME49810 with STD03N/P

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Sebastiaan said:

I am still surprised about the sound quality results. We compare this simple amp with many expensive respected amplifiers, and in most cases the LME49810/STD03 amp wins hands down in terms of 3D image speed and bass control. Just with a singe pair of transistors it has an awesome control like the bass is constant in a “rails”
Bas

I'm not surprised.

If you're not wasting money carving the amplifier chassis from a solid billet of aluminum, advertising, ergonomics and dealing with multiple layers of markup you can make a really good amplifier for a fraction -- but you've invested a lot of time and intellectual energy into the project.

I have now used most of the National driver chips -- they really are a breakthrough.
 
jackinnj said:


I'm not surprised.

If you're not wasting money carving the amplifier chassis from a solid billet of aluminum, advertising, ergonomics and dealing with multiple layers of markup you can make a really good amplifier for a fraction -- but you've invested a lot of time and intellectual energy into the project.

I have now used most of the National driver chips -- they really are a breakthrough.

Agree. In fact I designed a 4 layers multilayer board. Star ground for the power supply sections, one global shield and a layer for line signals. I can literally put my cellphone on top of the PCB with a phone call in the running without hearing any interference over the speakers :D

With kind regards,
Bas
 
Less is better? With 30x1000uF caps? :D
I'm joking Sebastiaan ;) Very nice design according to me!
I agree with you about simple res/cap zobel filter, a stable and optimized amp never needs a coil at the output, it's only my kind of view...
What i really like is the attention put in the 220V side and chassis/ground/switch/socket filtering.
Strange thing that 2 bias settings: with minimum bias (40/60mA) distortion had to be less according to my simulations and tests.
I'm going forward with my design where i use a simil-darlington with 15030/1 and 3281/1302 by ON. Emitters of drivers connected to opposite voltage as in Hiraga/Aloia designs so to always remain in A and avoid on-off switch and cob problems.
Current gain over 10.000 in the entire range of use, sanken darlingtons can't do that, plus the advantage of diodes not connected to the base, so i can put them among bias pins, surely more efficient way to operate for thermal stability.

What i wanted first to know, if i can ask, is about the passive bias method, infact i don't understand the use of that 470R among bias point and C5/C6, C7 because i don't need them in my design without lme49810 (at the moment).
Regards,
Gianni
 
gianxdiy said:
Less is better? With 30x1000uF caps? :D
I'm joking Sebastiaan ;) Very nice design according to me!
I agree with you about simple res/cap zobel filter, a stable and optimized amp never needs a coil at the output, it's only my kind of view...
What i really like is the attention put in the 220V side and chassis/ground/switch/socket filtering.
Strange thing that 2 bias settings: with minimum bias (40/60mA) distortion had to be less according to my simulations and tests.
I'm going forward with my design where i use a simil-darlington with 15030/1 and 3281/1302 by ON. Emitters of drivers connected to opposite voltage as in Hiraga/Aloia designs so to always remain in A and avoid on-off switch and cob problems.
Current gain over 10.000 in the entire range of use, sanken darlingtons can't do that, plus the advantage of diodes not connected to the base, so i can put them among bias pins, surely more efficient way to operate for thermal stability.

What i wanted first to know, if i can ask, is about the passive bias method, infact i don't understand the use of that 470R among bias point and C5/C6, C7 because i don't need them in my design without lme49810 (at the moment).
Regards,
Gianni

Dear Gianni,

Thanks for your response.

You are absolutely right about the Thermal trak transistors by ON. It is much more convenient the have full access over the thermal track diodes, and that is a downside of the STD03's.

But stubborn as I am, I don't want to add extra driver transistors after the LME49810's (even though the Darlington has in fact the driver in the package, but I want to use as less as PCB tracks as possible).

R5 (470R) set the current source of approx 2,5mA to bias the diodes in the STD03's

C5, C6 and C7 are to damp the current charges. The bigger C6 the tighter the bass in my experience. I steel this from Johan Ketelaar a Dutch designer for JK acoustics. As far as I remember I even saw him use values up to 1000uF!

The bias setting is something I need to experiment with. I can't measure difference in THD, but the IMD+N get better with higher bias settings. Almost twice as good on the audio precision ATS1

You have any schematics already of your design with the ON transistors? I am very curious. If I remember right Bryston use the same transistors but the version without the thermal trak diodes.

With best regards,
Bas
 
I was searching for a nice SS pre and driver stage, original Aloia was with a paralleled ecc83 in srpp configuration.
Important: I want an integrated amp with a total gain of 70 for 50w output. differentials, cascodes, ecc, tested but i don't really liked them. so this time i want to give the lme a try, even if i wanted a no-total-feedback amp in my project.
Emitters back-to-back are a really nice feature of this topology, always remaining in A and driving with brute force the 3281/1302.
I think there's no additional track compared to std, i don't even need a servo or a protection circuit because i think i will use a floating ground and so a dedicated and unregulated supply stage for each channel (only 3281/1302s). I can't do that with darlington transistors, together with many other limitations far more important than 2 tracks or more to design, i don't need bypass caps over collectors, i don't need base resistance with a hfe of 100 and over, always stable even at low currents.
Bias shunt capacitors are normally used to bypass active vbe stages and their 100% feedback current, in passive mode they make only damages to sound in my opinion. If a true symmetric signal is fed through the bias points, for example using a cap for each bias rail or with a perfect provided signal as the lme can do, there's no sound signal through the diodes and so the bypass is useless. That's what comes out of my tests and first of all, Hiraga and Aloia studies and real amps.
See my output stage as in Hiraga/Aloia great amps, it a pre-design sheet, work in progress.
Let me know, Sebastiaan and thanks for the infos until now...
 
Dear gianxdiy,

Thanks for your comments. I will give it a try and a listening session with the shunt caps out and let you know ;)

I love your ideas about a floating circuit. If you keep the signal truth differential this should be possible and give you a lot of benefits. Ground scheme's and grounding is a very critical factor for sound quality. Ground planes and ground scheme's if not designed careful will degrade the THD+N ratios. And not matter how hard you try your best, ground's are often dirty. Keep the circuit floating and no where referred to ground can give a lot of improvement in sound quality.

Brings me back to the servo. If you look at my schematic closer you can see my servo design has actual two reasons. One obvious to prevent any DC appear on the outputs, but second (and I think overseen) to create a virtual ground for the positive leg of the input. I don't want R4 to referenced to ground. The servo opamp output create a virtual ground. In the sake of a better CMNR it is good to know that the positive leg sees more or less the same impedance as the negative leg. Both look at a virtual zero output of an output stage.

If I made the amp a bridged design, and made the servo also work truth differential it would be possible to let the whole amp float above ground.

I think not famous in the USA but dutch designer Ad van Medevoort did such a tricks with his old line of amplifiers.

With kind regards,
Bas
 
Hi Bas,
I posted a pdf but i'm moderated and maybe i cant' attach... :(
Grounds can be such a beast!!! Not noted before you told me that your servo is double use. Surely very high class solution.
I will only let the final couple float, not the entire amp.
Really really uncommon the attentions you put in your grounds and in the whole amp layout, so nice!
I hope this time my pdf can be uploaded.
Regards,
Gianx
 

Attachments

  • double emitter follower 50w.pdf
    62.1 KB · Views: 346
Dear gianxdiy,

Thanks now the schematics work! :D

Lovely simple design. I think you can omit R2 since the bias current is already set by R1, R3 and your diode string.

Other members will flame me for this.. :D But in my opinion you can leave D1 and D3 out. The baker clamp of the LME49810 prevent the output stage from saturating (I test it with letting the amp clip heavily). Toose diodes induce a little bit leakage which decrease the sound quality. of course on your own risk, but in my LME49810 design I let them out. Since you don't plan to add the inductor and resistor in the output what is the function of R11?

I see you want to keep it simple, but why not replace R6 and R7 for current sources? Maybe cascode current source? :D

With best regards,
Bas
 
Dear Bas,
too much hurry posting my schematic :D:D:D
I thought R2 would set output impedance of lme, as read in its specific post here in diyaudio, maybe i misunderstood...
Consider that my pdf comes from a totally different voltage gain stage, it also explains D1 and D3.
In effect, baker clamp could do a better work, i will leave them out with the lme, thanks.
R6 and R7 work as reference to opposite rail to never let the 15030/15031 switch off, no need for current source, i could even remove them and try a real discrete darlington. Up to date, they are here, wasting a bit of power (3w resistors) to bias in class A...
First design is infact a composite of 3 emitter followers, with predriver and driver in class A (with resistors in back to back configuration) and final stage in class B/AB (60mA), all driven by a tube srpp which provides all voltage gain.
The current gain is enormous, with great linearity and soft clipping.
What i'm trying to do is to arrange this to be driven by the lme and so removing first follower, the lme can almost provide 50mA by itself, right?

P.S.
R11? Multisim thinks it's a loudspeaker!!! :D:D:D:D:D

Regards,
Gianx
 
Hi Jackinnj,
i use multisim only for preliminar tests, i save a lot of money but i don't blindly trust in simulations.
So R11 is there for simulation only, sometimes adding a cap to it to test the amp stability. We all know that printed circuits, real components and many other things can affect the quality of our projects...
I saw your pcb's, they are very pro!!!
Regards,
Gianni
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.