Project: "Taurus" hi-end audio amplifier (Revision A))

This is a major revision of this project here, and that is why I decided to open a new thread. Basically it is a similar amplifier, having essentially the same characteristics. However some measures were taken to prevent potential problems, making it more stable. The amplifier is now entirely AC coupled, and thus it cuts any DC bias on the input. The output uses boucherot cells and coils to stabilize the LM3875s against HF oscillations that might occur.

List of components:
C1/2/7/8 – 10mF electrolytic capacitor (50V);
C3/5/9/11 – 100µF electrolytic capacitor (50V);
C4/6/10/12/16/20 – 100nF polyester capacitor (63V);
C13/17 – 1µF polyester capacitor (63V);
C15/19 – 220pF ceramic capacitor (63V);
C15/19 – 22µF non-polarized electrolytic capacitor (6,3V);
D1-4 – 6A05 rectifier diode;
F1 – 2A slow-blow fuse;
F2/3 - 6,3A slow-blow fuse;
HS1/2 – 1,2°C/W passive heatsink;
IC1/2 – LM3875 audio power amplifier (LM3875TF);
J1 – IEC C14 connector;
J2/3 – RCA female connector;
JW – 26AWG stranded wire;
L1/2 - Air core coil, 16 turns with 6mm dia. of 22 AWG wire;
LP – 230V~ resistored neon bulb;
R1 – 10KR double gang potenciometer (10% tol., 1/8W);
R2/9 – 4,7KR carbon resistor (5% tol., 1/8W);
R3/10 – 47KR carbon resistor (5% tol., 1/8W);
R4/6/11/13 – 1KR carbon resistor (5% tol., 1/8W);
R5/12 – 33KR carbon resistor (5% tol., 1/8W);
R7/14 – 2.7R carbon resistor (5% tol., 2W);
R8/15 – 10R carbon resistor (5% tol., 1/2W);
S – DPST switch;
T – Toroidal transformer 230V~ pri. 2x18V~ sec. 225VA.

Electrical parameters:
- Va min = 207V RMS
- Va max = 253V RMS
- Ia (Va = 230V RMS, Z = 4Ohm) = 991.9mA RMS
- Ia (Va = 230V RMS, Z = 8Ohm) = 686.7mA RMS
- P (Va = 230V RMS, Z = 4Ohm) = 228.1W
- P (Va = 230V RMS, Z = 8Ohm) = 157.9W

Amplification characteristics (each channel):
- Input impedance: 3,197-10KOhm
- Input sensitivity: 500mV RMS
- Gain: 30,63dB (34V/V)
- Frequency response (-3dB): 7.234-120000Hz
- Output power (Z = 4Ohm): 45W
- Output power (Z = 8Ohm): 29W

Thanks to the guys here in AAC and those from diyAudio for all the help and advise while doing this revision. It wouldn't be possible without them.

I will translate the schematic diagram and post the translation, as soon as possible.

The schematic diagram is now available in English. Sorry for the strange new font. I didn't have the old font available on the computer I was using.

I think you are expert of Amplifiers...

RRITESH KAKKAR said:

I think you are expert of Amplifiers...

Click to expand...

I'm not an expert, but thanks.

Would it be a good idea to include an NTC inrush limiting device on the input? I imagine when you plug it in it will draw a large current and might be prone to blowing fuses.

I have a DVD player with combined surround sound here. It has 4700u + 4700u 35V caps (single supply) and uses a 24V bus. It is rated to 65W. When power is disconnected from the amp it keeps playing for about 1 second. So it makes me think the 20000u capacitance is a bit excessive for a 45W amp. Does it noticeably improve audio quality, going with such large capacitance?

There's one important metric missing, total harmonic distortion.

tom66 said:

Would it be a good idea to include an NTC inrush limiting device on the input? I imagine when you plug it in it will draw a large current and might be prone to blowing fuses.

Click to expand...

Yes, some current limiting like that would be good, but fortunately, in this case is not necessary. Slow-blow fuses are required, but they will withstand the inrush current by a large margin. Indeed I used fast fuses on the primary and sometimes they didn't blow (surprisingly).

tom66 said:

I have a DVD player with combined surround sound here. It has 4700u + 4700u 35V caps (single supply) and uses a 24V bus. It is rated to 65W. When power is disconnected from the amp it keeps playing for about 1 second. So it makes me think the 20000u capacitance is a bit excessive for a 45W amp. Does it noticeably improve audio quality, going with such large capacitance?

Click to expand...

Large caps do improve audio quality, especially for bass. If it plays for a second, don't worry. This one keeps playing for 4 to 5 seconds after being turned of (when the volume is sufficiently low). The diodes should be able to withstand the inrush current demanded by those capacitors, though. The 2 x 10000uF caps on each rail are the ideal considering the average current, so the ripple doesn't go beyond 10%. Tests revealed that those could be a bit bigger, but most of the problems come from the voltage drop on those rectifier diodes anyway. I'm considering using Schottky diodes in future versions.

sceadwian said:

There's one important metric missing, total harmonic distortion.

Click to expand...

Thanks, but unfortunately I can't measure THD, and I don't know how to estimate it. I would appreciate some help. Could you PM me?

cumesoftware said:

Thanks, but unfortunately I can't measure THD, and I don't know how to estimate it. I would appreciate some help. Could you PM me?

Click to expand...

You need a distortion analyser: http://www.barrytech.com/tektronix/tektm500/tekaa501.html

However they are quite expensive.

Consider using the method described on this page: http://www.dogstar.dantimax.dk/tubestuf/thdconv.htm

You will need an oscilloscope and signal generator, and some way to cancel the input and output signals (can be done with A+B with B inverted on some dual trace scopes, or on many digitising scopes.)

tom66 said:

...
Consider using the method described on this page: http://www.dogstar.dantimax.dk/tubestuf/thdconv.htm

You will need an oscilloscope and signal generator, and some way to cancel the input and output signals (can be done with A+B with B inverted on some dual trace scopes, or on many digitising scopes.)

Click to expand...

I actually have an o-scope, but it is a single channel one. I will try. Thanks!

cumesoftware said:

I actually have an o-scope, but it is a single channel one. I will try. Thanks!

Click to expand...

You could build a differential amplifier with a precision low distortion op-amp. Remember that its THD will affect the measurements, so get that as low as possible.

You could also consider calculating FFTs. If you have a high sampling frequency (96 - 192 kHz) sound card (usually in a desktop PC), that would be ideal. There are many free software FFT programs for sound cards.

tom66 said:

You will need an oscilloscope and signal generator, and some way to cancel the input and output signals (can be done with A+B with B inverted on some dual trace scopes, or on many digitising scopes.)

Click to expand...

Watch out for any phase shift introduced by the amplifier. Even a small phase shift will make this method unreliable.

sceadwian said:

There's one important metric missing, total harmonic distortion.

Click to expand...

The datasheet for the LM3875TF would indicate that at 29w into 8Ω, THD would be around 0.01% with a ± 35v supply. I imagine it will be somewhat higher at these voltages but very acceptable.

In any case, I spent some time as a professional technologist in the sound side of the movie business. Not as glamorous as you might think, but educational. What I did learn was, despite their fanatical search for absolute minimal THD and IMD, most audiophiles could not tell the difference between 0.01% THD and 0.00001% THD if their lives depended on it.

They waste a huge amount of money on these super-low distortion amplifiers and then play them back through speaker systems that have 10 to 100 times as much distortion. Not sane.

Like the guy that buys the Ferrari F50 and never exceeds the speed limit. As dumb as a garden boulder.

The graph in the datasheet for the LM3875 shows that its "typical" distortion is only 0.003% at 1kHz when its supply is plus and minus 35V, its load is 8 ohms and its output power is 40W to 50W.
It shows no maximum distortion which might be as high as 0.1%.

This project has a much lower supply voltage (about plus and minus 24V) so its output power is much less and its distortion might be much more.

BillO said:

...
They waste a huge amount of money on these super-low distortion amplifiers and then play them back through speaker systems that have 10 to 100 times as much distortion. Not sane.

Like the guy that buys the Ferrari F50 and never exceeds the speed limit. As dumb as a garden boulder.

Click to expand...

Indeed! The best way to "fight" distortion is to invest in good speakers (with a good linear response from at least 20Hz up to 30000KHz).

Audioguru said:

The graph in the datasheet for the LM3875 shows that its "typical" distortion is only 0.003% at 1kHz when its supply is plus and minus 35V, its load is 8 ohms and its output power is 40W to 50W.
It shows no maximum distortion which might be as high as 0.1%.

This project has a much lower supply voltage (about plus and minus 24V) so its output power is much less and its distortion might be much more.

Click to expand...

In this case the expected output power is 29W for 8 ohm loads. The distortion can be much more than 0.1% if the level of the input signal is above 500mV RMS and with the volume pot cranked up. Otherwise it will be much less than that.

Audioguru said:

The graph in the datasheet for the LM3875 shows that its "typical" distortion is only 0.003% at 1kHz when its supply is plus and minus 35V, its load is 8 ohms and its output power is 40W to 50W.
It shows no maximum distortion which might be as high as 0.1%.

This project has a much lower supply voltage (about plus and minus 24V) so its output power is much less and its distortion might be much more.

Click to expand...

I was just being conservative. Yes, it could be much worse if driven to hard. I'd be tempted to rate it at more like 6 to 8 watts (at the appropriate input level) given the 24v supply. Just to have the headroom (≈6 db) and to satisfy the spec jockeys.

But still, anything less than 0.01 % distortion would be very acceptable. At least to the ears, if not the spec sheet.

BillO said:

I'd be tempted to rate it at more like 6 to 8 watts (at the appropriate input level) given the 24V supply. Just to have the headroom (≈6 db) and to satisfy the spec jockeys.

Click to expand...

The supply is about plus and minus 23.5V which is 47V, not 24V.
The output power at clipping is about 29W into 8 ohms, not 6W or 8W.
Worry about headroom only if you use a slow VU meter instead of a peak detecting level meter.

The datasheet shows that most ICs will have very low distortion at clipping but a few will have higher distortion.

See below...( all in good nature, of course...)

Audioguru said:

The supply is about plus and minus 23.5V which is 47V, not 24V.

Click to expand...

Nomenclature, and beyond that you're splitting hairs. 23.5 vs, 24!?

Audioguru said:

The output power at clipping is about 29W into 8 ohms, not 6W or 8W.

Click to expand...

Well, yeah! I can read. I did said to leave 6db of headroom.

Audioguru said:

Worry about headroom only if you use a slow VU meter instead of a peak detecting level meter.

Click to expand...

Not so my educated friend! You cannot react to any kind of meter if you are flesh and blood. If you want live, or at least realistic dynamic range on your playback, you need to leave a little headroom. I'm not talking FM radio here, but the response required by modern motion pictures or the latest orchestral recordings (not that I can hear the range any more). 6db would be minimal

Audioguru said:

The datasheet shows that most ICs will have very low distortion at clipping but a few will have higher distortion.

Click to expand...

You must mean just before clipping, right? Once clipping starts, THD WILL go off the charts, ... and that is exactly what you want to avoid.

Audioguru said:

...
The output power at clipping is about 29W into 8 ohms, not 6W or 8W.
...

Click to expand...

Not AT clipping. 29W is the actual RMS output power for 8 ohm loads considering a total harmonic distortion below 0.1% (as the chart says). When clipping, any amplifier will show a distortion well above 1%.

BillO said:

...
Nomenclature, and beyond that you're splitting hairs. 23.5 vs, 24!?
...

Click to expand...

24V is an approximated figure. It will depend on the voltage drop across the diodes, that can vary from 0,6V to 1,2V.

BillO said:

You must mean just before clipping, right? Once clipping starts, THD WILL go off the charts, ... and that is exactly what you want to avoid.

Click to expand...

Indeed, you want to avoid clipping distortion. The slightest clipping can be noticed.