New Doug Self pre-amp design...

In an attempt to retribute Mooly's help, I upload here the Tone Controls but using NE5532 and I've added several electrolityc capacitors. Other than that, the schematic is the same as Mooly has uploaded.

The file NE5532.txt should be located in the same folder as the .asc schematic.
 

Attachments

  • DSelfToneControls.zip
    2.7 KB · Views: 214
If anyone is still worried about this issue as regards my preamp, permit me to quote from Small Signal Audio Design, second edition, p367:

"Going back to the loading on the inverting opamps at low volume, we have 383 Ω at Mk 1, which is 766 Ω per opamp and no cause for alarm. However, I have heard doubts expressed about a possible rise in distortion at very low volume settings below this, because the inverting amplifiers U1:B, U2:A then see even lower load impedances. The impedances may be low, but the current to be absorbed by the inverting stages is actually very limited because almost the whole of the pot track is in series with the input at low settings. To prove there is not a problem here, I set the volume to Mk 1 and pumped 20 Vrms in, getting 1.6 Vrms out. The THD residual was indistinguishable from the GenMon output of the AP SYS-2702. In use the input cannot exceed 10 Vrms as it comes from an opamp.

To push things further, I set the volume to Mk 0.2, (ie only 2% off the endstop) and shoved 20 Vrms in to get only 300 mVrms out. The THD+N residual was 0.0007%, composed entirely of noise with no trace of distortion. I then replaced the 4562s in the U1:B, U2:A positions with Texas 5532s (often considered the worst make for distortion) and the results were just the same except the noise level was a bit higher giving a THD+N of 0.0008%. There is not a problem here."
 
One day (winter) on my list of audio diy projects, is to make/design the active volume ckt (basically the Elektor 2012 pre-amp) including the new var freq tone control ckt (why did Elektor, not use this?), using motorized pots such as Bourns PRM16, relays, etc.
Mother, me being next in line :) has to get up on her walker to adjust/turn off the old Pioneer SX-950, this just not right :) The new 50" Panasonic flat screen has lousy sound/volume.
If I ever manage to do it, I'll post it here for everyone to build/use.
We are getting close to be able to have affordable colour graphic LCD, (FT800(EVE), AVR xmega (Arduino) to run/control it, from an Android &/or a IR remote device too.
It blows my mind that we are still using ne5534a and this was used in my second pre-amp design back in college,1979. I used a ne5533 that never caught on for some reason. A monumental step from the previous LM381A design.
So I'll pair up some ne5534a with my si4770 radio design. :)

Can someone answer a nagging ? I have, why do the majority of today's new bjt opamp's have higher current noise than the old ne5534a?

Cheers
Rick
 
Last edited:
One day (winter) on my list of audio diy projects, is to make/design the active volume ckt (basically the Elektor 2012 pre-amp) including the new var freq tone control ckt (why did Elektor, not use this?),
?

They didn't design it, I did, and as the emphasis was on very low noise, a simple Baxandall tone-control was more appropriate. I also wanted to show off the new split-drive tone-control technology I had thought up.


Can someone answer why the majority of todays new bjt opamps have higher current noise than the old ne5534?
Presumably because they are optimised for low voltage noise, which is almost always what counts. An MM cartridge input is the notable exception, and the 5534 matches that impedance much better than, say, an LM4562 with its higher current noise.
 
Wow, did not expect such a fast answer :)
I have to get/read/test the LA article to see what the noise spec's are on the new var freq tone control. It is a great ckt to use. I do not think that one needs a parametric/SV eq/filter, although I have one that I put into the old pre-amp.

More ?'s, if you not mind.

1) why do you stay away from a jfet first stage in the design for a MM cartridge?
The issue, that was raised, something about an bjt opamp has varying/non-linear i/p cap, against a high Z source, causing IMD.
It was written about this in a few places, one being the Amber 3501 THD analyzer that I have. They used a 2sk146 dual.
Bogus :)
I'll post the circuit, I'd like you thoughts on it, a mess for me to figure out :)
2) how can you tell from bjt data sheet what the current vs voltage noise, I assume you have to look at those NF curves plotting source R :)

Cheers
 

Attachments

  • Amber3501_input atten&balance ckt.pdf
    249.3 KB · Views: 179
1) why do you stay away from a jfet first stage in the design for a MM cartridge?
It is very difficult to improve significantly on the noise performance of the 5534A with an MM cartridge. A J310 FET would in theory give an improvement of 1.7 dB, but hybrid discrete-opamp circuits can be rather tricky. I have deployed thousands of them but I am still wary of issues with HF stability.

The issue, that was raised, something about an bjt opamp has varying/non-linear i/p cap, against a high Z source, causing IMD.
Other way round. JFET opamps have the capacitance to substrate problem.
BJTs are not free of CM distortion but there is usually less and the mechanism is quite different.

It was written about this in a few places, one being the Amber 3501 THD analyzer that I have. They used a 2sk146 dual.
Bogus :)
I'll post the circuit, I'd like you thoughts on it, a mess for me to figure out :)
Pretty standard instrumentation amplifier made from two JFET-opamp hybrid circuits.

2) how can you tell from bjt data sheet what the current vs voltage noise, I assume you have to look at those NF curves plotting source R :)
Cheers[/QUOTE]

Yes. You rarely get plots of vn and in vs Ic on data sheets. I don't know why.
 
Specifications
Test conditions: supply voltage ±17.6 V; all measurements symmetrical; tone control defeat disabled.
Test equipment: Audio Precision Two Cascade Plus 2722 Dual Domain (@Elektor Labs)
THD+N (200 mV in, 1 V out) 0.0015% (1 kHz, B = 22 Hz – 22 kHz)
0.0028% (20 kHz, B = 22 Hz – 80 kHz)
THD+N (2 V in, 1 V out) 0.0003% (1 kHz, B = 22 Hz – 22 kHz)
0.0009% (20 kHz, B = 22 Hz – 80 kHz)
S/N (200 mV in) 96 dB (B = 22 Hz – 22 kHz)
98.7 dBA
Bandwidth 0.2 Hz – 300 kHz
Max. output voltage (200 mV in) 1.3 V
Balance +3.6 dB to –6.3 dB
Tone control ±8 dB (100 Hz)
±8.5 dB (10 kHz)
Crosstalk R to L –98 dB (1 kHz)
–74 dB (20 kHz)
L to R –102 dB (1 kHz)
–80 dB (20 kHz)


Apparently absent, possibly the most important spec: source impedance.

Max output voltage on the low side compared to inexpensive competition.
 
AP2700 Source Impedances
Balanced 40 ohms, 150 ohms or 600 ohms.
Unbalanced 20 ohms or 600 ohms.

Good enough for test equipment, but totally inadequate as a general purpose headphone amplifier. But this device was never designed to be a headphone amp!

Even the headphone amps in $39.95-69.95 digital players and headphone amps have source impedances that are down around 1 ohm or less over the audio band and actually even down to DC (no coupling capacitors). They have low distortion while driving 16 ohm loads with from 0.75 to 3 volts, which covers the needs of about 99% or more of all headphones and earphones.
 
Hi Guys

Most power amps have a voltage gain of about 28.3 to be "theatre compatible", yielding 100W into 8R with 1V input. For just one watt of output, the PA puts out 2.83V and needs only 100mV input.

For my own system, 1W per speaker produces 90dB of sound from each which is a loudness I can't bear. My usual listening levels are in the 60-70dB range if I want it loud. For this, the input voltage to a typical PA is only 10mV or so.

What kind of THD does the Self preamp have at these lower outputs?

At this point in history, there are a few commercial preamps that have THD unmeasurable by an AP. I would think that ALL preamps should meet this performance. The ones I've built sim at 10s of ppb and are all discrete. I cannot measure their THD with the equipment I have access to. I believe that electronic distortions are wholly unnatural inasmuch as we did not evolve having to deal with it - so even unimaginably small quantities are "audible". Our other senses all work to ppm and ppb - why not our hearing?

IC opamps tend to all fall into the same sonic group that Self's PAs fall into, being mostly based on the same circuit and using the same type of compensation. I think that is the problem. I was disappointed to see such high THD figures from the latest version of his preamp, but just the same it is not surprising considering the active devices chosen. Doug is very good at what he does, but what he does falls into a narrow category.

Have fun
 
At this point in history, there are a few commercial preamps that have THD unmeasurable by an AP. I would think that ALL preamps should meet this performance. The ones I've built sim at 10s of ppb and are all discrete. I cannot measure their THD with the equipment I have access to. I believe that electronic distortions are wholly unnatural inasmuch as we did not evolve having to deal with it - so even unimaginably small quantities are "audible". Our other senses all work to ppm and ppb - why not our hearing?

Because of the raft of evidence that our ears aren't all that sensitive and that fact even used to keep us alive when tigers prowled our back yards.

I guess you never did any proper listening tests that are highly convincing of the fact that it doesn't take really low distortion (related to what can be easily measured) to have sonic transparency or if you will performance that is indistinguishable from a plain old straight short piece of relatively pure copper wire.

Thus, may folks are condemned to unendingly chase numbers for the sake of numbers as we see above. Lots of people in that boat, but it doesn't make it right.

Some of them like it that way, and the First and Second worlds are rich enough that if not too many people waste their time that way, society as a whole does not suffer that much. Party on, dude!
 
I added a 10r resistor to my headphone amp output.
Right next to the buffer IC output pin.

That's good enough for legacy headphones with impedances >100R.

Many modern headphones are 16 ohms or even less nominal, and varying lower over the audio band. I hear tell of headphones that go down to 5R at some frequencies.

If you calculate or measure the FR differences with a 10R source impedance and one of them, you probably shouldn't be all that comfortable.
 
Hi Guys

Arnyk - I've been listening all my life and have intact hearing. I fall into the group who prefers equipment that truly has lower THD and IM. However, I am not like some people I know who cannot enjoy a piece of music without knowing everything about the system. I DO believe that our hearing is sensitive to much tinier distortions than are discussed here. That we can tolerate copious amounts of distortion is actually something we DID evolve to do. Sounds in nature are rarely 100% fidelity, due to reflections, wind, materials between us and the source absorbing some of the energy at different frequencies - yet we can still distinguish the voice of a friend, say, or a certain bird. We needed this for survival. However, with much of the modern electronic processed sound our brains have to work overtime to sort out the missing bits and outright corruption of the sound we expected. But that is a whole other issue...

I am usually pretty happy when I can't measure the distortion of what I build. Of course, over the years test equipment has gotten better but so has the components that go into amps, and the knowledge base of what goes on and what can be improved has also advanced. So, what i am building has gotten much better over the years and I am again at the point where I can't measure its deviations with what test equipment is currently available.

Regarding headphones: Back in the day, most affordable headphones were the same impedances as speakers, typically 8R or so. The trend has been towards higher impedances with headphones to accommodate battery-powered portables and the limitations of opamps, with high-end types bucking that trend. Audeze has a range of units from 20R to 110R. Which is "best: from them is difficult to determine from their marketing descriptions.

To my ear, and for my aesthetics of which designs I like, fully symmetric circuits sound best. VFA vs CFA is neither here nor there if the circuit is designed properly and compensated well. - the differences between the types is negligible at the limits of performance anyway. I feel that low phase delay through the pass band is essential, just as Marshall Leach suggested and it may be slightly easier to do this with symmetric rather than asymmetric circuits.

I do not feel pained or inconvenienced by chasing lower THD and I fully enjoy the music at each step along way.

Have fun
 
Last edited:
Regarding headphones: Back in the day, most affordable headphones were the same impedances as speakers, typically 8R or so. The trend has been towards higher impedances with headphones to accommodate battery-powered portables and the limitations of opamps, with high-end types bucking that trend.
It actually isn't as simple as that. Crap-quality headphones back in the day used to be 8 ohms most of the time, true enough (I suppose because more often than not, they tended to use speaker drivers). Then in the 1970s, the first microphone-derived driver designs appeared, most notably the super popular Sennheiser HD414 at 2 kOhms nominal (no typo, I still have some HD424s like that floating around). Around 1980, 600 ohms had gotten pretty common on this side of the pond (K240, HD420/430, DT880/990 etc.). With portable devices becoming popular, manufacturers started offering matching headphones, with 32 ohms being typical even back in the 1980s (look at the trusty PortaPro). Sennheiser's mass-market all-round hi-fi models went to 300 and later 150 ohms in the late '80s to mid '90s, with the ~2000 lineup containing models with 32 to 100 ohms, while the high-end ones remained at 300 ohms. And that's more or less where things have remained.

Impedances for in-ear drivers tend to be lower on average. Regardless of whether they are conventional dynamic types or balanced armature, their voice coils are tiny, so you typically don't see more than a few dozen ohms (the highest I've seen would have to be the Yuin PK1 at 150 ohms). Multi-driver BA models may have significant impedance variation due to crossover construction constraints (RC only, Ls would be much too big). The good ol' triple.fi 10 Pro, known as being particularly critical, has a response that rises to over 60 and dips down to about 6.5 ohms. By contrast, the typical dynamic in-ear has a virtually flat impedance response.
 
........................

Many modern headphones are 16 ohms or even less nominal, and varying lower over the audio band. I hear tell of headphones that go down to 5R at some frequencies.

If you calculate or measure the FR differences with a 10R source impedance and one of them, you probably shouldn't be all that comfortable.

................. The good ol' triple.fi 10 Pro, known as being particularly critical, has a response that rises to over 60 and dips down to about 6.5 ohms. By contrast, the typical dynamic in-ear has a virtually flat impedance response.
Could you two talk to each other and agree what we need to hear?
 
Hi Guys

It can be safely said that the impedance of a headphone element is no indication of its quality.

The very first headphones that I know of were those used by telegraphers and later for telephony. These were 600R or so and were quite dreadful. Good enough for speech and for hearing dots and dashes.

Today you see good and bad examples of quality for every impedance.

Damping of the headphone element is as important as damping of a full speaker, in my view. It is pretty much dismissed in most designs since most headphone amplifiers are thought of more as a preamp stage that can drive a tiny speaker, rather than thinking of it as a small power amplifier. The latter approach puts one in a different mind set and the circuits will be a bit different - more like a "real" PA. Then all of the methods used to reduce THD and improve performance that apply to a PA can be applied to the headphone driver with similar results.

The varying headphone element impedance is no different than the varying impedance of a full loudspeaker and should managed in the same way.

As an example, in a drive circuit I designed for the Audezes, doubling the output stage cut high-impedance THD to one-half but cut low-impedance THD to one-third, so I went for this option. These reductions of THD were at crazy high signal levels (up to 10W peak) and both the single and double output stages sim at all-zeroes (less than 10ppb) in LTspice at moderately crazy to normal listening levels. I assume that the THD reduction holds to some extent at the levels below the simulator's resolution - possibly incorrect but if it is true I'd rather stuff the extra parts and have the better performance.

Have fun
 
Last edited: