DIY 2 channel preamp

[Art Vandelay]

In this day and age where 2 channel analog is fading, I decided to design and build a 2 channel solid state reference preamp, within a budget of around $1000.

I have several 2 channel preamps as it is, including an old Krell KSP-7B and a newer Marantz SC11S, but when I listen to the Bricasti DAC direct into the power amplifier vs via either preamp I hear a more pure and resolved sound, indicating that the preamps are both adding and subtracting from the original signal in their own ways.

So my objective was to (attempt) to build a neutral preamp that sounds like the conceptual ideal straight wire with gain.

Such a design requires fundamentals such as quality switching & ladder attenuator, and a big dual mono linear power supply.


For the input attenuator I decided to use a 2 channel DACT module, and the switching is done with double reed relays per input, with left and right channels on separate pcb's.

The power transformers are housed in a separate enclosure, which also includes dual choke filters, a bank of dual filter caps, and a discrete linear regulator that sets the rail voltages to +/- 25 VDC.

The rest of the power supply is housed in the main amplifier case and contains dual banks of Elna cerafine caps followed by a combination of solid state series and shunt regulator stages, with separate regulators for the amplifier input and output stages. Rail volts are +/- 20VDC.

The amplifier modules are fairly simple, using precision Linear System dual pnp / npn transistors for the input LTP and current mirrors, fed by discrete temperature compensated current sources.

The transimpedance or VAS is a high beta 2 transistor stage fed by a discrete current source, and the output stage is single ended with bias set by a current sink with selectable high / low bias (100 / 200mA) from the front panel. Feedback and compensation are my own design, and the preamp bandwidth is DC - 1.6MHz, with capability to drive a 10 ohm load to full output and with unconditional stability and THD / IMD < 0.002%, 20 Hz - 20kHz.

I've included some screen shots below of distortion and square wave performance, and also included the same tests on the Krell preamp for reference. Note, the distortion was tested at 6kHz, 2vrms output and driving a difficult 100 ohm load.

I'm very happy to say that every objective was 100% met. Both objectively and subjectively, this is now my reference 2 channel preamp, and the closest thing to a straight wire with gain that I've heard.

 

[slipperybidness]

In this day and age where 2 channel analog is fading, I decided to design and build a 2 channel solid state reference preamp, within a budget of around $1000.

I have several 2 channel preamps as it is, including an old Krell KSP-7B and a newer Marantz SC11S, but when I listen to the Bricasti DAC direct into the power amplifier vs via either preamp I hear a more pure and resolved sound, indicating that the preamps are both adding and subtracting from the original signal in their own ways.

So my objective was to (attempt) to build a neutral preamp that sounds like the conceptual ideal straight wire with gain.

Such a design requires fundamentals such as quality switching & ladder attenuator, and a big dual mono linear power supply.


For the input attenuator I decided to use a 2 channel DACT module, and the switching is done with double reed relays per input, with left and right channels on separate pcb's.

The power transformers are housed in a separate enclosure, which also includes dual choke filters, a bank of dual filter caps, and a discrete linear regulator that sets the rail voltages to +/- 25 VDC.

The rest of the power supply is housed in the main amplifier case and contains dual banks of Elna cerafine caps followed by a combination of solid state series and shunt regulator stages, with separate regulators for the amplifier input and output stages. Rail volts are +/- 20VDC.

The amplifier modules are fairly simple, using precision Linear System dual pnp / npn transistors for the input LTP and current mirrors, fed by discrete temperature compensated current sources.

The transimpedance or VAS is a high beta 2 transistor stage fed by a discrete current source, and the output stage is single ended with bias set by a current sink with selectable high / low bias (100 / 200mA) from the front panel. Feedback and compensation are my own design, and the preamp bandwidth is DC - 1.6MHz, with capability to drive a 10 ohm load to full output and with unconditional stability and THD / IMD < 0.002%, 20 Hz - 20kHz.

I've included some screen shots below of distortion and square wave performance, and also included the same tests on the Krell preamp for reference. Note, the distortion was tested at 6kHz, 2vrms output and driving a difficult 100 ohm load.

I'm very happy to say that every objective was 100% met. Both objectively and subjectively, this is now my reference 2 channel preamp, and the closest thing to a straight wire with gain that I've heard.

What is your background and education?

Why are you running the rail voltages that high? That should be way higher voltage than you need for a pre-amp output, right? I guess you didn't mention what voltage you are regulating it down to?

Could you give more and better pics of the internals?

Did you etch your own PCBs? Pre-fabbed? Sent out Eagle files?

What is your S/N? For a pre-amp, that is a more important metric than the THD+N.

Very large contributors for the final performance are proper grounding and grounding planes, as well as the layout of the PCB traces. Just as important, possibly more important than the component selection.

Looks nice! We could use some more electronics DIY on here!

 

[Art Vandelay]

What is your background and education?

Why are you running the rail voltages that high? That should be way higher voltage than you need for a pre-amp output, right? I guess you didn't mention what voltage you are regulating it down to?

Could you give more and better pics of the internals?

Did you etch your own PCBs? Pre-fabbed? Sent out Eagle files?

What is your S/N? For a pre-amp, that is a more important metric than the THD+N.

Very large contributors for the final performance are proper grounding and grounding planes, as well as the layout of the PCB traces. Just as important, possibly more important than the component selection.

Looks nice! We could use some more electronics DIY on here!

I'm keeping the schematic a bit guarded in case it evolves into a commercial production, but to answer the questions, the main pcb employs star earthing as well as an RF ground plane, with the layout optimised to cancel fields where possible. Actually, without all of the above it would be difficult to achieve unconditional stability with a 1.6MHz bandwidth, so yes, all of that was definitely necessary.

The pcb's were etched by me, so this is really a prototype, but if does evolve into something commercial I'll be combining the two power supply boards with the amp pcb, so there's just one multi layer pcb per channel. Actually, I've already done that with a view to retrofitting existing pcb's in the near future. For pcb design I use Target 3001, and for simulations, like most DIY'ers, I use LT Spice.

The rail voltages are a bit higher than necessary (+/- 21vdc), which does result in some extra heat dissipation in the output stage, but it's very much a design decision that helps to reduce high frequency distortion by keeping the bjt junction capacitance relatively low, which minimises Miller /Early related non-linear distortions.

As far as S/N goes, it's a low noise solid state preamp with a voltage gain of 4, so it's on par with most commercial low noise preamplifiers. Noise figure is mostly determined by the LTP transistors, bias current, and the small amount of emitter degeneration required for DC and high frequency stability. I could have made the noise floor slightly lower but it would be at the expense of some other parameter, such as slew rate or high frequency stability etc.

Fwiw, my background is engineering, but professionally mostly RF and wireless network engineering. However, audio has been a hobby for me since I was a teen, so I've always been interested in audio technology and engineering and over the years I've built and modified several amplifiers, speakers and DAC's.

 

[slipperybidness]

I'm keeping the schematic a bit guarded in case it evolves into a commercial production, but to answer the questions, the main pcb employs star earthing as well as an RF ground plane, with the layout optimised to cancel fields where possible. Actually, without all of the above it would be difficult to achieve unconditional stability with a 1.6MHz bandwidth, so yes, all of that was definitely necessary.

The pcb's were etched by me, so this is really a prototype, but if does evolve into something commercial I'll be combining the two power supply boards with the amp pcb, so there's just one multi layer pcb per channel.

The rail voltages are a bit higher than necessary (+/- 21vdc), which does result in some extra heat dissipation in the output stage, but it's very much a design decision that helps to reduce high frequency distortion by keeping the bjt junction capacitance relatively low, which minimises Miller /Early related non-linear distortions.

As far as S/N goes, it's a low noise solid state preamp with a voltage gain of 4, so it's on par with most commercial low noise preamplifiers. Noise figure is mostly determined by the LTP transistors, bias current, and the small amount of emitter degeneration required for DC and high frequency stability. I could have made the noise floor slightly lower but it would be at the expense of some other parameter, such as slew rate or high frequency stability etc.

Fwiw, my background is engineering, but professionally mostly RF and wireless network engineering. However, audio has been a hobby for me since I was a teen, so I've always been interested in audio technology and engineering and over the years I've built and modified several amplifiers, speakers and DAC's.

Good Stuff!!!

Good to see a hard-core engineer posting some DIY work! There isn't much traction on AH for this type of work. I would guess that you are more involved with some of the other DIY-focused forums?

Very good to see your attention to grounding and ground planes. From the reading and research that I have done, these items tend to be very critical for end performance. Honestly, I'm nowhere near understanding how to actually design the boards to achieve this. Is it a lot of trial and error? Some software that gives you a head start? A lot of experience and intuition?

Same story for me with the higher (than I would expect) rail voltages and BJT capacitance, I have honestly never even considered that, and I don't recall ever reading about that anywhere.

How did you etch the PCBs? Did you do the "laserjet ink transfer" tecnique? Or some other more manual technique? Etchning PCBs--another task that I have never done before. If I were to get into that, I would likely try to go with the "photolithography mask technique".

And, are you following the philosophy of any particular designer or from a particular book? I have a couple books from G Randy Sloan and Bob Cordell, we actually used a couple chapters out of Cordell's book for my Solid State Class at the community college. And, of course I am familiar with Douglas Self, but I don't own any of his books.

Edit: FYI, here is the current pinnacle of my work, a DIY TI4780 Chip Amp with a beast of a P/S

I decided that I would take my education for audio electronics in a step by step approach. For this project, my goal was to choose all of my own components based upon the TI4780 spec sheet and use the recommended circuit from the data sheet as a starting point, set my gain to 29dB, etc. Furthermore, I started my real hard-core education and design at the power supply.

So, at this point, I am extremely familiar with the needs of power supplies for amplifiers and pre-amps, and how to design the P/S to achieve the goals of the amp or pre-amp.

Dunno why some of my pics don't work anymore:
TI4780 Chip Amp (My first real design as far as choosing components, setting the gain with feedback loop, complete design of the power supply, etc)
http://forums.audioholics.com/forums/threads/new-chipamp-project-thread.92513/

 

[<eargiant]

Fantastic, I love it!!! You're definitely on to something, I have no doubt it sounds fantastic.

Some may scoff at a design like that as overkill but I think you're spot on. Much of what you strove for in your pursuit of the proverbial wire with gain reminded me of the design goals of my Sansui AU-X1.

That whole front section is pre-amp. What I love most about it is that while it's already "direct" (no tone controls), you can further attenuate the gain of the flat amp down -14db (not sure what the final gain is). At that point, you're in the high SNR (125dB) bliss zone. Clean, clean, clean sound. As you can see from the picture it has a big toroidal for the output stage and a separate EI for the pre-amp and plenty of filter cap reserves.

The internals of my AU-X1

Of course, the X1 wouldn't be an X1 without the 12 unobtanium Sanken Linear High Speed NM-LAPT output transistors. I love the sound of this amp so much that when the opportunity arose to buy a complete NOS spare set from a reputable individual (and master technician), I didn't hesitate (just in case). Yes, there are 12 of these high speed outputs in the amp section.

My spare set of Sanken high speed output transistors

This beast can run an square wave at 60kHz. The speed, bandwidth and high SNR is what I believe make it sound so amazing. As you know, once you hear something designed like that, with that goal in mind, with no holds barred, it's a revelation.

This square wave of an AU-X1 (post-restoration) was posted by a master technician know as Kale on various audio and DIY forums. To quote him "That is square signal frequency of 60kHz taken from speakers output with dummy 8 ohm resistor. You see, it is stable and very fast."


I wish you luck and hope you can bring your pre-amp to production, I'm sure it'll impress. Nice work!

 

[TLS Guy]

That's very nice work. I particularly like your ground plane. Star grounding is my favored approach. In a lot of commercial products the design of the ground plane seems to be an afterthought and may be that is putting it too strongly, as I suspect there is often no thought to this crucial aspect of design.

What was your design for the volume control circuit?

I would not be too worried about someone pinching your design. People who are any good will favor their own approach and the rest won't be able to implement it anyway.

There is a lot of good stuff in the public domain that no one takes any notice of. Top of that list would be Peter Walker's current dumping amps.

If you do go into production, publish a good service manual and make the circuit freely available. If you should run into trouble after production you will find you have plenty of free help.

The more you make your design widely available the better it will be for you and all concerned.

That is the way it used to be and those were much better times.

Back in the days of the "Golden age of British Audio" the competitors were frequently collaborating to solve problems and everyone gained.

What we have now are products that go wrong, and can't be easily serviced and everybody is mad, and rightly so.

There is no substitute for peer review in getting a design right.

 

[slipperybidness]

That's very nice work. I particularly like your ground plane. Star grounding is my favored approach. In a lot of commercial products the design of the ground plane seems to be an afterthought and may be that is putting it too strongly, as I suspect there is often no thought to this crucial aspect of design.

What was your design for the volume control circuit?

I would not be too worried about someone pinching your design. People who are any good will favor their own approach and the rest won't be able to implement it anyway.

There is a lot of good stuff in the public domain that no one takes any notice of. Top of that list would be Peter Walker's current dumping amps.

If you do go into production, publish a good service manual and make the circuit freely available. If you should run into trouble after production you will find you have plenty of free help.

The more you make your design widely available the better it will be for you and all concerned.

That is the way it used to be and those were much better times.

Back in the days of the "Golden age of British Audio" the competitors were frequently collaborating to solve problems and everyone gained.

What we have now are products that go wrong, and can't be easily serviced and everybody is mad, and rightly so.

There is no substitute for peer review in getting a design right.

OP hasn't responded to at least one of my questions yet, so I'll ask the same to you.

What is the "secret formula" when designing, for grounding and ground planes? How do you start, how do you get it right? How do you know when you've achieved your goal?

This is an area where my education is currently a bit lacking! I have read plenty of technical papers that indicate the importance of this for designs and final performance. But, it's not clear to me how you start on that, and how you improve upon the grounding and trace layout for optimum performance????

 

[MrBoat]

I can't even imagine the things that must keep you folks awake at night. Me personally, it's a certain blonde woman I have kept company on and off with for the last 18 years with a penchant for tight black dresses and heels, that reminds me of just how crude I really am.

 

[Art Vandelay]

OP hasn't responded to at least one of my questions yet, so I'll ask the same to you.

What is the "secret formula" when designing, for grounding and ground planes? How do you start, how do you get it right? How do you know when you've achieved your goal?

Ground planes are very commonly used at RF frequencies, and the merit of a ground plane is to provide an ideal ground reference point for ideal power supply decoupling. An ideal ground has zero resistance and zero inductance.

Personally, I like to combine the merits of star earthing and ground planes, so I use both.