A preamp with tone controls

[tcpip]

I am trying to build a preamp with tone controls, based on the circuits in Randy Slone's "Audiophile Projects Sourcebook." I've combined the circuit for his basic gain stage + loudness + volume control, with his circuit for a three-band tone control, and inserted a switch for bypassing the tone controls.

Even though the essential design is in most respects a straight copy from his book, I have many questions about the actual implementation. And those are the questions I'll list below, but first I thought I'd include the schematic and give some explanatory notes.

Tarun

[tcpip]

A lot of the explanations may be unnecessary if you have read Randy Slone's book from which I've got the schematic. Those of you who are beginners to audio amps like me, and who want to build a good preamp, may like his book. I know I learned a great deal from it.

Now for the notes:

• The preamp uses two opamps per channel, one for the gain stage, and one for the tone controls. I've used one dual-opamp chip to implement the gain stages for both channels, and the second dual-opamp chip for both tone controls. (I know the crosstalk will probably be a bit higher this way, but I wanted easier PCB layout. I'm a newbie, and my free Eagle software can do only very small PCBs.) The diagram in the earlier post is only for the left channel.
• The switch LSW-L is "Loudness Switch - Left." There is an identical "LSW-R" on the right channel.
• The "VOL-L" pot is the volume pot for the left channel. I am intending to use a linear volume pot, followed by a smaller-value balance pot, to get approximately log taper, as per the idea given by Rod Elliott on his Website. His approach to connecting a balance pot is to use a mono pot with the wiper connected to ground, not a contra-rotating dual-ganged stereo pot.
• IC1B is a non-inverting gain stage. Its output goes out via R8 (100 Ohms) to the output terminal (OUT-L). It also feeds the non-inv input of the tone controls.
• TREBL-L, MID-L, and BASS-L are three tone control pots, each 100K, in the feedback loop of the second opamp, labelled IC2B. I hope I have not made any mistakes in copying the schematic of this stage. The output goes out at TONEOUT-L.
• There are a lot of solder-pads, e.g. VSL1, VSL2, VSL3, LSL1, LSL2, LSL3, etc. All these are meant to solder the shield of shielded hookup wire which I'll use to connect the various switches and pots; all these pads connect to ground.
• The input selector stage is not shown in the schematic, because it'll be implemented using a rotary switch and point-to-point wiring, and the selected input will feed into the schematic at INPUT-L. Similarly, there's a final switch at the output, which will select between TONEOUT-L/R and OUT-L/R. That switch too is not shown here.
• I have put one pair of caps for supply decoupling when the supply rails enter the PCB, and two pairs of caps for each opamp chip.

Hope all this makes sense. Comments please?

Tarun

[tcpip]

The PCB layout that I've done so far barely fits everything onto one PCB for both channels. (Remember, I use Eagle's free version, where PCBs cannot exceed 3.2"x4".) It's a single-sided PCB; all red tracks will be executed using jumpers.

• It's a single-sided PCB. The top half handles the entire work for the left channel, and the bottom half handles the right.
• It expects clean regulated 15-0-15 (or thereabouts) supply rails. The PSU circuit is not on this PCB.
• There is no separation between power supply ground and signal ground... there's no star grounding. Can't see how I can do it without increasing board size or adding half a dozen or more jumpers.
• All switches and pots are assumed to be panel-mounted, i.e. off-board. Hence, shielded hookup wires are expected to be used for all that wiring. The shields of such hookup wires will need to be grounded. Solder pads have been provided for the worst-case scenario for such wires, i.e. assuming that all your hookup wires are just 1 core+ shield. In reality, it is likely that you'll get 2-cores + shield or 3-cores + shield, in which case you won't need to use all those shield solder-pads. They can be left unused.

How does this look? Will it work? Comments invited. This is my first preamp PCB, so please don't lose patience with me.

Tarun

[Bill Fitzpatrick]

Personally, I would make every effort to come up with a layout that would allow for separate signal and power grounds.

[tcpip]

Originally posted by Bill Fitzpatrick
Personally, I would make every effort to come up with a layout that would allow for separate signal and power grounds.

Ouch. I was dreading someone would say this. But I guess I'd better struggle now than get all sorts of problems when I build it.

Tell me, do the shield grounds for the shielded hookup cables qualify as "power" or "signal" ground? I really can't figure this out.

Thanks for the feedback. Anything else you think I should fix?

Tarun

[Bill Fitzpatrick]

The shield grounds are signal grounds.

[ashwin]

Hi Tarun,
it looks like a small section of the GND trace is going in a complete circle (see the top left corner of the PCB, basically encompassing C1, R3 and Vol1). I''m not sure if this is a good idea, the loop may pick up noise.

Also, I'm a bit surprised that you didn't simply mirror the pattern for one channel to get two. That's what I did for my phono preamp with no problems. Any reason why you didn't do this?

Also, you might want to put slightly bigger values in parallel with C17/C18 for the power supply entry point, maybe 10uF/100uF ? (not sure if this is really required, it's just something I do for my PCBs).

- Ashwin

[runebivrin]

Near as I can tell, this circuit has no power ground beyond what goes to the filter caps. There's no low impedance return from a loudspeaker to consider, so currents in the ground should be insignificant.

Rune

[tcpip]

Originally posted by ashwin
it looks like a small section of the GND trace is going in a complete circle (see the top left corner of the PCB, basically encompassing C1, R3 and Vol1). I''m not sure if this is a good idea...

You're damn right! I fixed it, attached below.

Also, I'm a bit surprised that you didn't simply mirror the pattern for one channel to get two. That's what I did for my phono preamp with no problems.

I didn't even know such a thing was possible. It was only during this PCB design that I fully understood how to mark groups and move/cut/paste them. For the schematic layout, I did one channel and then did a group cut/paste to get the other channel, and thought myself totally brilliant. Then I used gateswap to get the unnecessary IC opamps out of the schematic and replace them with spare opamps of the earlier ICs, and considered myself positively Einsteinian. Now you tell me about mirroring of PCB? Sigh...

Also, you might want to put slightly bigger values in parallel with C17/C18 for the power supply entry point, maybe 10uF/100uF?

Will do. Can't harm anything. As long as I can find the board space.

Thanks.
Tarun

PS: BTW, if anyone wants these layouts, just ask. I can send you the Eagle directory contents, zipped. Or I can send you PostScript or PDF of track side, component side legends, or solder mask, 1:1 or 1:2. Remember, it's work in progress, so you'll get whatever the current snapshot is. I won't post the files to the forum unless a lot of people ask; send me email.

[tcpip]

Originally posted by runebivrin
Near as I can tell, this circuit has no power ground beyond what goes to the filter caps. There's no low impedance return from a loudspeaker to consider, so currents in the ground should be insignificant.

That's exactly the point. I get conflicting views about the need for star grounding in circuits like this. I asked Randy Slone, one of the Gurus I have learned many things from, and he says it's not necessary for signal-level circuits. I asked Angshu, another of my Gurus, and he says that he always does star grounding, even for low-current stuff. Others on this thread seem to have one or other opinion. I don't know what to follow.

Tarun