Announcement

Collapse
No announcement yet.

PCB Design for Preamp + Power Supply

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • PCB Design for Preamp + Power Supply

    Hi all,

    I'm trying my hand at translating a turret board design to a PCB. The goal is to reduce my build time as much as possible without sacrificing much in terms of reliability. For now I'm just going to have the preamp and the power/bias supplies on the board. The power amp construction is going to remain turret board w/ twisted pair heater wires.

    I've tried to follow the same layout and lead dress principles for the most part, but am stuck at how to handle grounding. I usually follow some sort of star ground scheme in which the signal grounds relating to each power supply node are fed to their corresponding filter cap. Then each filter cap is grounded to the same point on the chassis with its own wire. What do you guys recommend in terms of grounding scheme? Here's my rough sketch so far:
    Click image for larger version

Name:	Screen Shot 2020-05-25 at 2.51.47 PM.png
Views:	1
Size:	712.0 KB
ID:	876697
    Click image for larger version

Name:	Screen Shot 2020-05-25 at 2.54.01 PM.png
Views:	1
Size:	601.6 KB
ID:	876698

    I also have an unused 5V heater supply on my PT secondary that I'd like to use for DC heaters. I'm thinking a rectifier bridge with a low forward loss might get me close to 6.3VDC and save me the headache of having to carefully route/wire the heaters on the preamp/PI tubes.

    Any insight is appreciated, thanks!

  • #2
    I think you will struggle to get 6VDC off a 5VAC supply, even using schottky diodes.

    It's not unusual to see discrete twisted pairs for the heater supply, even on a PCB. But that would be counter to your aim to reduce labor. You can run the heater traces one on each side of the board so that one is under the other to minimize the loop area and reduce the AC fields. Make sure no sensitive signal trace runs parallel and close to the heater traces.

    As far as grounds go, again it's common to see each stage have it's own trace back to a star. On the other hand ground noise is a function of resistance so simply joining all stage nodes together with a big fat (low resistance) trace will probably work as just well.
    Experience is something you get, just after you really needed it.

    Comment


    • #3
      I star grounded the power supply, preamp and all except the output tubes in the Workhorse 30 and 60W versions. It takes some thinking. the basic idea came out as running a batch of parallel ground wires down the center of the length of the PCB. The amps were remarkably hum free without the fancy DC heaters or heater elevation. One could decide that a strip of PCB copper maybe 0.6" wide taking the same room as the paralleled ground traces would be as good, I guess. I didn't try that after not being able to hear any hum. Floor planning is a must in the design of a starred-pcb amp, from the positions of the controls and preamp tubes and power supply all the way back down to the ground traces.
      Amazing!! Who would ever have guessed that someone who villified the evil rich people would begin happily accepting their millions in speaking fees!

      Oh, wait! That sounds familiar, somehow.

      Comment


      • #4
        Originally posted by R.G. View Post
        ...One could decide that a strip of PCB copper maybe 0.6" wide taking the same room as the paralleled ground traces would be as good, I guess. I didn't try that after not being able to hear any hum. Floor planning is a must in the design of a starred-pcb amp, from the positions of the controls and preamp tubes and power supply all the way back down to the ground traces.
        I can speak to the wide trace option. I have used it my last 3-4 PCBs and they are all quiet and trouble free the first time. I run the fat traces for power and gnd down the middle of the board and decouple each tap that feeds local stages. In the end it's sort of a "distributed star" in the sense that the fat traces act as the "star-point."

        Ditto on RGs comment about signal flow. That keeps trace lengths short and avoids potential feedback coupling.

        IMO, I get a better overall result with a good PCB layout and mech design than with traditional chassis with turret or eyelet boards. YMMV.
        Old Tele man: Equations provide theoretical values, SPICE provides approximate values; but, the ears provide exact values.
        Hofstadter's Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.

        https://sites.google.com/site/stringsandfrets/

        Comment


        • #5
          I started a similar adventure a couple of years ago. I've recently posted my KiCad sources & gerbers here: https://github.com/tristancollins/amps

          I took a slightly different approach - instead of one big board ($$$) I went for linking together smaller modules ($) to take advantage of the cheap PCB fabrication options. Also allowed me more flexibility in design. The negative was a bit more off-board wiring - but I'm solving that with a few more daughter boards to tidy things up.

          Click image for larger version

Name:	IMG_0031.jpg
Views:	1
Size:	427.2 KB
ID:	857854
          Click image for larger version

Name:	IMG_0032.jpg
Views:	1
Size:	409.9 KB
ID:	857855

          I'd recommend copying how I did the heater runs. Notice I rotated the socket slightly to allow straight runs. No hum that I can detect. My footprint is on GitHub if you want to use it. Yes, you could do twisted pairs using the PCB and a heap of vias, but that might cause other issues. I could have done top and bottom instead, but I copied from some of the hifi boards and parallel seemed to work fine for them.

          For grounding I followed Merlin's books - so each module is its own 'star' and the stars are linked together (board to board in sequence), being connected to the chassis only at the input. I also had separate audio ground and earth ground throughout. So the copper pour is earth ground on both side, with connections to allow shielded wires to be connected. It's so easy to do I don't know why more people don't do that.

          I ended up building two amps using these boards - both a combined JCM800 style & plexi preamp, but one using 6V6s and the other a 6SN7. I added reverb and effects loops. I'm really happy with them.

          Tristan

          Comment


          • #6
            Oh, and I'd recommend a socket footprint with a centre hole - makes it so much easier to mark holes out in the chassis!

            Comment


            • #7
              @uneumann Hi prof fancy seeing you here!

              Thanks for all the inputs everyone. It's really helpful to hear from others who have been through the same design process. I had a little more time to work on this PCB layout and here's where I'm at: Click image for larger version

Name:	Screen Shot 2020-06-03 at 10.29.19 AM.png
Views:	1
Size:	279.8 KB
ID:	857895

              I'm attempting to adhere to good PCB design practices - even though this kind of thing is pretty forgiving when compared with something like a 4+ layer RF board. So, I'm trying to keep all signal and power traces on the top layer and leave the bottom layer just for return currents/ground planes. That makes it challenging to route the heater traces. I'm thinking of having the indivudal traces located at the top and the bottom of the board. Probably can't get away with this unless I go through with the DC heaters. I know they are generally considered less work than they're worth, but I'd argue that proper AC heater lead dress is even more time consuming...

              I was thinking it could work well to run each individual filter cap ground to one terminal - and then just run a single wire to the chassis. Could be over kill though so maybe I'll follow your example tristanc and just link the ground together. Love the modular approach to your PCBs by the way - opens lots of doors for experimentation!

              Comment


              • #8
                As a look sideways, it's worth remembering that all the wires don't have to go on the PCB. One old trick I used for distributing clean power was to cut a strip of double sided PCB stock and stand that up vertical to the plane of the PCB, soldering resistor-lead legs onto it where I needed to drop power off the strip and onto the board. This strikes me as possibly useful to you. You can define a really long, skinny component that's a double-sided PCB strip, with the heater voltages carried on each side. This can run heaters down the PCB and drop off heater power wherever it needs to. I can do some sketches if you need.
                Amazing!! Who would ever have guessed that someone who villified the evil rich people would begin happily accepting their millions in speaking fees!

                Oh, wait! That sounds familiar, somehow.

                Comment


                • #9
                  RG, good point I forget that there's a z-axis to take advantage of here as well. Might have to try that next time around.

                  I went ahead and had some boards made with this single PCB approach. Paid ~$50 + shipping through elecrow for 5x 2mm thick boards. I received the PCBs all the way in the US within a week and got an extra board for free - so I'm pretty happy with their service. So far so good. Couple minor tweaks I'll make for next time around would be larger pads for caps, and using all snap-in types for the electrolytic caps better mechanical stability. Many of the amps I've opened up use white adhesive to hold those large caps in place so I'll follow suit.

                  Click image for larger version

Name:	IMG_2567.jpg
Views:	66
Size:	1.67 MB
ID:	908195

                  The one major gaff so far is that I put the reservoir cap and rectifier prettttty close to the front-mounted FX loop send/return. Preparing myself for a buzzy, humming mess when I first power this board up. Maybe I can get away with it if I use shielded wire to these jacks... wishful thinking.

                  Click image for larger version

Name:	IMG_2566.JPG
Views:	63
Size:	2.06 MB
ID:	908197
                  Attached Files

                  Comment

                  Working...
                  X