S.E.X. on glass

[Bourney]

I've been wanting to try my hand at printed circuit board design for some time. Having populated boards for a variety of audio and non audio projects I really appreciate the work that the designers put into optimising the layout and dimensions. Time to have a go myself.

After watching many YouTube reviews of free PCB design software and searching various forums I opted for KiCad, but there were a couple of others that would probably have been equally as suitable.

I decided to use the Single Ended Experimenter kit as my guinea pig, because:
(1) It's the only current Bottlehead headphone amp I don't own. I built one as a gift for my nephew a couple of years ago and always thought I'd like one for myself.
(2) It has more complexity than Crack but much less than Mainline or Crackatwoa, so I'm hoping it will be a good middle ground for a first attempt.
(3) It says Experimenter on the tin, so OK let's experiment.

S.E.X. is proving to be a good choice so far. There's enough in it to get me learning KiCad and provide a balance between "OK, I get it, this is going well" and "WTF, why isn't this working".

I will be using 2 boards for this project. 1 for the amp and 1 for the power supply. Both will be manufactured on 4oz copper with a 3mm trace width, except for the integrated C4S circuit which has 2mm traces. By my calculation this produces a copper cross sectional area equivalent to 21 gauge wire and should be a good proxy for the wire supplied with the kit, but this is an experiment for me and only time will tell.

I ordered the board for the amp first as I wanted to see how it would turn out. Below are some photos of the board before and after population. I'm happy with the outcome so far and have ordered the power supply board. It should arrive in a week or two.

Any guesses as to how this thing is going to sound?
(a) Same as the original, copper is copper.
(b) Worse than the original, old school is the best school.
(c) Sound? What sound? I clearly have no idea what I'm doing and a puff of the magic smoke is about to take place.

[Paul Birkeland]

If your first board revision works properly and isn't noisy, that's a win regardless of how it sounds!

[ssssly]

You beat me to it. The boards I designed for my SEX rebuild are in the mail right now as well.

I designed mine with the amp and the power supply on one board and the DC heaters on another.

How it sounds will largely depend on how well you maintained proper power, signal, similars, trace to trace and trace crossover separation. And how well you managed your trace lengths.

If you are new to PCBs you might want to try EasyEDA. If you don't need the circuit simulations its quicker and easier to use than KiCAD. I use Fusion 360 Electronics(Eagle) myself.

Pay attention to the C4S section of that board. Given the standard package size for the 2N, using greater than 50mil traces could be problematic. Both bridging and heat wise. And you don't need anywhere near that. Particularly in 4oz.

And you may run into trace resistance issues with 4oz traces that thick in the signal part of the circuit. Static, buzzing. Just have to test it and see.

If it isn't already and you end up ordering future revisions, you might also want to ensure your ground plane trace is wider than the rest of the circuit. And preferably outside the rest of the circuit. If it isn't it can be noisy. You also probably don't want it running near the -bias trace.

Hope all the magic smoke stays where it belongs and it sounds good.

[Bourney]

Thank you very much for the tips. If the first version doesn't work out I'll be revisiting your advice for sure.

I'm looking forward to seeing a post on your build when the mail arrives.

[ssssly]

Mee too.

While I'm proficient at PCB design, im proficient at logic circuits and vhf/uhf signals design. Not linear power supplies. Or audio circuits. Although I suppose knowing how to make a good antenna comes in handy when you want to avoid them.

[Bourney]

After a little back and forth over my gerber files my PCB for the power supply supply has arrived.

I remembered to put a radius on the corners this time.

This board has the same specifications as the amp board, with the exception that this one is immersion gold (enig) with 2U" thickness gold, versus the amp board which is HASL with lead. As far as production quality goes they both appear to be of the same standard.

Below pictures are self explanatory. Mounting position copied from the Mainline.

I'm waiting on the custom front panel to arrive before I can go any further. Should be here next week.

[Bourney]

I completed the build and everything tested OK, but I have some hum/static.
I was suspicious of the 'balancing' resistors to ground on the XLR so I cut them. They weren't the problem.
I removed the components from the C4S section and installed the 150k resistors. Things improved. The static dropped away and left just a very low level hum in the left channel, silence in the right channel.
When I play music, even at low volume the hum is not audible, until there is silence in the piece, or between tracks. Not the end of the world, but not ideal either.
The hum exists with nothing connected to the inputs, but disappears when the tubes are removed.
When I locate my Mainline, Crack, Crackatwoa in the same location they have no noise, so it is not a noisy location, or perhaps the other amps are less susceptible.
I'll try the earth tab/diode mod next with some UF4007.
In any case I'll redo the amplifier board , move some components, reroute some traces and reduce the trace size on the C4S section because (a) I'd like the C4S and (b) the layout of my first design really made it a nightmare to wire up - a trap for the inexperienced.

On the positive side my CAD drawing for the plynth worked out nicely.

[Paul Birkeland]

Did you test for hum pickup when you moved the plate chokes and output transformers?

[Bourney]

No I didn't test, I took a chance, but it does seem that the cause is not external.

Any suggestions for the test method - moving and rotating and see what occurs?

[Paul Birkeland]

With the tubes removed and one of the high voltage AC wires disconnected from that PC board, you can turn the amp on to energize the power transformer and it will radiate its 60Hz magnetic field.  Set your DVM to read AC millivolts and probe between terminals 5 and 10 on each output transformer to see how much 60Hz noise is picked up. 

It might also help to look at how your ground traces are laid out on your power supply PC board, as that can cause some weird noise problems in ways that aren't always intuitive.

[Bourney]

Thanks Paul.

Tubes out and measuring mv AC between OT5 and 10.

With the left channel amp (noisy one) disconnected from B+ I get:
Left OT 4.3 - 4.4 mv
Right OT 4.0 - 4.2 mv

With the right channel amp disconnected from B+ I get:
Left OT 4.0 - 4.2 mv
Right OT 3.6 - 3.8 mv

So the left OT slightly higher mv in both cases. Are those numbers significant?

Here's a shot of the power supply board. The front and rear traces are the same width, though it doesn't look like it in the photo. The ground exits between the two +400V and is wired directly to the star ground point on the chassis plate. 0.1 ohm resistance from the ground point on the board to chassis ground or any ground point. The two B+ voltages are within a half volt of each other, as are the heater +/- voltages.

I also swapped the B+ output connections from left to right and ran the ground wire direct to the IEC ground point. No change, left channel still a slight hum. I also swapped the tubes between channels, no change, problem stays on the left channel.

Next step swap the OTs from one channel to the other?

Thanks again for your time on this non-standard build. No expectation from me of assistance, but much gratitude for any pointers.

[Paul Birkeland]

I would not swap the output transformers, that's not going to do anything productive.  You may have to run the amp with the output transformer on the noisy channel unblolted and move it around to see if that makes a difference.  The plate choke may also need to be repositioned as well.   I could see your L B+ being a little noisier than the right, but that would give you 120Hz buzz, which it doesn't seem like you have (be sure to listen to a 60Hz and 120Hz tone to be sure you're identifying the sound properly).