j4cbo

The controller board is designed primarily for the complete amps that Icarium, myself, and whoever else is interested will be building. For those, I'll be populating them myself as well as writing the firmware. That said, if other people are interested in using the design, I wouldn't mind selling the boards and preprogrammed processor chips. I could also just solder the surface mount portions, and leave the rest to be populated by the buyer.

I think the controller board compares favorably with stepped attenuators. A 4-channel Goldpoint attenuator plus a 2-deck selector switch winds up costing $350, and I'm estimating something like $250-$300 for the controller board in a balanced-only configuration, including a graphical VFD and knob. It'll be more like $150-200 for a two-amplifier-board configuration. Those prices don't include the additional $130 for a pair of transformers, if converting between single ended and balanced is desired. Electrical noise will be minimal. It's possible to shut down the microcontroller's clock entirely and turn off the VFD when the control interface isn't being used, but I don't anticipate that being necessary; I'll be careful to design the board for good separation between digital stuff and the actual signal path, and the UI portion can be shielded off from the rest of the amplifier.

I was looking at those, but even the lowest-ranging ones seem to have a significant amount of noise - equivalent to 80mA RMS or so on the rail even with a moderate amount of filter capacitance (from page 9 of the ACS713 datasheet). Are they actually "good enough" in practice? If so, I'll add a pair of those, an i2c ADC, and some temperature sensors right on the PCBs.

Well, here's what I really want.

What's the deal with the Femto? It's mentioned a few places, but Twisted Pear doesn't seem to have a page up for it. If that's not likely to be available, or if it doesn't do what we want, then I wouldn't be averse to designing something custom for volume, input, and so on.

There's a ground plane across most of the top; its area is marked by the dashed line, but it isn't rendered out in that image so that you can see what's underneath. The bottom left and right halves of the board have V- and V+ planes, so that there's a very low-impedance connection between the power connector, rail capacitors, and output devices; those planes do show up in the image.

New board: http://b.j4cbo.com/temp/gbf2-brd.png Hopefully this will satisfy more people

Compare the Vishay TNPW series: http://www.vishay.com/docs/28758/tnpwe3.pdf ... to their RN60D: http://www.vishay.com/docs/31027/cmfmil.pdf RN60D: Temperature coefficient 100 ppm, noise "Average for standard resistance range is 0.10

Looking again, it seems like adding 0.3" spacing as well wouldn't be too hard, which would allow most 1/4w resistors to fit. RN60 would still be out, but RN55s should fit fine. That said, I'd still recommend surface mount

If I'm reading Kevin's board layout right, there's supposed to be 10k resistance between the inputs of the diamond buffers even if that's not used as a feedback point; in other words, on this, R24 and R25 (just left of center) should be present and in series together even if their midpoint isn't connected to the rest of the feedback loop. As I said, though, I'm not entirely sure what the intent was...

Hmm, now I'm trying to be sure of Kevin's intent for the feedback setup. The original HeadWize post discusses optional multiloop feedback, which would explain having the intermediate feedback point as well. I don't remember if I'd forgotten one of the two at first, and I don't have old versions of the layout on hand. That said, it looks like Kevin's own version of the layout doesn't connect the intermediate point... Kevin, can you clarify how things are supposed to be wired up? Should I add in a jumper so that the board can be used in different configurations, or is the multiloop feedback not to be used at all?

OK, board and schematic updated (same URL) with all the suggestions so far. I removed the part labels until I get around to fixing up the numbers and changing the labels to display part number rather than value. I also changed the power rails on the bottom to be large copper planes, rather than wide-ish traces.

So it is... http://b.j4cbo.com/temp/gbf2-brd.png http://b.j4cbo.com/temp/gbf2-sch.png

luvdunhill, can you let me know the dimensions of those heatsinks?

In the process of switching to 0.2" + 0805 resistors, I wound up redoing a fair amount of the layout; it's a lot less cramped now. The offset trimmer pots are in, and there's room now to add 7815/7915 regulators for the servo. I'll post an updated version tonight.

Yep, I saw that. Very nice heatsink design. 0.2" for those very small through-hole resistors? Seems reasonable. I'll whip something up this weekend...

Hi, folks Is it still the plan to go with the same Dynahi-like board layout? I'm thinking that I'd actually like to play around with the design a bit here; in particular, using surface mount for resistors when possible, to keep the size down and signal paths short. I know a lot of DIY folks have a phobia of surface mount, but IMHO it's actually easier than through hole in quantity, and makes the end result a lot prettier. If I were designing a power supply for this sort of thing, I'd throw in a microcontroller, dot-matrix VFD, all sorts of voltage / current / temperature monitoring, and potentially dynamic bias adjustment - but that's just me. Is everyone set on using 2SJ76/2SK213 (or J79 and K216) output devices? edit: Fortunately I've still got my old board files around. Just barely within the size limits of EAGLE's free version.