hpl said:
Hi all, I'm just about to get started with my first DIY project. I've spent a lot of time reading but I thought I'd just get some feedback on my pcb-layout etc before i get to work.
PSU:
-----
Tx: 120 VA 2x18~
Filtered by anything from 1000 to 6600uf per rail (ill probably start with 3300 per rail As that is what i have available now)
Amp
-----
I'm using the schematics from Elliots site (Project no 72) but want to do my own boards. The layout is made in eagle and I've attached a picture.
[IMGDEAD]http://img146.imageshack.us/img146/7840/lm1875esppy4.th.jpg[/IMGDEAD]
Do you think the overall layout and specifically the grounding scheme is good, and what about the trace thickness? (signal traces 40 mil, power lines 100mil)
Components:
-----------------
What do people prefer for C1 ? bipolar electrolytic or some other type of capacitor?
Also, any specific concerns about the components in the zobel? film or ceramic?
that's it for now i guess..
thanks,
Hi hpl,
You've made a good start on the PCB. But there are some things that I think you might want to change, if you do another version. (Note that I have not checked the actual connection topology, versus a schematic. I only looked at the PCB layout issues.)
I would want to get the large electrolytic caps as close as possible to the chip's power pins. You could move the large input cap and all of the other input stuff to the left and/or upward, and put the big electros right up by the smaller bypass caps. You will probably want to leave enough room to be able to use up to at least 1000uF for the electrolytics, there, and maybe even 2200 uF. [The Nichicon low-ESR UHE-series, for example, has a 50V 1000uF model with a 12.5mm diameter (but 40mm height), with 5 mm lead spacing, and also 1000uF (w/20mm height) and 2200 uF (w/35.5mm height) models that have 18mm diameter and 7.5 mm lead spacing. (Their mouser.com part numbers all start with 647-UHE1H, with the remainder being 102MHD or 102MHD3 or 222MHD6.)]
You should probably also give the Zobel's R6 its own separate ground return, back to the star ground point (and probably also connect the speaker's negative terminal to that pad, on this board, so the zobel is directly across the speaker output terminals), which could also help make it easier to move the large caps closer to the chip.
After you build it, you might want to also try soldering a 0.22uF or so film cap directly between the LM1875's power pins, or on the bottom of the board. You could enlarge the traces around the power pins pads, in case you'd rather drill holes for that, instead of soldering directly to the power pins.
Usually, it's a good idea to not have any traces joining with acute angles (i.e. less than 90 degrees). PCB manufacturers call those "acid traps", and they can cause etching problems. So, after you move the large caps closer to the chip, you should probably revise the way the bypass caps' ground traces connect to each other.
You will NOT want to join the signal and power grounds, except at the star point.
I would also try to keep the V+ and V- traces as close together as possible, along with their bypasses' ground return trace, to minimize the loop areas, which should help to minimize field-induced problems. The three power and ground wires should be tightly twisted together, on their way to this board, too. The same idea should also be applied to your signal input and signal ground traces and wires.
I see that you have what looks like a lowpass RF filter for the signal input. Very good. You might want to consider moving it so that its capacitor is directly at the LM1875's positive input pin.
If you want to make the power, ground, and output traces even fatter, that won't hurt anything, either. You could also fatten the signal traces by 50% or more.
You might also want to think about using some type of wire-to-board connectors, and leave room (and matching hole sizes and spacings) for them. Soldering the wires directly to the board often turns out to be very inconvenient, later. Some small terminal blocks might be easy. I'd use two-position ones for the input signal/ground and speaker output/gnd, and a three-position one for V+/Gnd/V-. Examples: mouser.com # 651-1935161 (2 pos) and 651-1935174 (3 pos), which have 5 mm lead-spacing
The speaker ground wire (connecting to Zobel's gnd) and the separate ground wire from there to the star point would have to share a spot in the terminal block, which shouldn't be a problem. But it might be, with some other connector type. You could also just use a three-position connector, there, with the two gnd positions connected by a fat pcb trace.