[eric343]

http://www.drsue.net/files/gilmore2pcb/r5/

Version 0.9!
I've changed around the transistor packages (so they're actually accurate!) and modified the pots' footprints so one can use proper multiturn trimpots. Before I commit (read

rder), I'd like to get some comments... because there's always one more bug. So questions, comments, flames, derogatory comments about my board layout skills, and 'yo momma' insults are welcome.

To give you some idea of the scale here, the (one channel) board is 100x80mm - a bit smaller than two credit cards stacked side by side!

 

[Possum]

Did you use an auto-router and check the traces over? It just looks like there are a few traces with unnecessary 90 degree corners where a straighter, more direct path could be used.

 

[eric343]

According to Kevin, he prefers 90 degree angles over 45s, sooo...

Quote:

(talking about an earlier revision; r3 or r4 iirc) Also because of the high voltages around the output transistors, the traces are too close and would probably arc. You really have to have right angles with as much space inbetween as possible.

In any case, I did some mucking about and it's now at v0.91... (not v.091 like in the image, my bad) so reload the page if necessary to see the 'slightly improved' version.

 

[fyleow]

Well this is sort of nitpicking but you could make the stuff on the lower left more clear. Can't really see some of the text since it's overlapping. You could move the word "R22" up instead of squishing it with "R17". But I'm sure you were looking for wiring flaws and not these small problems right?

Otherwise, it looks good.

 

[eric343]

I can't; Eagle places those automatically.

It shouldn't matter, since you can always use the "info" command to figure out what the part is when you're assembling the board. (I will keep the layouts posted even if I do sell them, and if you're going to build it you should at least download and install Eagle Freeware ed. so you can use it as a digital stuffing diagram)

 

[Possum]

Quote:

According to Kevin, he prefers 90 degree angles over 45s, sooo...

Yes, and you can see from his PCB files, such as for his dynamic amp, that the traces are all 90 degrees. But the paths are still very direct, e.g. only a 90 degree turn when it's necessary. Your traces have 90 degree corners from points located diagonally from each other, creating a zig-zaggy pattern. Besides, when you have so many traces already deviating from a 90-degree mentality, why try to keep it that way? I'm pretty sure when he means he prefers 90-degree angles, it's for the whole PCB.

 

[eric343]

True; the reason the anlges are mixed is because many were done by the autorouter (which prefers 45s) but corrected by me, or simply ripped up and re-laid in some cases.

So what you see is a mix of autorouter and inept newbie layout skills

 

[zzz]

My main concern with this pcb is that some traces are too close to each other and to pins even when the board occupies the whole screen in front of me, and that's like 3 times larger than what the result would look like. You can still move some traces a bit further apart and I'd recommend you to try doing that. And be sure to wipe the board squeaky clean from both sides with spirit or something like that after soldering to get rid of all the flux. Do not stand anywhere close to the board after powering it up

.

I think you mixed up collector and emitter on U$2. I'm not too sure I understand the reason behind using 100K in series with 100K to get 200K but hey, it's your board and these are your parts

. Pins 1 and 7 of JFET are not connected to the feedback resistors, are you using wires for that? Feedback from `-` can be easily connected to the corresponding pin of JFET with a trace. Links near D5 and U$7 can be avoided if you route both traces to touch the pin and solder it from both sides. R30 would be a bit of a pain to adjust and accidentally shorting trimmer to R41 with a screwdriver would suck, so be careful with that.

Other than that the board is fine, but I haven't checked the whole thing. I recommend doublechecking all your base, collector, and emitter connections. Not all transistors used in the amp are the same when it comes to pinout.

Well, have fun.

 

[eric343]

Thanks for the excellent critique!

The traces - will do

Soldering - I've got a spraycan of organic flux remover just for that.

traces - I've mucked around with them a bit, I'll post images later.

u$2 - D'oh! When I did the libraries for the transistors it was 11pm, and I also replaced the transistors in the schematic with them... of course, half of them were the wrong way round, but I thought I'd fixed all that. Well, it's fixed now

Feedback - How did that happen? Looks like Eagle didn't put a junction in the schematic. Oops

How's it look now?

R30- well, I have some diddle sticks... (plastic non-conductive non-inductive screwdrivers for adjusting trimpots)

Transistors - I know that much

I did the parts according to the datasheets; I hope I got them right

BTW, since you'd asked me for the Wheatfield HA-2 schematic, check out the last page of this PDF file:
http://www.pmillett.addr.com/Wheatfield/manual.pdf

 

[Jupiter]

eric343,

Why do you have all those resistor series? (zzz mentioned one of these.)

 

[eric343]

According to Kevin (yes, another "according to Kevin"), the high-value Vishays are spiral-cut with a laser to create the requisite resistance. As a result of the spiral cut, there's increased inductance, something that's a Bad Thing at these voltages.

 

[Jupiter]

Why not use something else than Vishays for high values? Maybe Holcos would be OK? Non-inductive Caddocks should be good, but they're pretty expensive. If you only use Caddocks for the high values, it would cost around $25-30 extra for each channel. (I can check prices if that's an option.)

By the way, which of the Vishays are spiral cut? Values higher than 150K? And at what voltages is that a problem?

 

[eric343]

Well, since Kevin didn't use any higher than 150K for the high-voltage sections, I'm assuming it's something like that.

And as for the voltage thresholds, you'd have to ask Kevin.

 

[eric343]

As promised, I've posted the new gif of the layout... Is that what you were thinking of zzz?