[cetoole]

Quote:

Originally Posted by peranders It's an advantage if you use the correct symbols. It saves a lot of confusion.

Yes it is, but the eagle parts library doesnt have that part, and I havnt made one up yet. I assume the best way of representing that part would be as a jfet with the Source and Gate shorted?

Quote:

Originally Posted by peranders I saw also that you have a class B virtual ground. You have bypassed a bit but I wonder how good this will be.

No idea, thats why I am trying it out, and if its terrible, the transistors can be omitted, resistors shorted, and a high current opamp, like the OPA551u, of which I have several, can be substituted. I would like to, as much as possible, get away from using the parts we have now that have only one source, and a lot of availability problems, and I couldnt make another class AB buffer fit on the board without increasing size.

Quote:

Originally Posted by thedoctor just concerned since there's a few new designs surfacing up in here, and finalized in a few days after initial schematics.

I dont think of it as finalized, the first prototypes have just been ordered. I figure as much of my mistakes might exist in the orientation of the sot23 transistors as anything, although linked schematics and parts will help mitigate this. Also on that prototype order were pcbs for a diamond buffer board very similar to the one on this amp, just with jfet ccs instead of crds, and TO92 transistors and axial resistors that I will try using in a similar multiloop design (pimeta).

 

[cetoole]

Thanks to a suggestion by Peranders, I decided to switch the output transistors to SOT223 package, seeing how large the discrepency in power output there was between the two. Looking at the datasheets for some of the transistors I was looking to use, like the MMBT3904/MMBT3906 complimentary pair, confirms this, the SOT223 package can dissipate nearly 3x the power of the SOT23, and reading other places, the SOT23 can only handle 100mA or so, while the SOT223 can do 1A. Because of this, I made some changes tonight, but it was quite hard to route the PCB in the space I wanted, and use these output transistors. Anyways, here is what I came up with. Note the incorrect part is again used for the CCS, both on the diamond buffers, and now replacing the DO-35 package CRD, because I just couldnt make it fit while routed reasonably.

 

[cetoole]

Just a minor update to the layout, and I also finally have the schematic all correct; I finally made a library part for the SST502 CRD.

 

[kin0kin]

this is definitely not something that can easily built by even some of the well experienced builders. I suggest you introduce the oven baking method for soldering smds before you really decide to finalize the project...if i had to make this, im gonna oven bake the bottom of the board then slowly solder the few things on the top layer.

i think there will be a mich higher failure rate in building this than the pint.

 

[DaKi][er]

Quote:

Originally Posted by kin0kin this is definitely not something that can easily built by even some of the well experienced builders. I suggest you introduce the oven baking method for soldering smds before you really decide to finalize the project...if i had to make this, im gonna oven bake the bottom of the board then slowly solder the few things on the top layer. i think there will be a mich higher failure rate in building this than the pint.

doesn't look to hard at all, anyone with some skill in smd will easily do this one, just have to take your time and not settle for anything but perfect. No need for oven baking or any other overly complicated methods. I can see how the 2 layers overlaid like that makes it look complex but if you split the board into the separate top and bottom layers it will look less intimidating

Defiantly not for beginners but even for someone with little to no smd experience but enough in general soldering, would not have that much trouble. It’s all in the technique and getting the one that works for you

What was the failure rate of the mint due to bad soldering alone? I though they were from unstable circuits caused by cranky opamps

 

[cetoole]

Kin0kin: While this amplifier is not designed to be a first project, I dont see anything specifically that will make this impossible, or even very difficult to build. All it should take is a steady hand, and some experience. Just look at the success rate on Guzzler and 00940's USB dac, and the success reported during the proto runs of Alf's revision to that dac, and that SSOP28 PCM2702 is much smaller than anything here. The maxim IC, MAX1722, in SOT23-5, was quite simple to solder I felt, and thats about as small as anything here gets. DaKi][er is correct that looking at the layers individually helps, so here are the single side shots. As to the failure rate of the pint due to bad soldering vs. a cranky opamp from hell, who can tell? Some of the soldering jobs posted looked pretty bad, but then again, there were some people doing this as their first project.