[balou]

Hi,
Well, the cat is out of the bag about my new amp design, and instead of cluttering the other related threads with general discussion about the amp, I just make a new thread.

Here it is: edit: pic updated

Features:
-Uses compensation output per default
-Designed for NE/SA/SE5534A, AD744 or AD829
-External compensation to make the opamps unity gain stable
-All-discrete buffer (monofied sijosae buffer)
-Fits in an altoids tin
-Low build cost (parts only about 40$ including shipping, about the same as only the panels of some other portable amps *SCNR*

)
edit:
-JFET Power Rail isolation for the opamps
-Class A operation through the compensation pin
-Jung Multiloop topology
-Reverse voltage protection diode

 

[balou]

To continue from another thread:
Quote:

Originally Posted by NelsonVandal This is really something! A cool competitor to Mini3. Will this fit in a mint tin with 2 9V's? I was planning on building something like this on a perfboard with 3 AD829's and BD139/140 or MJE243/253 instead of BC327/337, and maybe BC550/560 as input transistors. It seems like there's room for JFET's to isolate the rails, and then it would almost be in the same league as PPA v2 and LISAIII - in a tin! Why not use SMD components (except for opamps) - they're easier to solder and takes less space and could be laid out on both layers.

As of now, the board is 58x50mm, so 9v will not fit in the tin I currently have. Which leads me to another question... how big are altoids tin anyways? I only have Grether's Pastilles tins, which seem to be roughly the same size as the Altoids.

The BD139 is quite big, that's why I didn't consider it.
JFETs: doesn't a JFET isolated design draw much more power? Yes, something like this would be possible.
SMD: well... I wanted to make a better cmoy, and so I also wanted to use through hole components. Depending on who you ask, people chose either TH or SMD as 'easier'. About space: compare TO-92 and SOT-23 layouts:

The red pads are SOT-23, green is TO-92
if you need a via, TO-92 is actually smaller. But I agree that SMD parts would be a far better choice for the bypass caps.
And I don't want to go lower than 0805 package, else it gets definitely harder to solder than through hole.

Do I get another vote for SMD? Then it will be SMD where practical

Quote:

If you're planning on selling bords you can count me in for a couple.

I will order at least one eurocard at olimex - meaning 4 boards, of which I intend to use two or so

 

[cetoole]

Looks pretty cool, kinda like a through hole version of something I laid out (but not build) a few months ago.
http://www.head-fi.org/forums/showthread.php?t=234342

 

[balou]

Yes, the two designs are indeed very similar. I'm actually thinking about going SMD too - it would allow to use two 9v batteries (well I don't really care about that) and it would also potentially allow to cram more boards onto a panel (much more important to me - that's also why I was excited when I brought down the size to 50x58mm, because at this size I was able to fit 4 onto an eurocard). I think the TO-92 transistors look kinda cool and oldschool, that's why I probably stick with them (edit: it not only looks like it, it is oldschool. you could've built exactly this amp twenty years ago). Hey, it's not _only_ about the sound

one question to consider: are there good sounding transistors which are available only in SOT-23/TO-92? Oh and another question, what type/package are those huge smd transistors in your design?

I think I have to add the JFET rail isolation now to make it less alike

and my design has a TLE2426 *nitpicking*

edit: yeah, you bad bad people confused me. clone cetooles design by going SMD, going even larger by allowing space for BD139 audio transistors, staying with TO-92...
what's sure for now, decoupling will be done by 0805 or 1206 ceramics

 

[MrMajestic2]

Dont finish this too soon, I have 6 other amps to build first

Looks interesting.

 

[balou]

I will either finish it next thursday or in september - olimex has summer holidays in between

Also, considering the SMD vs TH question: I think a dense SMD design will turn away some of the less experienced builders. And part of the initial motives to build this amp was easy to build, a replacement for the cmoy and easy to source parts (meanig also stuff already lying around in your home, school/university lab etc). And frankly, being a Grado owner, I really don't need 18v. The 5534 will swing about 4 volts lower than rail-to-rail (at +-15v, but it goes down to +-1.5v operation, so voltage swing must be higher at low voltages), and I barely use more than 2v output.

 

[dBel84]

http://www.head-fi.org/forums/showpo...postcount=1970 this was my take on a buffered 3 channel amp

, good luck with your design and build..dB

 

[Filburt]

Something that may be worth keeping in mind is that if you pull from the comp out on the NE5534 at unity gain and that's what's in the feedback loop, it may oscillate.

 

[balou]

dBel84, thanks!
Filburt: thanks for the tip. I'll have to see if it works if the NE5534A is both compensated and output taken out of the comp pin.

I now implemented the JFET power rail isolation
Current pic:

edit: only SMD parts at the moment are the 3 decoupling caps for the opamps (mounted directly underneath the socket)

 

[balou]

do you see now what happens if some stoopid JFET yells fire on an already crowded board?

sorry for the bad pun, but I just worked through the night on the amp design. Mind not so clear anymore. Good night everyone

 

[TzeYang]

is this amped being worked on using EAGLE CAD?
Looks pretty good so far. But why use SE5534?? :lol:

 

[balou]

Yes, I'm using Eagle, and Eagle3D/Povray for the 3d pics.
Why SE5534A? It has very low noise, and is still widely used in various hi-end and studio equipment. Low cost doesn't always mean low performance. I think the reason why it's so cheap - it has bipolar inputs, which requires some more care than the modern FET input amps. It's also not unity gain stable without external compensation, and it will oscillate if you don't take care about your layout.
It's in stark contrast to chips like the OPA2134 - you can drop this chip in almost any design, and it'll just work. The 5534 needs much more care because it's just a raw opamp without all the protection and stability circuitry in the OPA2134.

And I think another reason for it's low cost is that it's design costs are completely amortized and there are no patents on it - everybody can clone it, and due to it's simple construction, it can be manufactured cheaply at the IC fab, and it's low complexity also means it has a high yield.

But if you don't like this chip, you could of course also use the AD744, which is highly regarded here (also because of it's comp out pin)

 

[NelsonVandal]

IMO AD744 sounds better at 18 V. It's spec'd minimum supply is 9 V. I've never heard NE5534, but it's operating range is 10 - 30 V. I think your board must be small enough to fit two 9 V's. So you're probably right about extra features like JFET's etc to keep it small, unless you go SMD.

SMD parts are not only smaller, they can be placed on both sides of the board and you don't need as many vias. Also, some say tombstoning of resistors make the amp less stable. If you stick with TH, you could have a distance of the holes to fit 805 or preferably 1206, and then it's free to chose SMD or TH as you please. You could also add traces to fit SMD JFET's like SST310 that had to be cut if you want to use those, then it would be even more freedom of choice how good your amp is going to be. Because this is probably not only going to be a small and cheap amp, but a superior sounding one.

If you want input from the gurus here (I'm definately not one of them), please show the schematics and layout.

 

[heatmizer]

Looks real nice so far. Is the B^3 going to be a commercial board. Are you going to release any eagle files for it?

 

[balou]

NelsonVandal: yes, after reading the AD829 and AD744 datasheets, I also think that two 9v will be necessary. I'm getting close to it (maybe you could already cram two 9v in it). I will use as much SMD as needed to fit in two 9v batteries. But considering the buffer transistors: will SMD parts sound as good, will there be enough choice of different transistors and also important, will they have the same output power?

heatmizer: I will most likely order a 320x200mm panel, meaning 25 boards, and make a group buy out of it. Releasing eagle files... it doesn't look like this is common practice with the amp designs here. I'd have to see