[dsavitsk]

I have been looking for a self contained, small, AC powered amp for a while, and have not come up with anything. I also wanted to try the AD8397, but it seems to be tricky enough that it needs a PCB, so breadboarding is not an option.

Anyhow, here is what I have come up with, and I am looking for another set of eyes to look for errors or to see if there is anything that might be done better.

The spacing for C1 is odd as it is meant for both Blackgate N series 4.7uF caps as well as some long film caps I have. I also added 2 sets of output taps, one before the coupling caps for headphones, and one after to use it as a preamp -- they are both in strange spots, however. I also figure that if worse comes to worse, I can skip the output cap and put in a small inductor to kill HF noise.

 

[ericj]

The datasheet recommends both a .1uf multilayer ceramic bypassed to ground basically right on top of each power pin, and a 10 to 47uf tantalum nearby.

It also does not really like capacitative loads. states:

"CAPACITIVE LOAD DRIVE
When driving capacitive loads, many high speed operational amplifiers exhibit peaking in their frequency response. In a gain-of-two circuit, Figure 23 shows that the AD8397 can drive capacitive loads up to 270 pF with only 3 dB of peaking. For amplifiers with more limited capacitive load drive, a small series resistor (RS) is generally used between the amplifier output and the capacitive load in order to minimize peaking and ensure device stability. Figure 24 shows that the use of a 2.2 Ω series resistor can further extend the capacitive load drive of the AD8397 out to 470 pF, while keeping the frequency response peaking to within 3 dB."

So put an output resistor on it, probably inside the feedback loop.

Oh, is the orange field a ground plane? The datasheet also recommends against putting a groundplane right underneith it.

Edit: And maybe some more through-holes for C1, in case you get an uncontrollable desire to use input coupling caps that are some other length.

 

[dsavitsk]

Quote:

Originally Posted by ericj The datasheet recommends both a .1uf multilayer ceramic bypassed to ground basically right on top of each power pin, and a 10 to 47uf tantalum nearby.

The datasheet says the ceramics should be within 1/8". I was hoping that the MKPs within 1/8" would do. As for the others, there is 1360uF worth of caps just over an inch away.

Quote:

Originally Posted by ericj It also does not really like capacitative loads.

I'm not really sure what this means. If by capacitive loads it means the coupling caps, then they are orders of magnitude larger than what is being covered here.

Quote:

Originally Posted by ericj Oh, is the orange field a ground plane? The datasheet also recommends against putting a groundplane right underneith it.

I saw that.

Quote:

Edit: And maybe some more through-holes for C1, in case you get an uncontrollable desire to use input coupling caps that are some other length.

I don't know. I doubt I am going to have a desire to build more than one of these, but I suppose some more holes are cheap.

Thanks

 

[ericj]

Quote:

Originally Posted by dsavitsk I'm not really sure what this means. If by capacitive loads it means the coupling caps, then they are orders of magnitude larger than what is being covered here.

A meter of headphone cable is at least a few picofarads of capacitance. Perhaps even several.

I guess the way you've got it laid out, you can always shove an output resistor where those output caps are.

 

[DaKi][er]

I would be inclined to place 1206 sized smd pads for ceramic bypass caps as close to the power pins as possible, like right on top of them nearly
At least have the option to put them in

 

[dsavitsk]

Quote:

Originally Posted by DaKi][er I would be inclined to place 1206 sized smd pads for ceramic bypass caps as close to the power pins as possible, like right on top of them nearly At least have the option to put them in

There is plenty of room to move the film caps around, so I'll put these in.

I can also get an output resistor in so long as I can tombstone some of the others. I am concerned, however, that this will do more harm than good.

 

[amb]

dsavitsk, I know you're trying to make things as integrated as possible, but placing that power transformer right next to low-level signal circuitry is asking for hum. Just because it's a toroid doesn't mean it's free of stray magnetic field. I've done enough testing on this stuff now, I can just envison those nasty spikes on the spectrum analyzer without you having even built the amp yet.

My suggestion to you is to either use a wallwart, or to put the power transformer well away from the amp board (and I mean more than 6 inches). Get these basics right before talking about boutique capacitors and such.

 

[dsavitsk]

Quote:

Originally Posted by amb dsavitsk, I know you're trying to make things as integrated as possible, but placing that power transformer right next to low-level signal circuitry is asking for hum.

That bad that I should scrap the project? The case this is destined for is 6"x6"x2" and already has a DAC in it that uses this same transformer on the board. And indeed, there is some noise from the dac, but it is only audible at volumes much higher than where I listen. Also, the noise does not sound like transformer noise (it is a hash, not a hum, and I think the main culprite there are some AB resistors that I bought surplus since a version of the same DAC with metal flms is quite quiet.)

Anyhow, I am alergic to wall warts and batteries, so this is the only option. If this is the sort of thing where this won't be the greatest amp ever, and that it won't measure as well as other implementations, I can live with that. It is a cmoy after all, and is intended for portable use. If, however, you mean to suggest that it will be awful, unlistenable, audibly noisy, etc, then I won't bother with the project. I know you have talked about this issue elsewhere, but my recollection was that while you could measure the hum, and you thought it might be degrading the sound a bit, it was not really audible. Am I remembering this incorrectly?

Anyway, I'm not really talking about any expensive boutique parts here. The input caps (and the Noble pot) are some electronics goldmine specials, or some $0.50 Nichicon ES series, or maybe some extra N series that are sitting around. Nothing expensive, though.

 

[blueworm]

I am a pcb design beginner, but one thing seems to stand out and slap me in the face.
The routing of the input signal that passes right under the left output cap (C1L),
and then swithces from top to bottom just to bridge across a trace.
I would pay some attention to this area and re-route some of the lines.

 

[amb]

Quote:

Originally Posted by dsavitsk That bad that I should scrap the project?

No, but you will get undesirable noise that could be avoided simply by re-thinking about placement of the transformer.

Quote:

And indeed, there is some noise from the dac, but it is only audible at volumes much higher than where I listen. Also, the noise does not sound like transformer noise (it is a hash, not a hum, and I think the main culprite there are some AB resistors that I bought surplus since a version of the same DAC with metal flms is quite quiet.)

I don't know what your noise sounds like, not having heard it myself. But, transformer-induced noise is not pure AC mains frequency. It's that, plus a whole series of its harmonics. Some of those harmonics can be as high in amplitude as the fundamental. The result is a noise that might sound like a "hash" if low in level, or an outright buzz if loud enough.

See the following two noise spectrum graphs for comparison. The first is my M³ with its power supply in a separate enclosure, situated about 12 inches away from the amp. The second is my SDS Labs amp with its onboard Amveco toroidal transformer, which looms only about 1.5" away from the opamp.

Quote:

Anyhow, I am alergic to wall warts and batteries, so this is the only option.

By "wallwart" I don't necessarily mean junk, you could even roll your own if you want... the whole point is to get that transformer as far away as possible. The regulators could stay on your board.

Quote:

If this is the sort of thing where this won't be the greatest amp ever, and that it won't measure as well as other implementations, I can live with that. It is a cmoy after all, and is intended for portable use. If, however, you mean to suggest that it will be awful, unlistenable, audibly noisy, etc, then I won't bother with the project.

How audible the noise is depends on how sensitive your headphones are, and how quiet the ambient noise level is. Even though the amp might not be intended to be the ultimate, why compromise it when the problem is so easy to solve?

Quote:

Anyway, I'm not really talking about any expensive boutique parts here. The input caps (and the Noble pot) are some electronics goldmine specials, or some $0.50 Nichicon ES series, or maybe some extra N series that are sitting around. Nothing expensive, though.

OK, I saw Blackgates mentioned and that certainly qualifies as "boutique" in my book. Even if you already have them...

 

[dsavitsk]

Thanks for the information. I think I'll just have to sit myself down and have a frank discussion about the tradeoffs here. I do have the sense, though, that if I put the transformer on the board, I have the option to use it that way and try it out, and if it sucks, I can always take it off board. I still have one unpopulated DAC board too, so perhaps I could wallwart them both and share a supply. Or, I can put the DAC and the amp on the same board with a separate PS ... now see what you've started?

Is there any problem with supplying low voltage AC via a wart rather than DC? That is, leaving the rectification, etc. on the board.

Quote:

Originally Posted by amb By "wallwart" I don't necessarily mean junk

No, I don't either. I mean something else I have to carry around and lose. I moved recently and threw out a UHaul sized box (12x12x18) stuffed and overflowing with old wall warts and AC cables. I hate them.

Quote:

OK, I saw Blackgates mentioned and that certainly qualifies as "boutique" in my book. Even if you already have them...

These are $3 blackgates.

 

[dsavitsk]

Quote:

Originally Posted by blueworm The routing of the input signal that passes right under the left output cap (C1L), and then swithces from top to bottom just to bridge across a trace. I would pay some attention to this area and re-route some of the lines.

It is pretty ugly. The secret, however, is that I have more 10K Noble pots than I know what to do with, so I'll never use the panasonic. It is there for no particular reason. I will, however, take a look at it. I think that at least one trace has to go to the bottom layer, though it can probably be done better than it is.

 

[amb]

Quote:

Originally Posted by dsavitsk Is there any problem with supplying low voltage AC via a wart rather than DC? That is, leaving the rectification, etc. on the board.

It would be better to have DC coming into the case rather than AC, because it eliminates any possibility of AC hum components bleeding into other parts through wire crosstalk, etc. But, even with the transformer secondary's AC coming into the case it will be vastly better than having the transformer itself sitting less than an inch away from the opamp. If you're careful with routing the AC wires away from other signal wiring, then it should be fine.

 

[tangent]

Quote:

Originally Posted by ericj A meter of headphone cable is at least a few picofarads of capacitance. Perhaps even several.

Try several hundred. I just measured a Sennheiser HD-600 cable a few days ago, and got around 500 pF, with two different meters. That comes out to about 50 pF per foot.

 

[MASantos]

Quote:

Originally Posted by dsavitsk Thanks for the information. I think I'll just have to sit myself down and have a frank discussion about the tradeoffs here. I do have the sense, though, that if I put the transformer on the board, I have the option to use it that way and try it out, and if it sucks, I can always take it off board. I still have one unpopulated DAC board too, so perhaps I could wallwart them both and share a supply. Or, I can put the DAC and the amp on the same board with a separate PS ... now see what you've started? Is there any problem with supplying low voltage AC via a wart rather than DC? That is, leaving the rectification, etc. on the board. No, I don't either. I mean something else I have to carry around and lose. I moved recently and threw out a UHaul sized box (12x12x18) stuffed and overflowing with old wall warts and AC cables. I hate them. These are $3 blackgates.

If you remove the transformers and regulation circuitry from both boards( amp and dac) you should have space to put the dac and amp in one board and then build a PS board for both the dac and amp! You have 2 enclosures aswell, eliminate the issue regarding hum and interference, and also eliminate the need for output jacks, input jacks, interconnects. saving money and decreasing the number of connections.

Manuel