I was dreaming about putting in some balanced line drivers (THAT1646) and that takes 2 8pin dips. those require almost no supporting parts, just a nice bipolar power supply and bypassing.

if there was enough hole space in the scratchpad area, I could mount those 2 8pin dips and have a nice pro-audio style balanced out for XLR runs and have the 3ch pimeta outs drive local phones.

sorry to keep adding purposes to things.

---

[Quote]

Figure out a change plan, and we'll talk about it.

---

It may be better if you send me the PCB file, it's alot easier then photochopping one of the pics you posted.

---

[Quoting tangent]

Here's a thought: what if there are only 2 rail caps, and we take over the space given to C2A for more scratchpad area?

---

love it!

2 rail caps really are enough. the 'more is better' does not always have to invade every design, does it?

---

[Quote]

Plus, it doesn't actually prevent you from having a third C2 cap: you can just put it in the scratchpad space and do a little creative lead bending to hook it into the power rails.

---

yup! great. great!

btw, I really do want to build a DRV/SSM/THAT style balanced driver setup on this scratchpad. those DRV-style chips want an op-amp before them and the pimeta should be (mor than) good enough to drive this kind of chip. the pimeta is already extremely low noise (even for pro audio folks) and being able to have a balanced line driver right on the board might really make a great piece for the 'rackmount crowd'

I also like the fact that a lot of the board can be SMD now, if the builder so chooses.

[Edited by linux-works on 02-25-2009 at 10:07 AM.]

EDIT: Nope, still need another resistor. Still, maybe the space is there for it. When I rejected it earlier, it was because I needed space for both that and the cap. which now isn't a problem. Will revisit it when adding the additional perfboard pads.

[Edited by tangent on 02-25-2009 at 11:06 AM.]

Also, thanks for the SMT pads where you've put them!

Are the input capacitors necessary? Who wants them and for what? (Almost) all CDP's and DAP's have output capacitors. I think most people are going to use opamps with low input bias currents so there wont be problems with DC offset.

Do the C4 capacitors really make the amp sound better? I seriously doubt it.

I also seriously doubt the virtues of "class A biasing" the input opamps.

These components take a lot of space, and maybe the space could be used for something else.

Also, it's great to see you working on this project again, tangent. The PIMETA is one of the most important stepping stones in an audio DIYer's journey (up the stairs to audio heaven...?). I'm not crazy on the Serpac because the screws crapped out on me last time I used it, but I don't have a very wide knowledge of project boxes and I'll probably just throw it into another Hammond like my current PIMETA. Again, glad to see this thread exists.

is there some reason a dc-blocking cap has to be after the pot?

and its a valid question to ask: who has the responsibility to ensure that a signal has no DC component, is it the source's job or should the receiver 'be conservative' and assume the worst?

on a portable amp like the pimeta, maybe its safe to assume that all sources are 'safe' and that the blocking cap really is not a good use of internal space.

discuss...

---

[Quoting linux-works]

is there some reason a dc-blocking cap has to be after the pot?

---

Capacitor followed by a variable resistor (pot) would seem to make a high pass filter

---

[Quote]

and its a valid question to ask: who has the responsibility to ensure that a signal has no DC component, is it the source's job or should the receiver 'be conservative' and assume the worst?

---

To answer a question with a question: Who gets the blame when a set of headphones get burnt due to a bad source? Most likely the amp.

---

[Quoting linux-works]

is there some reason a dc-blocking cap has to be after the pot?

---

While I'm at it: Alps blue will make the amp "unportable" and you still have to hardwire the input-, output- and DC-jacks, so why not the pot. I think that space is a waste for the majority. I think it would be nicer to have on-board minijacks and a DC-jack.

I wonder what opamps PIMETA is aimed for when designed for Serpac H-65. Some of the better opamps need a higher voltage supply than that from a single 6F22 battery. With three opamps and three LMH6321's, the current draw will be quite high. An amp like this should be powred by AAA's. 7 or 8 AAA's don't take much more space than a 6F22.

---

[Quoting nelsonvandal]

Are the input capacitors necessary?

---

---

[Quote]

Do the C4 capacitors really make the amp sound better?

---

If it makes you happier, pretend they're tantalums, something you see recommended in all the literature.

---

[Quote]

I also seriously doubt the virtues of "class A biasing" the input opamps.

---

---

[Quote]

These components take a lot of space

---

You can't count the LED and its ballast resistor against the board space cost. They pretty much had to be there already as a power indicator. That's the cool thing about this -- novel, I believe -- biasing design: it adds very few components compared to a naive implementation of a typical bipolar CCS.

You can argue about whether we really need both power LED positions, but I like the flexibility it affords.

---

[Quoting joneeboi]

That's an awful lot of space which I'm sure could go to something.

---

---

[Quote]

You could throw R1L/R and R2L/R next to R3L/R, R4L/R, R5L/R and R6L/R

---

---

[Quote]

then you could shift the whole L/R amp circuit to where C1L/R, R1L/R and R2L/R currently reside

---

There's an idea worth toying with here, though, but not to get rid of C1. I want to try swapping the C1/R1/R2 group and the OPALR/BUF/R3-6 groups vertically, to put the Cs in the center and make the left and right post-pot traces take more nearly similar paths. Even if all it does is move the big Cs toward the center of the board, where big parts belong, it'll be worth it. We did that back in the META42 days, but somehow got away from it.

Hmmm...now I'm wondering if I should swap the C2 and BUFG groups as well.

---

[Quote]

I'm not crazy on the Serpac because the screws crapped out on me last time I used it, but I don't have a very wide knowledge of project boxes and I'll probably just throw it into another Hammond like my current PIMETA.

---

Generally, though, this is just the downside of using self-tapping screws. It'd be nice if more cases had brass thread inserts.

---

[Quoting linux-works]

who has the responsibility to ensure that a signal has no DC component

---

Thus, we should give the DIYer the tools to fix or prevent the problem, if we can. This is especially the case with a small, inexpensive amp like this one, as it's more likely to be used with cheaper sources. We went the other way on the PPA, mainly because it's a high-end design, and so should be used with high-end, more trustworthy, sources.

Also, I think the average DIY-for-someone-else builder will want input caps, at least as an option. The nature of the business makes you want to build more conservatively than you might do for yourself.

---

[Quoting cobaltmute]

Capacitor followed by a variable resistor (pot) would seem to make a high pass filter

---

---

[Quoting NelsonVandal]

Alps blue will make the amp "unportable"

---

This is not to say that we should give up the Serpac H-65 target size. You have to pick something. If we just let ourselves keep growing the board, we'll end up with a PPA sized board or bigger. We've already got those.

---

[Quote]

you still have to hardwire the input-, output- and DC-jacks, so why not the pot

---

Why just one board-mounted part? Because as soon as you board-mount just one more part, you change the thing from a typical DIY project to a precision panel crafting exercise, or require custom panels.

With just one part board-mounted, you have a huge range of places on the panel where you can put that one hole. With two parts, the second hole has to be a precise distance and direction from the first, or you have to make oversize holes to allow enough slop that both parts can fit through the panel at once. Add a third, and the problem only gets worse.

Others solve this problem by providing custom panels, which then requires a particular case. That sort of project doesn't interest me much. I believe a large part of the value of a DIY project is in doing your own casework. I have no interest in making "Heathkit" type projects, where you just solder it all up and stick it in the provided case. If you want a kit project, look elsewhere.

---

[Quote]

I wonder what opamps PIMETA is aimed for when designed for Serpac H-65.

---

---

[Quote]

With three opamps and three LMH6321's, the current draw will be quite high.

---

I'm not super happy about having such a high current floor, but there are no lower-powered buffers that I like better for a design like this. 4-6 hours is acceptable for me, and I haven't seen any other objections yet. I was pretty clear about this in the first post.

---

[Quote]

An amp like this should be powred by AAA's.

---

This goes right back to the multiple board-mount part issue, kits, custom panels, and all that. When you design the board so it can go into any case over a given minimum size, you have the freedom to choose whatever power source you like for it.

No, I didn't want to bring up the old class-A-biasing-war again.

Those two tweaks could easily be auditioned while playing. I've done it. I can't hear any difference. If you do hear the difference, please let a friend switch for you and try to tell when the tweak is on or not. I was very happy with this kind of class A biasing until I did it live.

I believe in science and progress, an increase in knowledge with time. If something can be observed (like a better sound), it could also easily be proved in randomized blind tests.

[Edited by NelsonVandal on 02-28-2009 at 07:19 AM.]

---

[Quoting tangent]

If anything, they want to go between OPALR and R7 on each side. That pushes the other resistors outward, into the input traces. There just isn't four resistors worth of space there, not without doing something silly like necking the input traces down to like 10 mils to get them around the corners.

---

If you got rid of C1L/R, then you could put R1L/R on the ends, maybe run R2L/R horizontally above the entire set of resistors. Moving PG somewhere else, you could put R1L somewhere near headphone guy with R2L somewhere nearby.

My attached image is a proud display of my intense MS Paint skills with a capacitive touch pad. R2R surely couldn't occupy that space, so perhaps there is some fancy work that can be done with the other layers. Of course, this is all moot as you don't want to get rid of C1, but I just wanted to illustrate my point. I like the idea of swapping the OPALR/BUF/R3-6 with the C1/R1/R2 groups. I can't fully grasp the mess of routing the "Class A" portion if you do, so I'll just let you run with the idea and catch up with you later. =P

Of course, I'm happy with whatever you come up and don't really desire to have any other features installed. For that matter, my minimalism doesn't really care for power LEDs either, but that shouldn't keep you from including one.

Also, I love what you did with the pots. Clever, efficient and effective.

[Edited by joneeboi on 02-28-2009 at 04:16 PM.]

EDIT: Removed board image and schematic links. They were replaced with newer versions, which you can see later in the thread; they don't apply to this revision level.

Changes:

- Upper left mounting hole and one C2 cap removed in favor of more scratchpad area. Now one more row than needed to support DIP-16. New front area could support some sorts of board-mounted switches. MisterX, does this suffice?

- GND bus in scratchpad area is no more. I decided it was better to let the DIYer decide what both busses are good for. There are two audio GND (IG) pads now, one of which is next to one of the busses for easy tying.

- Added a little bit of space between remaining C2s.

- Several layer access pads moved. Still just one of V+ and V-.

- Added B+ bus access pad: B+ is just the part of V+ on the other side of the power switch, good for making circuits that have to remain powered all the time

- Using a small plane for most of B+, not as many big, thick traces/heat sinks.

- Moved power inlet Molices to align both with each other and with H-65 battery door tab cutout; several nearby changes as a result.

- Replaced TO-92 version of LM317 with TO-220 version, and increased RCCS from 0.3" to 0.4" to allow for 1/2 W resistor. Previous parts weren't big enough to allow C/10 charge currents with AAA and up.

- Renamed LM317 to IC1 and TLE2426 to IC2

- Replaced power diodes with dual thru-hole/SMD version (1N400x/S1A) and decreased thru-hole pitch a bit to make routing a bit nicer.

- Moved one of those diodes to a different place in the circuit to prevent discharge of batteries through wall supply when latter is attached but not providing power. Double-check this, please! I'm not absolutely certain I can get the benefit of the three diodes discussed above with just two. It looks fine to me right now, but then, it's 3 am, too.

- As usual, many small trace and positioning tweaks

Things tried that didn't work:

- The swaps of C1/R1/R1 and the main amp section, and C2 with the ground channel just doesn't work.

- A minimalist bass boost (one 0.4" pitch thru-hole pad pair, an 0.3" nested within, and an SMT on bottom side) appears too fugly to bother with. This is bugging me enough that I may try once again, but I've given up for now.

[Edited by tangent on 03-09-2009 at 06:07 AM.]

[Edited by NelsonVandal on 03-03-2009 at 02:50 PM.]

---

[Quoting tangent]

A cleaner way to break the C1 trace is with a Dremel.

---

Given the thickness of one layer of PCB in a two panel (4-layer) board, do you think you have to worry about someone cutting through the full layer if they Dremel C1? And if they do, what to they cut?

---

[Quoting bdhornback]

I understand your reasoning... but I really don't like taking a Dremel to a perfect PCB. IF you're going to short the C1 traces... could you please at least do it on the bottom of the PCB (instead of shorting the SMT pads) so it doesn't show from the top? Thank you.

---

[Quoting tangent]

A cleaner way to break the C1 trace is with a Dremel.

---

---

I really do hate 'ruining a clean board' by doing a knife or dremmel trick.

a wire, otoh, is fully reversable with no ill traces (so to speak)

could I put in one more tiny vote for non-traces and a big sentence in red in the how-to saying 'please user, DO install a wire short here'. I know answering user questions is annoying but the wire option means that no board is ever harmed in the call of duty I kind of like that design over the surgery approach.

I don't like cutting traces.

Thank you for ME not being the only one.

---

[Quote]

I really do hate 'ruining a clean board' by doing a knife or dremmel trick.

a wire, otoh, is fully reversable with no ill traces (so to speak)

could I put in one more tiny vote for non-traces and a big sentence in red in the how-to saying 'please user, DO install a wire short here'. I know answering user questions is annoying but the wire option means that no board is ever harmed in the call of duty I kind of like that design over the surgery approach.

---