Would anyone be interested in a relatively high end portable?

[Jazper]

I'm toying with/researching the idea of designing a portable amp based on the TPA6120A2 and standard opamps to drive it (eg the ad8610 etc). It will require some surface mount soldering to build, at the very least for the tpa chip but possibly also for a DC-DC converter chip -though the latter isn't firm yet as I'm still playing with ideas for the power stage.

Now I know that most people here are pretty scared of touching surface mount components, so if I were to offer the boards for sale I'd possibly do that step for them before I sent them out (with the TPA and the converter). This is considering the TPA is only about USD$1.50 or so and would add very little to the price of the board.

I'm estimating that this will be a nice little amp to knock around with and be about the size of a mint (or perhaps a smidge smaller).

Right now I'm looking at running it on either 2x or 4xAA batteries (but this is subject to change).

So would anyone be interested in this or would it be an exercise in redundancy?

 

[steinchen]

DaKiller http://www.mypage.tsn.cc/dakiller/tpapages/ has designed a neat little amp based on tpa6120, you'd only need to alter the power supply section. peranders got a tpa6120 design and PCBs,too

I've built one based on a spare pcb from DaKiller and must say that the little and cheap tpa6120 is darn good. Don't expect too much from the chip though, concerning price and size it really rocks, but it won't wipe the floor with other amps.

 

[Jazper]

Quote:

Originally Posted by steinchen DaKiller http://www.mypage.tsn.cc/dakiller/tpapages/ has designed a neat little amp based on tpa6120, you'd only need to alter the power supply section. peranders got a tpa6120 design and PCBs,too I've built one based on a spare pcb from DaKiller and must say that the little and cheap tpa6120 is darn good. Don't expect too much from the chip though, concerning price and size it really rocks, but it won't wipe the floor with other amps.

I have one! (at least until my gf swipes it ) I was planning on using a similar design, abeit with through hole components.

It may not wipe the floor, but it'd potentially be a nice cheap, good sounding amp while being portable.

 

[steinchen]

Quote:

Originally Posted by Jazper I was planning on using a similar design, abeit with through hole components.

the datasheet recommends using SMD components and short traces. SMD 1206 package components are not too hard to solder by hand.

 

[Jazper]

Quote:

Originally Posted by steinchen the datasheet recommends using SMD components and short traces. SMD 1206 package components are not too hard to solder by hand.

yes, but most diyers don't like using smd components unfortunately - though I prefer them.

 

[DaKi][er]

Quote:

Originally Posted by Jazper yes, but most diyers don't like using smd components unfortunately

I would be blaming this squarely on the fact that most of them have never tried smd's, didn’t have a well made PCB's for smd or didn’t have good enough tools for the job

I remember back before I done any smd work at all and the thought of smd scared me some and got quite frustrated at parts only coming in smd instead of through hole but after actually working with them (on well made pcb's) I quickly came to see their superiority

(To pull a phrase from the cable argument) Don’t knock it 'till you try it, and make an effort as well

 

[mono]

IMO, surface-mount components aren't as much of a size reduction as other things, and should be among the last things done- only if other measures don't reduce board size enough.

By other measures I mean things like only using enough capacitance, like the opamp spec sheets dictate rather than throwing huge electrolytics on the board. Further, getting rid of large flim caps & opamp sockets. Idea being to get rid of the tall parts, as most layouts are taller than a AA, but wouldn't necessarily be longer than one if optimally sized caps were used. Looking at boards like DaKiller's (which I"m not knocking, it looks nice), you can see there IS free PCB space for non-surface mount components, but where there is wasted space (considering it's a portable amp you consider rather than Dakiller's transformer powered design) is the height, even ignoring that Dakiller's has the integral power supply circuit. Basically it's just optimized for a different purpose.

Once you get into implementing a switching power circuit, you don't necessarily even need AA anymore. I mean, with a slim circuit board you might try to make it no larger than the dimensions of 4 x AAA, so the board could sit on top of the AAA battery pack and use a case roughly 5.5x4.5x2 cm. That makes it truely shirt-pocket sized. I'm sure there are holes in this idea, just a passing thought...

 

[Jazper]

Quote:

Originally Posted by mono IMO, surface-mount components aren't as much of a size reduction as other things, and should be among the last things done- only if other measures don't reduce board size enough. By other measures I mean things like only using enough capacitance, like the opamp spec sheets dictate rather than throwing huge electrolytics on the board. Further, getting rid of large flim caps & opamp sockets. Idea being to get rid of the tall parts, as most layouts are taller than a AA, but wouldn't necessarily be longer than one if optimally sized caps were used. Looking at boards like DaKiller's (which I"m not knocking, it looks nice), you can see there IS free PCB space for non-surface mount components, but where there is wasted space (considering it's a portable amp you consider rather than Dakiller's transformer powered design) is the height, even ignoring that Dakiller's has the integral power supply circuit. Basically it's just optimized for a different purpose. Once you get into implementing a switching power circuit, you don't necessarily even need AA anymore. I mean, with a slim circuit board you might try to make it no larger than the dimensions of 4 x AAA, so the board could sit on top of the AAA battery pack and use a case roughly 5.5x4.5x2 cm. That makes it truely shirt-pocket sized. I'm sure there are holes in this idea, just a passing thought...

Capacitance is a non issue; (capacitance multiplier using smd transistors(only need 3-4 parts) if necessary provides huge capacitance, in a small space.

As for a switching power supply circuit using ics, that was the plan. But runtime is a consideration and noise is also an issue, I have some samples coming from maxim for testing purposes (and will be playing on perfboard). SMD is nice, and recommended for the tpa. While it may not be necessary to run AAs, I feel safer doing so as they have high capacity for their size and the TPA is a hungry sucker for current (30mA).

 

[mono]

Quote:

Originally Posted by Jazper Capacitance is a non issue; (capacitance multiplier using smd transistors(only need 3-4 parts) if necessary provides huge capacitance, in a small space.

I'm no expert on capacitance mulitpliers but believe this would be a misapplication within the context of replacing the large caps (which I contend weren't needed but to follow this tangent). The goal for the caps should be minimal inductance, while the capacitance multiplier is meant for cleaning up the supply side? Usually when I'm seeing large caps, it's with a linear or battery supply.

What I'm suggesting instead is that there should not be a goal to have large caps nor to mimic that with a cap multiplier, that it's just not needed, a waste either way. This is ignoring the switching supply though.

Quote:

As for a switching power supply circuit using ics, that was the plan. But runtime is a consideration and noise is also an issue, I have some samples coming from maxim for testing purposes (and will be playing on perfboard). SMD is nice, and recommended for the tpa. While it may not be necessary to run AAs, I feel safer doing so as they have high capacity for their size and the TPA is a hungry sucker for current (30mA).

My point about the battery size was that once you start thinking switcher, you're not just choosing (for example) 2 AA OR 2 AAA. Your choice could be 2 AA or 5 AAA, or whatever... to the extent that the battery combination will make a difference in how small the case can be if the dimensions of the battery pack coincide with one of the PCB's dimensions.

I hadn't looked at the power requirements of the TPA6120A2 though, 30mA for it alone is going to push this towards being a relatively hungry amp, even moreso with a switcher powering it.

 

[Garbz]

I think it was fixup or someone who managed to get 8x AAA batteries into hammond's smallest enclosure + amp. This shouldn't be very hard, for an amp with a low part count, and surely we can go one size up. Aside from the taboo subject of switching audio supplies I doubt most people will go for the added complexity. CMoy, Mint, and Pimetas all are the most common amps simply because everything is kept very simple, and simple also means understandable, and cheap.

 

[Jazper]

Quote:

Originally Posted by mono I hadn't looked at the power requirements of the TPA6120A2 though, 30mA for it alone is going to push this towards being a relatively hungry amp, even moreso with a switcher powering it.

the switchers I'm considering aren't that power hungry at all,
MAX1771 - 12V or Adjustable, High-Efficiency, Low IQ, Step-Up DC-DC Controller
MAX669 - 1.8V to 28V Input, PWM Step-Up Controllers in µMAX
MAX775 - -5V/-12V/-15V or Adjustable, High-Efficiency, Low IQ Inverting DC-to-DC Controllers
MAX774 - -5V/-12V/-15V or Adjustable, High-Efficiency, Low IQ Inverting DC-to-DC Controllers
MAX1605 - 30V Internal Switch LCD Bias Supply

for example: The MAX1771 step-up switching controller provides 90% efficiency over a 30mA to 2A load - 110µA (max) Supply Current

The TPA is a power hungry chip, but with opamps the entire amp will still consume less than 50mA of current- if you were to say use 2x 2300mAh AA HiMH batteries you'd still get a significant amount of operating time. Consider the TPA to be 2x 15mA buffers. (24v/.050A = 1.2w), my computer mouse draws more power (3v/250mA= 12w) and it gives me over 20 hours of use from said batteries.

As for the switcher being a problem - I doubt that greatly, all of these switch ABOVE the human hearing range (from 500khz->500mhz) so any "noise" will be minimal, plus I plan to have some capacitance after it. Further, the emi field that radiates from them is only about 2mm or so

Also, this will be a relatively low part count amp, think a mint, with 2 extra chips and say 5 extra components or so.

if one wanted to run 4x9v batteries(2 per rail, no switchers) I'd have to impliment regulation to keep them at about 14v per rail or so, the tpa can only handle 15v+/-, and this would burn off more current than switchers.

It could be done with a single switcher (say the 1771) and a TLE2426, but the maximum current rating on a TLE is around 40mA and when you factor in the 30mA pig of a tpa, that doesn't leave you much headroom for powering opamps. So you'd need at least 2x TLE2426s for this, in an effort to keep part counts down, it'd be better to use 2 switchers I'd think.

And beyond this, capacitance will be necessary after the switchers, a capacitance multiplier would be a simple way of using 3-4 components per side to get relatively high capacitance to avoid ripple (and keep inductance low)while having small sized caps.
http://sound.westhost.com/project15.htm <-- more info there

 

[mono]

Quote:

Originally Posted by Jazper the switchers I'm considering aren't that power hungry at all, MAX1771 - 12V or Adjustable, High-Efficiency, Low IQ, Step-Up DC-DC Controller MAX669 - 1.8V to 28V Input, PWM Step-Up Controllers in µMAX MAX775 - -5V/-12V/-15V or Adjustable, High-Efficiency, Low IQ Inverting DC-to-DC Controllers MAX774 - -5V/-12V/-15V or Adjustable, High-Efficiency, Low IQ Inverting DC-to-DC Controllers MAX1605 - 30V Internal Switch LCD Bias Supply for example: The MAX1771 step-up switching controller provides 90% efficiency over a 30mA to 2A load - 110µA (max) Supply Current

90% efficiency sounds good on paper, but once you start doing the math and consider the additional space to implement it, that may end up being a signficant amount of power. I suppose what I'm getting around to suggesting is that this could end up being more expensive but no better nor smaller than existing designs. NOT that this is for certain, only that it could end up drifting in that direction.

Quote:

The TPA is a power hungry chip, but with opamps the entire amp will still consume less than 50mA of power - if you were to say use 2x 2300mA AA HiMH batteries you'd still get a significant amount of operating time. Consider the TPA to be 2x 15mA buffers.

I think 50mA is a rather conservative figure and final result will be higher, especially if one doesn't want to be locked into least power hungry opamps. Don't forget some like class-A and an LED.

Quote:

As for the switcher being a problem - I doubt that greatly, all of these switch ABOVE the human hearing range (from 500khz->500mhz) so any "noise" will be minimal, plus I plan to have some capacitance after it.

While the general theory about switching frequency is true, I suspect there will be more noise to contend with at lower frequencies. Take a computer power supply for example, it does switch higher than audio frequencies but some have reported they are too noisey. Capacitance will certainly be necessary but I'm figuring a careful (maybe even 2 stage) inductor circuit will be necessary. That's probably going to require an additional cap too.

Quote:

Also, this will be a relatively low part count amp, think a mint, with 2 extra chips and say 5 extra components or so.

I suppose my issue is then, will it be any better than a mint? IMO, where a mint might be bettered is to remove the protection diode, implement a ground channel, but use discrete surface mount buffers. Think size-reduced and simplified PPA, but with a 2nd board on a riser for surface mount buffers, a la glassman but ALL the buffers on a single 2nd board with sip headers to connect to the mainboard. I suppose I'm drifting off too much though, that is a more time consuming project.

Quote:

if one wanted to run 4x9v batteries(2 per rail, no switchers) I'd have to impliment regulation to keep them at about 14v per rail or so, the tpa can only handle 15v+/-, and this would burn off more current than switchers.

I can't imagine calling a 4 x 9V battery amp portable in any realistic sense beyond being able to put it in a carryon bag. Have you determined yet what the minimal voltage would need be? I suppose my main point is I can't see going with surface-mount components when it doesn't seem like the amp will end up being very small. I don't consider Hammond's smallest case very small. In other words, I'm not sure what the niche is supposed to be for this amp, what makes someone want to built it?

Don't get me wrong, I have at least a passing interest or I'd not have posted anything, and even now just throwing around ideas, not criticism.

Quote:

It could be done with a single switcher (say the 1771) and a TLE2426, but the maximum current rating on a TLE is around 40mA and when you factor in the 30mA pig of a tpa, that doesn't leave you much headroom for powering opamps.

Agreed, but look at a Pimeta, as it uses more than double 40mA in some configs and still manages to get away with a TLE2426 as only a virtual ground reference point. I'm suggesting that doing without a ground channel AND using a switcher may result in a lot more noisey amp than a simplied surface-mount variation of a Pimeta.

Quote:

And beyond this, capacitance will be necessary after the switchers, a capacitance multiplier would be a simple way of using 3-4 components per side to get relatively high capacitance to avoid ripple (and keep inductance low)while having small sized caps. http://sound.westhost.com/project15.htm <-- more info there

What evidence is there that "and keep inductance low" is true relative to same PCB real-estate without it? Is it not necessarily a higher inductance? I'm thinking that IF you need the capacitance multiplier, it will replace nothing, only be a further filter prior to traditional localized IC rail filtering.

 

[Jazper]

Quote:

Originally Posted by mono 90% efficiency sounds good on paper, but once you start doing the math and consider the additional space to implement it, that may end up being a signficant amount of power. I suppose what I'm getting around to suggesting is that this could end up being more expensive but no better nor smaller than existing designs. NOT that this is for certain, only that it could end up drifting in that direction.

Definately something to be mindful of.


Quote:

I think 50mA is a rather conservative figure and final result will be higher, especially if one doesn't want to be locked into least power hungry opamps. Don't forget some like class-A and an LED.

This is true, but opamps are never really fully class-A. Considering we're using 2 ics, we can't really expect class-A operation. 50mA may be conservative, but even at 80mA we're not drawing as much as a pocket pimeta iirc.

Quote:

While the general theory about switching frequency is true, I suspect there will be more noise to contend with at lower frequencies. Take a computer power supply for example, it does switch higher than audio frequencies but some have reported they are too noisey. Capacitance will certainly be necessary but I'm figuring a careful (maybe even 2 stage) inductor circuit will be necessary. That's probably going to require an additional cap too.

Well the plan is to test this idea on protoboard first, before finalising anything. As for switchers in a computer power supply (which can't really be compared as we're dealing with much smaller voltages) - many sound cards rival some of the best sources out there and do it with very low capacitance (overall) I really doubt there is as much an issue here as people make out.


Quote:

I suppose my issue is then, will it be any better than a mint? IMO, where a mint might be bettered is to remove the protection diode, implement a ground channel, but use discrete surface mount buffers. Think size-reduced and simplified PPA, but with a 2nd board on a riser for surface mount buffers, a la glassman but ALL the buffers on a single 2nd board with sip headers to connect to the mainboard. I suppose I'm drifting off too much though, that is a more time consuming project.

I think the power requirements of such a venture would make it prohibitive, especially for high impendance headphones.

Quote:

I can't imagine calling a 4 x 9V battery amp portable in any realistic sense beyond being able to put it in a carryon bag. Have you determined yet what the minimal voltage would need be? I suppose my main point is I can't see going with surface-mount components when it doesn't seem like the amp will end up being very small. I don't consider Hammond's smallest case very small. In other words, I'm not sure what the niche is supposed to be for this amp, what makes someone want to built it?

Ideal voltage would be approx +12.5/-12.5v, minimal voltage would probably be about +10/-10v, I wasn't really suggesting the 9v option as something to consider, more a "if I were to do it that way, that's what I'd need"

I suppose the aim is to make something that betters the mint(not that the mint is bad), while having a relatively simply topology. Also to make use of the beauty that is the tpa, provided it can be powered well enough.

Quote:

Agreed, but look at a Pimeta, as it uses more than double 40mA in some configs and still manages to get away with a TLE2426 as only a virtual ground reference point. I'm suggesting that doing without a ground channel AND using a switcher may result in a lot more noisey amp than a simplied surface-mount variation of a Pimeta.

There is also the other problem with virtual grounds, running other things from the same power source. I'd prefer to run dual switchers for just this reason - using the neg side as a ground. (hence the inclusion of inverting switchers in the sample run.

Quote:

What evidence is there that "and keep inductance low" is true relative to same PCB real-estate without it? Is it not necessarily a higher inductance? I'm thinking that IF you need the capacitance multiplier, it will replace nothing, only be a further filter prior to traditional localized IC rail filtering.

Agreed on the capacitance multiplier, but it does allow an extra bit of isolation, at the cost of a 2-3v drop.

 

[amb]

On a Pimeta (and PPA, as well as M³), the TLE2426 passes nowhere near its 40mA limit. In fact it passes a very miniscule amount of current. This is because:

• The opamps' quiescent currents go from V+ to V-, not through the TLE2426.
• The ground channel output buffer source/sink the headphone return currents and diverts them to the power rails.
• The TLE2426 only creates a reference potential for the inputs.

On a MINT (also cmoy with a TLE2426-based rail splitter), due to the lack of a ground channel, the TLE2426 does source/sink the return currents from the headphones.

So, if you plan to use a TPA6120, and plan to use a TLE2426 as a virtual ground splitter, the TLE2426 will source/sink the headphone return currents unless you also add a ground channel output buffer.

 

[mono]

Quote:

As for switchers in a computer power supply (which can't really be compared as we're dealing with much smaller voltages) - many sound cards rival some of the best sources out there and do it with very low capacitance (overall) I really doubt there is as much an issue here as people make out.

Are you overlooking that many sound cards have onboard power regulation for this very reason?

I agree that they don't have a lot of capacitance- that was exactly my point previously, that a lot isn't necessary. What they do have however is fast ceramic SMD caps close to supply pins, more often than not.