OK, found it!!!
For those interested, Owens PS boards are also $12 and can be found here:
http://www.diyaudio.com/forums/head...-headphone-amplifier-pcbs-63.html#post2716995
Owen:
Do you recommend one separate front end PS for each amp channel? Would 1 be OK for 4 channels?
454Casull:
That question is far to vague to answer correctly.
I'm putting together a spreadsheet that will help to calculate the C/W of the heatsink you'll need based on a few factors that the user can pick.
You need to know the load, the rails, the idle bias current, the room temp, and the usage expectations. That last one is critical in the sense that if you expect to output full power indefinitely, you'll need a much bigger sink than you would if you use a 1/8th crest-factor (for music).
As an example, if you want to output 400W continuously to the load, then the output transistors will dump about 400W continuously into the sink. If you want to dissipate that much power while keeping the heatsink temp below 100C then you're looking at about 0.2 C/W which is essentially impossible to attain with passive cooling. If you're listening to real music though, with 400W transients, then you could get away with a much smaller heatsink and passive cooling becomes a viable option.
I'll post the spreadsheet tomorrow sometime and people can have a look.
Cheers,
Owen
That question is far to vague to answer correctly.
I'm putting together a spreadsheet that will help to calculate the C/W of the heatsink you'll need based on a few factors that the user can pick.
You need to know the load, the rails, the idle bias current, the room temp, and the usage expectations. That last one is critical in the sense that if you expect to output full power indefinitely, you'll need a much bigger sink than you would if you use a 1/8th crest-factor (for music).
As an example, if you want to output 400W continuously to the load, then the output transistors will dump about 400W continuously into the sink. If you want to dissipate that much power while keeping the heatsink temp below 100C then you're looking at about 0.2 C/W which is essentially impossible to attain with passive cooling. If you're listening to real music though, with 400W transients, then you could get away with a much smaller heatsink and passive cooling becomes a viable option.
I'll post the spreadsheet tomorrow sometime and people can have a look.
Cheers,
Owen
HYPERTUNE:
You could run all 8 channels off a single board as long as you don't push the voltage drop across the regs too high. Each channel draws roughly 15mA per rail, so all 8 channels together would draw about 120mA through each reg. As long as the input to output voltage differential is 10V or less, you'll be good to go.
In other words, if you have 50V on the mosfets, plan for 60V on the LME, and plan for about 65 to 70V before the regulators. That would be a transformer with dual 48V secondaries.
Cheers,
Owen
You could run all 8 channels off a single board as long as you don't push the voltage drop across the regs too high. Each channel draws roughly 15mA per rail, so all 8 channels together would draw about 120mA through each reg. As long as the input to output voltage differential is 10V or less, you'll be good to go.
In other words, if you have 50V on the mosfets, plan for 60V on the LME, and plan for about 65 to 70V before the regulators. That would be a transformer with dual 48V secondaries.
Cheers,
Owen
Very nice design, thank you for sharing and organizing the GBs.
Schematic Comments:
1) Datasheet says Cmute should be no less than 47uF. I see a 10uF cap.
2) Datasheet recommends a decoupling cap across Vcc and Vee. This is shown in the demo board design as Cs13. It seems to be missing.
Thanks!
Schematic Comments:
1) Datasheet says Cmute should be no less than 47uF. I see a 10uF cap.
2) Datasheet recommends a decoupling cap across Vcc and Vee. This is shown in the demo board design as Cs13. It seems to be missing.
Thanks!
Technical Q&A time:
The Saint:
Very good catch on that! I owe you one. It turns out if you fix a problem twice, you probably didn't fix it.
I had forced a pin swap in the layout to accomodate the lateral fets, then before I posted the schematic I thought "having the wrong pins number is going to confuse people" so I swapped the pins around there. When I forward annotated it put me right back to the old layout for non-lateral fets
Good thing I haven't sent the boards in yet... it's still an easy fix.
Again, good catch!
454Casull:
As qusp mentioned, it really is very easy to solder 0805. It's actually faster and easier than soldering PTH once you get the hang of it. No need for fine tip irons or anything else special, it's just technique.
It makes a huge difference in layout quality and I can assure you that a good 0805 SMD resistor or cap will drastically outperform any PTH part money can buy. They're just physically closer to the ideal and you can't get around that with PTH parts.
Watch a few Youtube videos on the topic, and you'll see how easy it is. I have personally soldered 0201 parts, and I can attest to those being a little more difficult. Our lab techs can do 01005 by hand which is hard to believe considering you can hardly see the parts with the naked eye. 0805 is walk in the park, and if you can do 1206 or 1210 well, then 0805 is no harder. Again, it's all about patience and technique.
kipman725
That would be a nearly ideal setup. You'd get much better performance than Class D, and you'd get very close to the same efficiency. Making a supply that can do that isn't trivial though. If you want to do one, I'm sure there would be several here who would be interested, including myself.
FloridaBear
I haven't measured it yet, but output impedance will be very low due to the exclusion of the emitter resistors and an output filter. I'll measure that too when I get the board finished up in a few weeks.
Cheers,
Owen
Glad to be of help Owen, when I saw it I thought I hope he hasn't send the Gerbers off yet! Could have been a very expensive mistake.
Cheers
Anthony
Aussie Amplifiers
Hi Guys,
I've updated the list and it should be accurate up to this post. Please check it every now and again to make sure everything is correctly entered.
I've also updated the schematic and the layout to remedy the error with the D-S swap. I've attached both for review.
Q&A:
Arius:
Thanks for the comments! the more eyes on the schematic the better.
They are specifying a 47uF to avoid clicks and pops when using a mechanical switch to mute and unmute the amplifier. I think it's essentially acting as a an RC debounce for the switch.
In our application, we won't be using a switch, but rather tying the mute pin to the VCC supply for the LME itself. We do this by using a high value resistor and a capacitor to essentially create a turn on muting time constant. In that application, the capacitor and resistor are used to keep the amplifier in a mute state until the power rails and up and stable, usually about 1 or 2 seconds. For that application, all that is needed is a good 10uF ceramic and the appropriate resistor for the chosen rails. The resistor value is listed in the spreadsheet. Hope this clarifies why a 47uF is not being used.
I generally don't feel that this is necessary as long as the existing rail to GND caps are properly implemented in the layout. the LME reference board is surprisingly poor in its layout, and the use of PTH caps makes for generally poor decoupling. In this board, the large SMD caps allow for nearly perfect decoupling of the rails directly at the input, and again on the LME pins. If these caps are properly placed, as they are on this board, there is no benefit to adding a rail to rail cap.
HYPERTUNE:
Your numbers for the transformer are exactly correct. Use that, and all will be well!
As for the HV caps, I would suggest these on the input and output:
Digi-Key - 565-1737-ND (Manufacturer - EKZE101ELL121MJ25S)
They're only 120uF, but you really don't need much at such a high voltage with very low (and constant) current draw doesn't require large capacitance values. Plus you've got the regs to take care of ripple.
454Casull:
For the answer to your question, read the reply to what Arius asked about the mute pin.
Basically, there's an RC attached to the mute pin that keeps the amp in mute mode during power up. This means that an input signal would not be passed on to the output stage until the rails are stable. If you have a very long soft start, then you may need to tweak these values.
Cheers,
Owen
I've updated the list and it should be accurate up to this post. Please check it every now and again to make sure everything is correctly entered.
I've also updated the schematic and the layout to remedy the error with the D-S swap. I've attached both for review.
Q&A:
Arius:
Thanks for the comments! the more eyes on the schematic the better.
They are specifying a 47uF to avoid clicks and pops when using a mechanical switch to mute and unmute the amplifier. I think it's essentially acting as a an RC debounce for the switch.
In our application, we won't be using a switch, but rather tying the mute pin to the VCC supply for the LME itself. We do this by using a high value resistor and a capacitor to essentially create a turn on muting time constant. In that application, the capacitor and resistor are used to keep the amplifier in a mute state until the power rails and up and stable, usually about 1 or 2 seconds. For that application, all that is needed is a good 10uF ceramic and the appropriate resistor for the chosen rails. The resistor value is listed in the spreadsheet. Hope this clarifies why a 47uF is not being used.
I generally don't feel that this is necessary as long as the existing rail to GND caps are properly implemented in the layout. the LME reference board is surprisingly poor in its layout, and the use of PTH caps makes for generally poor decoupling. In this board, the large SMD caps allow for nearly perfect decoupling of the rails directly at the input, and again on the LME pins. If these caps are properly placed, as they are on this board, there is no benefit to adding a rail to rail cap.
HYPERTUNE:
Your numbers for the transformer are exactly correct. Use that, and all will be well!
As for the HV caps, I would suggest these on the input and output:
Digi-Key - 565-1737-ND (Manufacturer - EKZE101ELL121MJ25S)
They're only 120uF, but you really don't need much at such a high voltage with very low (and constant) current draw doesn't require large capacitance values. Plus you've got the regs to take care of ripple.
454Casull:
For the answer to your question, read the reply to what Arius asked about the mute pin.
Basically, there's an RC attached to the mute pin that keeps the amp in mute mode during power up. This means that an input signal would not be passed on to the output stage until the rails are stable. If you have a very long soft start, then you may need to tweak these values.
Cheers,
Owen
Attachments
454Casull:
I would say that it's definitely worth your time, especially if you only have to do it once to improve 8 channels.
In terms of sonic degradation, you'll be missing out on 10V of headroom which comes for free (extra output power without extra dissipation) and your noise floor will look just like the one posted in the measurements.
Will it be audible? The 60Hz+ harmonics noise might be slightly audible with efficient loudspeakers, but otherwise I would imagine general sonic quality should remain the same. The big benefit is the reduced PSU noise that the regulator provides to the front end.
I'll let you decide.
Cheers,
Owen
I would say that it's definitely worth your time, especially if you only have to do it once to improve 8 channels.
In terms of sonic degradation, you'll be missing out on 10V of headroom which comes for free (extra output power without extra dissipation) and your noise floor will look just like the one posted in the measurements.
Will it be audible? The 60Hz+ harmonics noise might be slightly audible with efficient loudspeakers, but otherwise I would imagine general sonic quality should remain the same. The big benefit is the reduced PSU noise that the regulator provides to the front end.
I'll let you decide.
Cheers,
Owen
They're available in the kit
Honestly guys, unless you already have the mosfets and the LME49830 sitting in a bag on your desk, I can assure you that the kit will be your cheapest option. I know the DK prices listed aren't the cheapest out there, but placing orders from 6 different suppliers to save a few dollars here and there will not be any cheaper. Digikey alone charges $9 for shipping, and that's if you live in NA.
I think a lot of people see the actual price required to build the amp ($78), and they think "Oh, that's a lot... I can do it cheaper than that". But when you add up all the little costs like shipping, picking up parcels, taxes or VAT, etc... you'll end up with the exact same parts, and might have saved $5 for your troubles. Chances are you will have spent more but won't realize it because you did it small bits at a time.
Again, I know a lot of you (like myself) have stashes of parts sitting around, and if you've already got half the kit then it's obviously not worthwhile. If you're starting from scratch though, the kit really is your best choice.
Cheers,
Owen
Honestly guys, unless you already have the mosfets and the LME49830 sitting in a bag on your desk, I can assure you that the kit will be your cheapest option. I know the DK prices listed aren't the cheapest out there, but placing orders from 6 different suppliers to save a few dollars here and there will not be any cheaper. Digikey alone charges $9 for shipping, and that's if you live in NA.
I think a lot of people see the actual price required to build the amp ($78), and they think "Oh, that's a lot... I can do it cheaper than that". But when you add up all the little costs like shipping, picking up parcels, taxes or VAT, etc... you'll end up with the exact same parts, and might have saved $5 for your troubles. Chances are you will have spent more but won't realize it because you did it small bits at a time.
Again, I know a lot of you (like myself) have stashes of parts sitting around, and if you've already got half the kit then it's obviously not worthwhile. If you're starting from scratch though, the kit really is your best choice.
Cheers,
Owen
Last edited: